GunLink Software Release Notes 3.X

If you have just installed this update then you must restart this computer before use!

If you currently have a 3.1.0 patch installed and you have the unlikely situation of needing to unroll this update then you will also unroll to the previous patched state, but it may not seem so as you may not be able to see the patch release notes in GunLink Web. The patch will still be active, but there is no problem re-installing the latest version of the patch if you want to. However you cannot unroll the patch to get back to the original 3.1.0 state whatever you do and whatever it looks like as the original 3.1.0 file backups are gone.

New Manuals. There is a new version of the Operator's manual included in this update covering all the new features.
 
Please replace any older copies both printed and PDF files, or at least mark them 'not latest'. There are also some other new manuals that were not previously distributed electronically, these may be newer than any paper copies you already have. See here for details. If you think you cannot see the latest manuals from GunLink Web try clearing the browser history or cache. You can also always access manuals directly from a file browser at /opt/gunlink/share/doc.

New Firmware. There are new versions of firmware for a number of devices. Check the firmware update notes to see which you need to install urgently. The new facility to disable bad timing links requires a firmware upgrade.

New gun error specifications. - Added warning level for gun errors, these are reported to the logs but not counted as errors for the statistics. These are configured in the same way as normal gun errors, either set the 'single error spec' flag when building the configuration and set the new warning level parameter in 'host prefs / specs' to a non-zero value, or clear the flag in the configuration parameters and set per gun warning limits using Configure.

Added cluster spread error limit. Again if 'single error spec' is set you can enter this once in 'host prefs', if not you can set a value per cluster using Configure.

The existing array spread error and the new cluster spread errors are recorded to the database in a way that allows you to see in the line stats tab or GunLink Web which guns were involved in these errors. Note only two guns are recorded for a spread error, it is possible there were other equally bad guns or some less bad ones which would still have led to a spread error if there hadn't been an even worse gun.

Both the start of shot and end of pulse solenoid voltages are now recorded to the database and viewable as graphs on the line stats. tab of display and the GunLink Web error reports. Previous confusion between 'fired volts' and the end of pulse volts has been clarified, the values are refered to as 'Pulse Start' and 'Pulse End' volts everywhere.

Separate 'sensor correction' and 'beam-steering' gun offsets. - Previously there was a single value called 'offset' for each gun which was typically used to correct for differences between gun sensor pick points (such as peak signal or zero crossing) and the true peak acoustic pressure as determined by hydrophone data or other measurements. Other possible requirements for offsetting gun aim-points individually include beam-steering and dynamic depth compensation but attempting to use the single offset values for either of these tasks didn't work if you needed shot by shot changes as they interacted with the tuning algorithms.

There are now two offsets per gun. The original value is renamed 'Sensor Offset' on the gun parameters tab and is associated with correcting for sensor offsets, if you had any offsets in use previously they are retained. If you change these substantially then the gun tuning will need to 'correct' them in the normal way. A new column is labelled 'Beam Offset'. This value is used to adjust the solenoid start time and the expected gun fire time on a shot by shot basis outside of the tuning mechanism.

A new message has been defined as part of the GCS90 navigation interface protocol which allows the navigation system to dynamically change the beam offsets on a shot by shot basis. Simulation of this capability has been added to the 'gnits' navigation simulator.

Split hardware error messages from Seismic Log. There are now two log choices, one includes all messages as previously, the second includes only messages related to seismic activity - gun errors etc.. The small scrolling log window in Display continues to show all messages. The Log tab of Display has an option box 'all messages', deselecting it will show only seismic messages. Similarly there are new options to print each type of log separately on the database tab and a new 'auto-print' option on the controls menu. All messages continue to be retained in the database, the only change is to select which (sub)set is displayed.

The 'all messages' option retains the original pdf file prefix of 'gunlog_...', the new option produces files called 'seislog_...'

Ability to disable bad timing links. Even with recent timing receiver firmware improvements a really intermittent timing link can still occasionally make the other link look bad even if it is actually reliable. This is mainly of use on GL4000. You can now force the TEM to ignore one or the other link, but not both. This is achieved by control buttons next to the link status in the panel applet node status display. The disabled state is not memorised in the TEM firmware as most reasons for wanting to do so are not faults in the TEM but the connection back to the ship. So both links will be enabled during the boot phase and any link disabling will take place when the node connects to the host but timing links are not required at that point. Requires a firmware update see here.

The new firmware version also greatly extends the time a link is automatically disabled after an error is detected from a few milliseconds to a few seconds.

Display changes.

• Several of the less frequently used tabs can now be enabled or disabled from a new 'Views' menu, and dismissed when not needed. This should allow a less cluttered workspace.
• The 'Colours' menu has been simplified and moved to the Plotting menu.
• The Positions tab has been rewritten to produce a graphical plan view of the source layout.
• Removed 'layout' file in Display. The layout file no longer exists.
  It often caused confusion and the ability to load multiple files was seldom if ever used.
  Settings are now saved as you change them to preference files and only the layout of the signature windows and trace gain settings now potentially depend on the configuration.
  Note when updating to this version any non-standard setup of signature window sizes or positions will be lost.
  You will need to arrange the windows once more, after which the changes will be persistent, until such a time that there is a very significant Configuration change. Similarly settings like 'Confirmation Requests' on the controls menu will be defaulted.
• Clicking Solenoids - if the depth of the selected gun indicates it is not in the water and it has pressure on it then you will be asked if you are really sure you want to click it. This will also happen if the depth or pressure for this gun is derived or nominal. If the pressure is low or the gun seems to be in the water you will not be asked.

Configure. New plan and Depth (side-on) layout views show scaled graphical views of the array making dimension mistakes easier to spot. The Nodes section has been rearranged to make parameters clearer. It now works more like the string tabs, once nodes have been added to the left pane clicking on a node brings up it's parameter settings on the right pane instead of everything being in one table which needed scrolling.

You can no longer set a node type to 'Unused', it must be set to the correct physical type. If you are not using it either because you are not using the strings connected to that node or you just want to be able to power it up and check it there is a new 'Spare' flag you should tick.
Until you edit and send to host any configurations with 'Unused' nodes you will get error messages when you start Display and Configure or the host programs which should indicate what to do. The system will still operate but the status of any 'Unused' nodes may not be displayed correctly and you may not get the correct power or fan controls for GL4000 nodes. When editing any such configurations ensure you have the correct node type selected, it may not default to your system type, you do not need to create a new configuration to make this change.

Forced host and Display restart on Configuration change. Following incidents where the clear warnings about the need to restart the host programs after configuration changes were ignored for extended periods resulting in statistics corruption you no longer have a choice. When you select 'update database and send to host' Configure checks if the system is in shot disable and if not refuses to do anything. Once the system is 'safe' and you select the option the host programs will restart automatically and this will trigger an automatic restart of the Display program. Similarly if you change the window sizes in 'host prefs' the host programs will restart either immediately or twenty seconds after the end of the line (during which time you will get reminders that the change is still pending).

The same mechanism also now applies to use of the GunLink 2000 'patch panel' feature of Display. When you apply changes the system will restart, you get a final chance dialog warning you of this. Previous unreported bugs with the patch panel feature should no longer exist and it should be usable again.

Disable Guns on MOB Enhancement - The guns are always disabled on activation of the MOB alarm via the TCU inhibiting fire signals in the timing links and also in the main software. On GL4000 this also switches off power to all the TEMS via the MOB connection to the UPSUCs ensuring a minimum disabled time. However for GL2000 it only disables the 24V solenoid power in the GCUs, if the MOB signal is cleared for whatever reason then firing may re-commence. As an additional safety feature the system will now switch to shot disable if the MOB or remote 1 or 2 connections to the TCU are asserted. It does not do so if you activate the front panel TCU key-switch.

Long-Phones increased sample rate Enhancement - The long-phone option now supports 0.5mS sample rate in addition to 1, 2 and 4mS. The total buffer size is unchanged at 32K samples so the maximum record length at 0.5mS is 16 seconds (actually 16x1024mS), at 1mS 32 seconds etc..

NOTE: This doubles the volume of long-phone data to be transmitted from the nodes compared to 1mS data. If you have marginal or faulty private Ethernet hardware it may increase the incidence of write blocking errors and worst case lead to loss of data or entire shots. Use of an Ethernet switch with a 1Gbit link to the host is very strongly recommended when recording long phones although not essential. If your system have been upgraded from release 2 you definitely do not have such a switch as original equipment, early shipments of new release 3 systems also do not. Contact Seamap if you want to upgrade or advice on sourcing a suitable unit.

XY dimension conventions. There is some confusion over numbering and coordinate conventions both within GunLink and in general. The GunLink convention is that the first string in the Configuration layout is the stbd. outer and the first gun on that string will be the first gun in the GCS90 header. This has not changed and usually reflects industry practice although we have no way of knowing or enforcing this if you want to use some other convention.
Our convention for cross line dimensions was opposite in sign to UKOAA which specifies going from port to stbd. is positive. The manual and revised layout tab now reflect UKOOA. To achieve a correct plan view on existing installations you may need to alter the sign of the string to string offsets on the Layout tab or switch to specifying them symetrically about a centre line. New installations will default to UKOOA convention nominal values until you enter the correct values.
UKOAA would require negative offsets measured from the head of the string to clusters and devices and for in-line offsets of complete strings but this is contrary to usual industry practice when describing gun arrays (except in final positioning data files) so is not going to change.
None of these changes have any impact on operation if you do not change anything as everything is simply mirrored. But if you are using far-field estimation and change the string to string offset values you should review the sign of your cross line offsets for cluster guns.

UPSUC monitoring of TEM current. Most of the dual channel UPSUCs shipped have broken main power rail current monitoring which always reads close to zero. As this is a source of repeated confusion all locations that display this now say 'not measured'. There is also an updated version of UPSUC firmware (1.4.4) which removes this value from the front panel display data. The default display on the UPSUC will now just say 'On: Lkg xxmA' showing any leakage current only. If you want to check the main supply current look at one of the Lambda supply readings.

NVIDIA driver update. A new driver version provides support for some newer graphics cards.

dandp Fix - pressure drop simulation was broken in 3.1.0.

Instrument tests Fix - Handle GPM/GFSM odd/even signal driven decisions correctly based on settings instead of trying to guess from the signal level. This prevented any results being shown for some bad channels.

tcucom Fix - UPSUC simulation data didn't update and was not always set even once.

master Enhancement - Added a lot more information to the configuration report created when master starts up. This is stored in the file /opt/gunlink/LOG/master-init and can be usefull for diagnosing problems.

adread Fix - Setting the 'long phones' record length less than the cycle time - to match the streamer record length for example - is not an error so don't report it every shot.

Fix - The master clock maintained by adread that other programs use to control timeouts could run slow. The most obvious result were spurious 'short predict' warnings if the arm to fire delay was less than one second. Log times and data are not affected.

tcucom Changes - Reduce the specification for reporting possible short arm to fire delays from 0.65 to 0.5 seconds to allow for more delays both in communications from navigation and internal processing. The overall system specification remains at 0.75 seconds.

Improve Node status Enhancement - The system now works out which strings a node controls, normally for a GL4000 this will be one string but for a GL2000 it may be more than one. This name is then displayed on the node status. If it controls only one string then it is the name assigned in Configure, otherwise it will be something like 'String 1+2'.

ad Fix - for a short period (up to five minutes) the day of the year kept by the nodes was not updated after midnight. This caused the node/TCU shot timestamp comparison code to flag errors.

+_12V limits Changed - Changed the lower limit to 10.5V to cope with production spread of GL2000 power supplies (ten percent).

pecom Removed - The 'feature' that started a new line if you were in external and on-line but didn't fire a shot for five minutes has been removed. The precise original reasoning behind this is lost in the mist of time - probably in case you forgot to close a line - but recently some complex multi-vessel operations have exceeded five minutes between shots. It was also inconvenient when collecting multiple patches of in-fill under one line name.

ISCT files Added aim-point time. - The isct.asc files (in /opt/gunlink/var/LOG) which contain space separated shot data including gun errors, depths and pressures now have the aim-point time as the second field after the shotpoint number. If you analyse these files using for example external spread sheets then you will need to make adjustments. The change makes the files more useful for one off experiments where the standard reporting mechanisms don't quite fit.

Database Enhancement - There are additions for the new spread and warning error limits. Also some additional data is recorded for possible future use - desired volume, actual volume and aim-point time.

writeSegd Fix - If only 'long phones' were selected for recording the process was not started when the host programs re(started).

stats Fix - When applying changes from the Display patch panel various errors could occur due to a compiler issue, the code has been changed to avoid them.

Device shotpoints Fix - when shooting flip-flop or other multiple source sequences the shotpoint counts added to devices other than guns and solenoids (DTs, PTs, phones etc.) were not necessarily correct. Depending on the system configuration history the count could be from the wrong string (not likely to be noticable in flip-flop) or the total line shot count. Note device shot counts for all devices on a string are calculated from the gun that fires the most shots on that string, not from individual gun or cluster positions.

gpsSim Enhanced - The NMEA and TSIP options now replace the time data in the sample data files with 'now', previously only the Motorola option did this.

gnits Enhanced - Added simulation of the new 'set beam offsets' message in the GCS90 protocol.

Various Various changes to provide host program changes.

New jars - New library jar files for Google Protocol Buffers.

TRecv1p4p12 Enhancements - Enables one timing link to be disabled even if apparently working, this is controlled by buttons on the node status applet display. You can only disable one link at a time, if you disable A then disable B for example, A will be re-enabled before B is disabled. The disabled status should be shown in all the places link status is displayed.

Automatic inhibiting of a timing link after any detected error was previously set to 256 timing words - about 1.6mS. It is now dramatically increased to 2.5 seconds to avoid spurious re-enabling of intermittent links.

TCU1p4p10 Fix - the time transmitted to nodes when the TCU is running without valid 1PPS from GPS was 3 timing words (18.75uS) slow. It's always been like this but only revealed when cross checks of the node shot time stamps were added in the last release. The faulty time stamps are not used for any data.

UPSUC1p4p4 All the dual channel UPSUCs shipped have broken main power rail current monitoring which always reads close to zero. This firmware removes the display of the incorrect value from the default front panel display message leaving just the leakage current. If you want to check the current look at one of the Lambda supply readings. Installation of this update is optional it does not affect operation and the host programs will not display any faulty current readings whether it is installed or not. (Note: release 1.4.3 was not a field release).

Split log support Enahncements - required for splitting the seismic and all messages logs.

get_line_log New - Customer requested script to extract line log to a tab separated text file for import into spreadsheets. Running '/opt/gunlink/scripts/get_line_log' in a terminal will present a list of line names and start times from which you need to select the index number. Output is created in a file in the current directory.

v3SysDetails New - Field engineer script to save all machine settings for transfer to a spare or as a backup. Saves network settings and many other configuration files not saved by the contract export tool. This may become a documented feature in future.

'no GPS fix' deletion script Added - Specific script - /opt/gunlink/scripts/deleteGpsNoFixFromLog - to delete 'GPS no fix' from the database and thus logs. Should be less required now the logs are split.

deleteFromLog Enhanced script - Added ability to select a line from a list (using '-t' option) so you can select one line based on start time if there are multiple lines with the same name.

Monitor Host Status Enhanced - The status of the host programs are now displayed at the top of the window - on/off line etc.. This is used to determine whether you can do an 'update and send to host' operation which now results in an immediate host program restart.

Node Section Enhanced - The left panel now only shows an entry for each node, clicking on it shows it's parameters in the right panel as for other devices. The previous table representation didn't have enough room without scrolling.

Plan and Depth views Enhancement - New scaled graphical plan and depth (side on) views allow checking of offsets in a more immediate format. It is important to note that the string to string spacings used for the plan view are notional (5m between strings) to achieve a viewable display. The Configuration contains no knowledge of even nominal string spacings, you enter these on the 'Layout' tab of Display as they may need updating manually during a line or in future may be updated automatically from navigation data.

Cluster spread error parameters Enhancement - As part of the new cluster spread error capability the cluster parameters will now include a spread error parameter if you have not selected 'single error spec'. If you don't want this feature or want one spread error for all clusters leave these at zero and set the 'single error spec' option in the configuration main parameters and the 'cluster error spec' options in Display / 'host prefs'. If you need different values per cluster set them here and clear the single erorr spec option this means you will also need individual gun error specs set in the configuration and you cannot change them on-line.

Sensor and Beam steer Offset graphs Enhancement - Added ability to graph both types of offset value from the line stats tab.

Optional Tabs Changed - Some of the tabs are now optional and controlled by settings in the 'views' menu. This is mainly future proofing if we add more tabs which you may not want all at once. Rarely needed ones like the 'Software messages' (the old 'half tab') and temporary ones like Instrument test results have close boxes in the tab title. The more important ones can be only removed from the views menu.

Removed layout file Changed - The layout file caused a lot of confusion and mixed configuration dependent data - like the layout of the signature windows - with static settings like colour choices. Additionally the ability to have multiple settings files was not used. All settings are now saved in 'prefs' files (stored in /home/display/GunlinkPrefs) and updated whenever you change a setting. Signature panel layouts and settings are saved similarly and will be defaulted if they don't match a new configuration. Existing settings will NOTbe migrated from your layout file automatically, you need to check and re-enter them.

Bubble Test Changed - Increased number of shots per gun limits to very large numbers (from 15 to 250) at customer request.

Bubble Test Fixed - The code to find the nearest phone if there is not a dedicated cluster phone was broken and always used the last phone on the string.

Bubble Test Fixed - Using the 'force line name' or some other similar options while recording a bubble test could cause it to start a new line and lose track of the correct test results. Remember You cannot change the name of Bubble Test lines even if using external cycle. They are always called 'Bubble-Test' and any navigation line name, start or end messages are always ignored while in bubble test mode.

Positions Tab Enhanced - This is completely re-written to draw a true x/y plan view of the source. The previous layout needed scrolling for all but the smallest configurations and didn't show dimensions to scale. Note using the conventional 'String 1 is stbd. outer' configuration then to use UKOOA sign conventions and a true plan view with the vessel on the left of the screen you need to enter negative offsets for the other strings. Or you can use symetrical offsets from a centre line with stb strings positive and port strings negative.

Solenoid Volts Clarified and enhanced - Previously there were confused references to 'Fired Volts'. On the Gun Status tab this actually was the volts at the end of the pulse providing a measure of how much energy was removed from the capacitors. But on the line stats. tab the graph refered to as 'fired volts' was the volts at the start of the pulse which should be very close to the desired voltage setting on the Gun Params tab. Now the names are 'Pulse start volts' and 'Pulse end volts'. You can also graph both values (and also for GI gun injectors) from the Line stats tab.

Colour customisation Moved - The colour customisation menu is now on the 'plotting' menu. You can only change trace colours now, not the gun colours.

Shotpoint Report Enhanced - The timestamps now show the full micro-second precision. There are new options that allow you to produce a list of just shotpoints and times.

Line Stats displays Enhanced - to show array spread errors and new cluster spread and gun warning limit data.

Line Error display Enhanced - Shows both start of pulse and end of pulse solenoid voltages. If GI guns are in use the table also shows injector start and end voltages and peak current (this may make the table overflow the formatting slightly on narrow screens).

Custom Query Shot data search Enhanced - Now includes ability to search for start of pulse or end of pulse solenoid voltages and the same for GI gun injectors.

No Change Recompile required to match library changes.

Remote system Time sync Enhancement - Using the system type tool to set a system to 'remote' will now create a 'cron' job that synchronises the remote system clock to the host using 'rdate' once per hour. Note you may need to have the time roughly correct to start with.

Desktop Wallpaper Enhancement - The various desktop background images have been updated to include the contact details for gunlink_support@seamap.com the primary source of support.

Remove getty sessions - Remove getty (terminal log-in prompts) on most spare GCU and TEM serial ports to assist testing and future use for dedicated purposes.

Updated Navigation interface specification document - Added new 'offsets' message.

GFSM Field Fault Report form - Added a copy of the doc file you should fill in for any GFSMs returned for repair.

General Field Fault Report form - Added a copy of the doc file you should fill in for any other items returned for repair.

3.1.0 Updated Operator's manual covering new features.

30/09/2011 Latest licence file.

New video driver - Supports some newer video cards.

If you have just installed this update then you must restart this computer before use!

If you currently have a 3.0.4 patch installed and you have the unlikely situation of needing to unroll this update then you will unroll to the previous patched state, but it may not seem so as you may not be able to see the patch release notes in GunLink Web. The patch will still be active, but there is no problem re-installing the patch if you want to. However you cannot unroll the patch to get back to the original 3.0.4 state whatever you do and whatever it looks like as the original 3.0.4 file backups are gone.

New Manuals. There is a new version of the Operator's manual included in this update covering all the new features, also updated Moxa serial to Ethernet adaptor instructions and Far-Field manuals.
 
Please replace any older copies both printed and PDF files, or at least mark them 'not latest'. There are also some other new manuals that were not previously distributed electronically, these may be newer than any paper copies you already have. See here for details. If you think you cannot see the latest manuals from GunLink Web try clearing the browser history or cache. You can also always access manuals directly from a file browser at /opt/gunlink/share/doc.

New Firmware. There are new versions of firmware for a number of devices. Check the firmware update notes to see which you need to install urgently.

'Bubble' Test enhancements - Add spinners to select the number of shots to fire for each gun. This is at customer request but you should remain aware that 'averaging' values for this sort of test may be of limited value. The table in 'display' will show the data from the most recent shot for each gun. There are two spinners, one for 'warm-up shots' which you can set to zero if you want and one for 'recorded shots'. Each gun will be fired for the sum of the two counts but SEG-D files (if enabled) will not be recorded for the 'warm-up' shots and while the processed data for these will be displayed on the screen it is not reported to the log (see next paragraph).

The system will start a new line called 'Bubble-Test' each time you select ''start all' or start selected', the start shot will always be 1001 and file 0001. As well as producing a line log for the test sequence this will also ensure any SEG-D files being recorded are grouped in one directory. But if your SEG-D directory name includes only the line name it will be the same directory for all bubble test sequences, and if the file name convention includes only the shot number and/or the file number then files will be over-written - this is true for any lines of course. The log produced contains messages reporting the measured and calculated parameters for each shot, it does not contain some of the normal start and end of line messages - about disabled guns for example - as these are confusing in this context.

Average values of the computed parameters are displayed in the results table in brackets after the value for the most recent shot, and reported as separate log entries at the end of testing for each gun. The display averages are reset at the end of any warm-up shots and new ones calculated for the 'recorded' shots.

A CSV (comma separated variable) file suitable for import into a spreadsheet is produced in /opt/gunlink/var/LOG. The file will be called 'Bubble-Test-xxx.csv' where xxx is the date/time of the start of the test sequence. The file contains a header line which has descriptions for the data followed by one line per shot. The data includes the shot number, gun, gun firing parameters (error, depth, pressure etc.) plus the measured and theoretical bubble calculations. Both 'warm-up' and 'recorded' shots are logged to the file.

The separate cycle time spinner in release 3.0.4 has been removed and an internal / external cycle selector added. If you select 'internal' then the normal internal cycle time is used and can be changed at any time. If you select 'external' then the system will switch to external mode and cycle time is under the control of the external navigation system. In external mode any start/end line or line name messages received are ignored in order to produce the new 'Bubble-Test' line logs. The array select commands issued by navigation are also ignored which may cause it to object to the returned headers, it is also likely to report errors if it tries to double check which guns/strings/arrays fired. If you select external mode the system retruns to 'shot disable' at the end of the sequence or when you cancel.

Long SEG-D Hydrophone records In order to provide continuous recorded coverage between shots there is now a 'long phones' option. This enables you to record hydrophone data at 1, 2 or 4mS sample rate (only) for up to 32, 64 or 128 seconds respectively rather than just for the screen window length.

You need to specify the sample rate in the SEG-D host parameter settings as normal. You also need to specify the maximum record length. This determines the file structure and size and ensures all files for a line are identical in format. The length needs to be larger than the longest expected shot cycle time to ensure continuous coverage but choosing a very long value 'just in case' will consume unecessary disk space. This also has an impact on network traffic if you are transferring the files to another system.

Any trace data allowed for after T0 for the next shot will be zero filled. Data in the first portion of the long phone channel set will overlap that in the normal hydrophone channel set but may be at a different sample rate. If as would normally be the case the configured record length is longer than the cycle time then the absolute times of traces in consecutive files will overlap, but the overlapping portion of the earlier file will always be zero filled. If the only use for the long data is autofire confirmation then you should only need to record it at a slow sample rate thus saving disk space.

The long phone channel set is always the last in the file, so will come after any notional and far-field data if you are recording that. If you try to stitch together the data from consecutive files there will be a small extra zero filled gap of between 5 and 25mS depending on sample rate, this is half the length of the resampling filters and should be too small to cause an autofire to be missed.

Sercel Air shut-off valve support For GL4000 GFSM only at present, requires a firmware update, see below.
These valves operate by being wired in parallel with the gun solenoid using a 'Y' pig-tail. Two diodes are incorporated in the pig-tail so that one pulse polarity energises the solenoid and one energises the valve. The valve needs to be energised for a variable amount of time depending on the severity of the leak as it needs a pressure differential to keep it closed. A small leak may thus require a longer operating pulse than a large leak but firing history will also have an effect, if the gun is still re-pressurising after a shot the valve is more likely to seal (even if there is no leak at all). Because of the way the valve works there is NO way to correct the situation if you accidentally operate the wrong valve with the guns deployed.

First you need to tell the system which guns have valves fitted. In the array configuration the parameters for each gun have a new 'valve' option. This defaults to 'none'. If you select 'Sercel' for a G-I gun then it is assumed to be wired to the Initiator solenoid. Once you have told the system there is a valve on a gun then the 'reverse solenoid' control on the gun parameters tab of Display is disabled (you can still control the polarity of the Injector solenoid of a G-I gun). You must arrange the gun wiring so the 'normal' polarity operates the solenoid, not the valve.

To operate the valve once you are absolutely sure you have identified the leaking gun right click on the appropriate gun symbol on the main gun status screen. The normal 'right click' menu will have an extra item 'Energise Valve'. Selecting this brings up a box allowing you to select a range of energisation times from 2 seconds to 24 seconds. If you have previously tried to energise the valve then it defaults to the last used value (for that copy of the Display program). The minimum value of two seconds should suffice for all significant leaks, use of longer values may cause more heating in the fire module so should be avoided unless really necessary. Once you select the time and 'activate' there is a final chance to cancel before the valve is energised. Messages will be issued to the log and screen reporting progress, in particular the peak current through the valve is checked to be at least 2A or activation will be abandoned - the valve is probably open circuit.

The valve is 'fired' like a normal solenoid using the normal solenoid volt setting from the gun parameters tab, this ensures the solenoid is pulled in properly but unlike for a gun solenoid the power supply is kept permanently enabled so the valve solenoid is held in. This required a much lower current than to close it in the first place, typical holding current achieved is around one amp. Once you energise the valve the gun is disabled, you can also activate the valve on a disabled gun without needing to enable it. If you want to try re-energising the valve you can also do so without re-enabling the gun, the system will recharge the capacitors before activating the valve again. You can also re-enable the gun but as there is no way to reset a sealed valve until recovered on deck the gun cannot fire again if the operation succeeded.

Improved xdumpsegd plots xdumpsegd now uses 'gnuplot' to do the plotting. This means the plots will resize if you change the window size. You can also zoom the plot - right click on the plot to select one corner of the zoom window, move the cursor to the other corner and click again (any button). You can zoom in further by repeating, to unzoom use the plot menu, this will unzoom both plots if you have added an FFT plot. You can also place markers using the middle mouse button. The cursor position in graph coordinates is shown at the bottom left of each plot. Note using the 'print' or 'save as pdf' options will always produce unzoomed output (and unzoom the screen displays).

Log message deletion facilities - Some common error messages caused by hardware faults can overwhelm the logs obscuring the source performance data. Pending a future complete split of the hardware error messages from the source performance logs two new facilities have been introduced to allow log messages to be deleted before printing. Note both these mechanisms are permanent with no 'undo' facility so need using with care. The first is on the 'log' tab of the display program - see here. The second is a set of command line scripts that can delete either operator entered or several common messages from one or all lines - see here. Any PDF log files already created are unaffected, you will need to re-print them to get clean copies. Use of the 'autoprint' option for the main log file is pointless if you will need to delete error messages after the end of the line.

Corrected SEG-D filter descriptions - Some of the filter frequencies recorded in SEG-D files have always been incorrect. No changes to the actual filters or data result from these changes just the header values.

• Low cut frequencies for sensor and phone data were recorded as 3Hz, should be 0.3Hz but as this cannot be reported in SEG-D the value written is now zero. SEG-D defines this as 'out' which is a reasonable description.
• High cut frequency for 0.1mS sample rate data ('raw') is 4800Hz and has always been correctly reported
• High cut frequency for 0.5mS sample rate data is 800Hz and has always been correctly reported
• High cut frequency for 1mS sample rate data was 400Hz corrected to 280Hz
• High cut frequency for 2mS sample rate data was 200Hz, corrected to 140Hz
• High cut frequency for 4mS sample rate data was 100Hz corrected to 70Hz

gnuplot 4.4.2 - updated for use with new version of xdumpsegd.

synergy 1.3.1 - Added. This is a package that allows multiple computers with (possibly) mixed operating systems to share a single mouse and keyboard. When the mouse goes off the edge of one computer screen it can appear on another screen on a different machine in the same way as it can move across multiple monitors on a single GunLink machine. This is neither supported or tested by Seamap, if you use it you need to support it yourself although it is believed to be a stable package and has been tested by at least one customer.
 
The configuration file '/etc/synergy.conf' may need configuring for your set of machines, you can probably use the same file on all the machines sharing a keyboard and mouse. If you run the synergy master server on a different machine then the configuration file may not be required on GunLink machines. If it is needed then names in this file need to appear in the /etc/hosts file. You will also need to ensure either the client or server daemon processes are started when the X server/desktop starts. You cannot enable them at boot time like a normal service so there are no scripts in /etc/init.d. Ways to auto-start the services for the Gnome desktop used by GunLink are documented in the synergy documentation available on the program's web site. Remember any custom files and modifications will not be backed up or maintained by GunLink backup systems and you will need to document your changes in case of needing to re-install the system.

New Scripts to analyse navigation interface timing - '/opt/gunlink/scripts/analyse_nav_timing.pl' looks at the arm to fire delays and shot cycle times recorded in the database for unsummarised lines. It checks for arm to fire delays that vary by more than a selectable precentage from the average. It also checks for shot cycle time variations that exceed a selectable percentage. Any such shots are listed together with ang log messages for that shot. Finally the arm to fire delay and the shot interval are plotted using 'gnuplot'. Run the script with no parameters to see how to use it and get a list of the line names it can analyse (the ones which have not been summarised). The plot data and analysis are saved in the file '/opt/gunlink/var/gnu_plot.data', you can examine this (large) file with a text editor if you want to analyse things further. The file will be over-written each time you run the script so you need to copy and rename it if you want to look at multiple lines. '/opt/gunlink/scripts/replot_nav_timing' will replot the results of the first script from the saved data file by default but you can specify any other file(s) you have saved on the command line.

master - Fix. Using the 'rear entry strings' option there were incorrect 'sensor data for unassigned channel' and 'not all sensor data received' messages issued (by 'adread') if you had a GFSM fitted with two GFM modules but only a single gun connected. Only the messages were wrong, the system was operating correctly.

nodeAD - Ensure correct time used in node messages if no timing link (or TCU) available at start-up.

nodeAD - Issue warning message if TEM fan controller (GL4K only) stops responding and flag node status dialog data as 'bad'. Warning messages are issued at 15 second intervals. Note this only applies if the controller initially works then dies, if it is not detected at node boot/restart it is assumed not to exist and ignored.
Reject impossible (negative) values sometimes issued by the first revision fan controller firmware.
After an initial fan speed error report increase the interval between reports gradually from 30 seconds to 5 minutes to avoid too many in the log if a fan fails.

nodeAD - Fix. 'failed to see child exit' errors sometimes issued when checking the status of down Ethernet links if the link recovers during the check.

nodeAD Apply stricter checks to string device error messages to ensure corrupt ones caused by bus errors do not cause problems.

nodeAD - Fix. GL4000 only. If you moved a GFM or GPM exactly one bus position further away from the TEM but failed to update the inventory the code that detected unexpected device resets during initialisation was triggered incorrectly preventing bus initialisation. Typically this would happen on the GFM bus if you move a GFSM one cluster or on a GPM bus if you move the GFSM two clusters. The same issue could prevent detection of unexpected resets if the inventory content for the previous device on the bus was wrong although other bus integrity checks would normally find these problems later in the initialisation process. The cause was checking against the expected inventory serial number rather than the actual found serial number of the previous device.

nodeAD - Fix. Sensor autofires for G-Gun sensors were not being reported. There is a 0.5 second autofire inhibit period after gun fire for G-Gun sensors (not for other sensor types) so we avoid reporting autofires when the sensor closes even if this is outside the sensor window. This code was incorrectly interacting with other code which copes with a sample count overflow every six days. This bug has been present in all versions of release 3. You can see the problem by doing a solenoid click on a working deployed gun. You should see both sensor and hydrophone autofire indications, on previous versions of release 3 you only see a hydrophone autofire report.

stats - Fix. If the database was slow due to a lot of un-summarised lines then lines could be created with '1970' start times instead of a default value of 'now'. Normally this would be corrected to the T0 time of the first valid shot on the line but if the line didn't contain any valid shots (typically a test line where the system was disabled with a keyswitch or navigation did not issue closures after the arm event) then the line start time did not get updated. The resulting lines caused the contract archive system to refuse to operate due to overlapping times. Lines are now always created with a default start time of 'now' (when the SOL message arrives) even if the database is slow, so even if no valid shots exist it has a usable start time.

tcucom - Check UPSUC powers off/on within three seconds of operator command and issue error message if not, retry up to three times.

tcucom - Fix. For Motorola M12 GPS receiver type the host decode was wrong if the number of satellites exceeded 9 or the HDOP/PDOP exceeded 9.9. This affected only the informational decodes shown in hardware status or message monitor, the data is not used by the host programs.

tcucom - Add a warning 'DUPLICATED' to the GPS decode messages displayable with message monitor if two consecutive messages have the same time stamp. This has been seen occasionally on the Motorola M12 - see TCU firmware update.

tcucom - Fix. The decoded GPS status message visible in monitor/GPS for the internal Sony SS12 and NMEA data types said 'Fix:poor' when the fix was actually good.

tcucom - Simulate UPSUC status in hardware simulation mode.

tcucom - Send node number and gun string name to UPSUC for it's status displays if it is running firmware 1.4.1 or higher.

tcucom - Fix. Problems with self-update of UPSUCs. Previously which UPSUC (if any) actually got selected for self-update was essentially random.

tcucom - Fix. Bug in capture of external GPS data on the host meant tcucom could go into an infinite loop consuming 50% of available processor power (on a dual core machine) and preventing correct operation of the TCU.

pecom - Fix. Possible loss of navigation commands on GCS90 protocol interface.

pecom - Fix. Fix external header capture size limitation of 255 bytes when using UDP network protocol. Maximum external header size is now 10240 bytes as intended and as possible using serial or TCPIP connections.

pecom - Fix. Fix time stamp in GCS90 headers if the shot is aborted due to lack of fire closure. Previously the date / time was 1970, now it will be somewhere around the shot time - but still not what the navigation system expected as we don't know what the correct shot time should be. Note if the shot is aborted early enough a header may not be produced at all. The problem was usually seen when the fire pulse was extremely early so the nodes did not have time to arm but the TCU did (just) have time to arm.

shotSeq - Fix. Sensor amplitude change warnings. This was based on the largest of the positive or negative peaks of the sensor signal but if you extend the window to show the sensor reset pulse for a G-Gun sensor then the negative peak may be the larger and being more erratic cause incorrect alarms. Also the database value was always the positive peak and so not appropriate if you are using a negative threshold/peak detection method. The system now uses positive or negative peaks as appropriate for the detection method and ensures the detected peak is within a window extending from solenoid start plus the minimum firing delay setting to 20mS after the expected gun fire time. The same value is always used for the change warning and the database (which is used for the graphical displays).

adread/nodeAD - Send explicit disarm command to nodes if some don't report armed or fired for a shot but others do so the entire shot is not aborted. This avoids them staying armed and a misleading 'arm when not idle' error being reported on the next shot.
If one or more nodes fail to report armed status or 'saw T0' status by the allowed timeout an error message is issued but provided some nodes have reported correctly the shot continues. If the missing message(s) arrive after the timeout and no data is lost then the end of shot messages do not tell you which nodes didn't report. Changed to always report which nodes have not reported by the timeout, this may result in a duplicate report at the end of the shot if the node communications do not recover in time but allows you to always see which node is having the communications (Ethernet) issues.

ad module - Lost scan handling. Only try to substitute lost scans (indicating a timing link problem) if only one scan lost. If more scans are lost or appear to be lost there is no point trying to recover the missing data.
Report timing link 'compare' errors where both timing links receive data that passes the parity and stop bit tests but the two are not identical and we cannot guess which may be wrong. The system has always detected these but didn't previously report them.
Report timing link errors as 'errors' rather than 'info' so they appear on logs. But only the first two or three messages every ten minutes will be reported as 'errors' so the log is not flooded in the event of an intermittent link. If you want to monitor the status of suspect links more closely you should use 'message monitor' or one of the other hardware status displays.

nodeAD/shotSeq/tcucom - If the T0 time stamp message from the TCU gets lost for any reason but the nodes fire ensure we get the correct timestamp from a node, previously the headers had a copy of the previous shot time stamp which results in a justified 'duplicate shot now' message from Spectra QC.

monitor - Apply 'search string' processing to 'special messages as well as normal messages.
Enhance search string capability, if the first character of the string is '!' then only messages not containing the rest of the string are displayed.

dandp - Fix. On the hardware status tab the depth and pressure channels on the TCU and any on a GL2000 GCU (including the atmospheric reference transducer) always had status of 'dead' due to some status only used for this display not being correct. They now show the raw frequency count as intended.

dandp - Change depth and pressure simulation so that hardware status tab shows correct displays.

Atmospheric Pressure - Enhanced hardware status tab and GunLink Web live data displays to show the full atmospheric pressure status string. This allows you to see if for example it has been ok but is currently bad or a manual value is being used.

gunlink-panel app - All node status display, don't show dead water temperatures as 'dead' since new style TEMs don't have any transducers fitted due to lack of space on the faceplate. The single node display was changed to cope with this in the previous release. It does mean if you do have a transducer fitted but it's faulty there is no error indication.

gunlink-panel app - Fix - Negative GL4000 TEM internal pressures. The TEM internal pressure sensors are absolute so they read aproximately 14PSI when open to atmosphere. To avoid confusion we subtract a nominal 14PSI to get a roughly gauge reading, that is the displayed value should be roughly zero when the TEM is open or unpressurised. If the TEM is not pressurised this can lead to the displayed value being slightly negative. A problem with signed/unsigned numbers meant these values actually indicated something like 655PSI although the values in 'gunlinkInfo' or the hardware status tab were correct. Note a very slightly negative reading does not mean the TEM has a vacuum or real pressure below atmospheric. A large negative value probably indicates a broken transducer.

gunlink-panel app Show 'golf ball' and data for any extra node entries you create in the configuration whether they are required or not. Previously nodes with no configured channels were only displayed if they were lower numbers than required nodes, or if they booted up and connected. Now extra higher numbered nodes will also show up allowing monitoring of status data for the UPSUC and any data recovered via the reverse timing links. (the changes to do this are actually in the javanet program).

gunlinkInfo - Enhanced. Now shows info for all nodes either connected or defined in the configuration. Also shows UPSUC data (for GL4000) and reverse timing link derived information if the node is not connected by Ethernet. Previously no data was shown for a node if it was not fully booted and connected via Ethernet.

askgunlink - Enhanced. This program is mainly an internal utility for other programs to see the system state but it can be used from a command line - from a remote ssh connection for example - to see if the system is idle. Added a readable status decription to the default help output if you just run 'askgunlink', similar to the existing 'askgunlink status' result.

sendSegd - Enhanced. New (host prefs) option to prevent creation of per line directories on ftp server. If set all files will be transferred to the directory specified by the 'transfer path' setting.

xdumpsegd - Enhanced plots now use gnuplot allowing zooming and stretching of windows.

upsuc_test - Add ability to program board serial numbers (the user controls were always there but non-functional). Fixed some minor display issues.

segdlib - Fix. If you selected the 'always use rev 1 fields' option and recorded a header (either the internal GCS90 one or an externally captured one) then the external header size field of the general header was written offset by one byte resulting in corrupt files. This was a new bug introduced in release 3.0.4. It remains the case that if you need to use the 'always use rev 1 fields' option it means your processing system(s) do not accept valid SEG-D formats.

Bubble Test - Fix exception if none of the selected guns are enabled and you try to start a sequence. Add 'shots per gun functionality, the screen displays will show the calculated data from the last shot fired by each gun if you select to fire more than one shot per gun. Skip past guns without a suitable phone (needs a cluster phone) without waiting. Create a new line (beginning at shot 1001) and suitable log messages for each test sequence. Refuse to start sequence while on-line or in external.

Importing from release 2.5 - Fix. Bug importing host preferences if the system uses a Ethernet to serial adaptor to connect the UPSUCs on a GL4000 system. Causes system to not start correctly with one CPU using 100% as it repeatedly tries to convert the 'system.ini' file.

Log message delete facility - New. The log page has a new button allowing log messages to be deleted. This is intended to allow deletion of for example known hardware errors that are filling the line logs. It is not an ideal solution and should be used with great care, future updates may include a better mechanism for separating hardware and source error logs.To use it enter a suitable search string in the 'search' box which will identify only the messages you wish to delete. To provide protection against accidentally deleting the wrong entries the search string must be at least 10 characters (upper/lower case does not matter), and you cannot use the 'invert' search option. Clicking on 'delete' will tell you how many entries in the selected line match (and display them), there are then further layers of confirm/cancel dialogs.
NOTE: There is no undelete facility unless you have made a database backup immediately before deleting messages and restore that.
NOTE: This facility is protected by an entry in the file / Display Preferences / Host Permissions dialog. The default is DISABLED.

Sensor Single Trace with no hydrophone on cluster - Fix error message which appeared in the 'half tab' error pane when you created a single trace sensor window for a gun where there is no hydrophone on the cluster. The underlying bug also prevented the gain adjust feature working on these windows and attempting to adjust gain caused large numbers of exception errors.

Corrupted Gun Shape drawing - Fixed. Drawing issues that sometimes caused the gun shapes to only be partly drawn or strangely compressed have been fixed.

Depth Transducers - Removed the 'on board' tick box. This was never used and the pop-up help that indicated this is how you specify the atmospheric pressure transducer wrong. You define the atmospheric pressure transducer by placing it on the 'system devices' tab of the layout rather than a string. Note the parameter is still visible if you examine the 'ini' file using an editor but ignored as it always has been.

Device Pools sort by serial number - The devices in the device pool tabs are now sorted by serial number to make finding the one you want easier. This is a string sort as we cannot know the format of serial number you have chosen to use except for GL4000 CMs. This means the resulting order for numeric serial numbers will be 1, 2, 3, 10, 20, 30, 100, 200, 300 for example.

importNewDevices - Updated. Field engineer's tool for importing initial GFSM and GPM inventory into database on GL3000/4000 system installations. Fix missing database name. Enhance serial number format checking. Default to just checking file contents without updating database.

Log message deletion scripts - New. These scripts provide an alternative command line mechanism for deleting unwanted hardware error messages from the logs (mainly for GL4000). The same warnings apply as mentioned for the similar facility in the display program. There are five scripts, the first is a general purpose script that can delete any messages containing a given string from one or all lines in the database. The other four strings use this general script to delete four common types of message from all lines (there is no single line option when using these scripts), you still get to review all the messages that would be deleted and cancel the operation if necessary.

TRecv1p4p11 - Enhanced operation where you have two working timing links (GL4000) and one of them has a multi-bit error which passes the parity checks. This results in a 'compare' error and previously the data was ignored which can lead to a 'lost scan' error. Now the code tries to see if one of the links produced an expected value. Since 15 out of 16 timing words are normally all identical this extra check enables the faulty link to be identified most of the time. Work around for host code node type detection bug in releases 2.5.2-August2009 patch and 3.0.2 and 3.0.3.

This is an important update for GL4000 systems only, install as soon as possible. It has less urgency on GL2000 systems but you should always try to have firmware at the latest release. Note version 1.4.10 was never released to the field. That said this is not a complete solution to intermittent timing links. It is still possible for one badly intermittent timing link to cause node errors even though the other link is perfect. Typical errors are 'lost scan' or time went back', these always indicate timing link errors unless the controller board itself is broken.

Tcu1p4p9 - Version 1.4.8 has been released in previous patches so you may have it, this contained: Fix Julian day sent to nodes. Work around any GPS receivers that issue the same time in consecutive messages, this issue has been seen with the Motorola M12 when not fixing but also very occasionally when fixing with good number of satellites. Fix front panel display of number of satellites if greater than 9. Add some extra checks for buffer overflows when sending messages to the host. In addition version 1.4.9 fixes an issue with Motorola GPS receiver types never outputting data if they are factory fresh or the front panel 'reset gps' option is used. Previously the receiver was told which messages to output if you had that type selected before it was auto-detected. In version 1.4.7 this was removed to avoid sending Motorola format messages to a Sony receiver if you had selected the wrong type. But as a result a receiver that was not configured to output any messages was never detected. The code now issues a Motorola data request 10 seconds after power up if no data has been seen. This should only be required once as the receiver saves the setting in battery backed up RAM.
Note: version 1.4.7 was never released to the field, the previous field version was 1.4.6. Recommended update for all systems, highly recommended if you have version 1.4.8 already installed and Motorola receiver type.

FireModule1p4p4p13 Support for Sercel air shut-off valve operation. Also has modification to disable the high side solenoid driver after the pulse, it gets re-enabled at arm time for the next shot. This may reduce noise and other problems if you have bad leakage by preventing power supply recharge signals reaching the sensor. Not an essential update unless you have Sercel valves.

UPSUC1p4p2 - Fix ability to re-program board serial number. Force reset after sucessful self-update. Changes in release 1.4.1 (never released to the field) - Extra front panel display status menu items to show the assigned node number and gun string name. These are what you have put in the configuration it doesn't necessarily mean you really have it wired to that string. Since the data for these items has to come from the host it is only valid if the UPSUC is in communication with the host. Not an essential update, but if you don't update you will not see the new status displays.
NOTE: Older versions of UPSUC firmware do not reset after being updated which prevents multiple units being updated in one pass by 'updateAllFirmware'. You need to update one, reset it, update the next etc.. updateAllFirmware will detect this condition, tell you and stop trying to update any more units.

Fix node network mask and broadcast addresses - The bonding device defaulted to a mask of 255.255.255.0 and broadcast address of xxx.xxx.xxx.255 rather than useing the values sent by the host dhcp server. This doesn't seem to have been causing any problems but was not correct as the private network uses a 27 bit 255.255.255.224 mask.

Fix log rotation - System log creation got confused after the first time the daily cron job ran. They should now get rotated correctly. To take '/var/log/messages' as an example. New messages are written to 'messages', the previous 24 hours to messages.0 (uncompressed) and the previous day's are compressed to messages.1.gz messages.2.gz etc.. Rebooting may cause rotation to run more frequently.

Fix private network broadcast address - Network config tool. This should have been the base IP address plus 31 but was plus 30 for some reason. Again this doesn't seem to have been causing any problems. Also fix the display of xxx.xxx.xxx.xxx/26 on the interface description should have be /27, the value is not used, it is for information only.

Manual SQL query - This query now has added checks to ensure only SELECT statements can be run. Also fixed to allow queries with quotes in.

Release notes - Allow display of patch release notes if a patch is installed.

Line deleting - The deleting of 'off-line' SEG-D files and directories when the related off-line is deleted failed if you used capital letters for the directory naming fields in 'host prefs'. That is if you had %L, %D or %T rather than %l etc.. You can use either upper or lower case but the pop-up help uses upper case as examples. Note if you change the SEG-D directory naming convention it is possble some unwanted SEG-D files will be left behind as the deletion code has to use the current convention to try to find the directory. Attempting to use field width options with the L, T or D options will also break this cleaning up activity (and is pointless). Finally note using just '%L' as the directory name means all 'off-lines' share the same SEG-D directory called 'Off-line' and it's contents will get deleted at the end of every line rather than 5 off-lines being preserved as normal. Similarly all Bubble-Test 'lines' will share the same directory (although these never get deleted). To prevent this add the date and/or time using %D and/or %T to make all directory names unique.

New greatly expanded Far-Field manual.

New Remote Pressure Display (RPD) manual.

Reformated Navigation interface specification document.

New TCU Jumper setup document. This duplicates data in the installation manuals but has more examples and drawings emphasising the jumpers associated with each signal.

New 'iperf' manual. iperf is a network performance testing tool which is usefull for testing node links.

3.1.0 Updated Operator's manual covering new features.

Updated SEG-D file manual covering 'long phones'.

Updated Moxa Ethernet to Serial adaptor configuration instructions. Added options for use with UPSUC.

08/12/2010

If you have just installed this update then you must restart this computer before use!

There is a new version of the Operator's manual included in this update covering all the new features.
Please replace any older copies both printed and PDF files, or at least mark them 'not latest'.

Rear Entry Strings - For GL3000 and 4000 only. There is a new option in the 'Basic Parameters' section of the array configuration. If you have the umbilical / TEM / IWM connected to the tail end of the strings but wish to have a UKOOA numbering convention with gun 1 nearest the vessel set this. Then build the configuration in the normal way with the cluster / gun / devices nearest the vessel at the top of the page. The device bus address assignments will be recalculated and reversed during an extra stage of file parsing as the host programs are started.

Note: This is not the same as the 'connection side' option which you use to tell the system which way around the GFSMs are fitted, you need to set both parameters correctly.

Remember that the GCS90 header is built with the gun at the top of the first (left hand) string in the configuration as 'channel 1', this is not the same as the gun name or the physical channels for solenoid(s) or sensor which you could set to anything you like in the configuration. Using this option only affects the headers and the normal operational views of the system, diagnostic views like 'hardware status' and 'busMonitor' continue to follow the electrical layout since this is most likely to be of use during fault finding.

There is no migration tool for existing configurations. You will need to create a new configuration with a new name and the string elements reversed.

New Soft Start Features. - The existing soft start mechanism spreads the enabling of guns (or clusters) evenly across the requested time interval. An alternate new mode enables guns in a series of operator defined steps with the steps evenly distributed across the desired interval - so that generally multiple guns are enabled together. A table is used to permit entry of the step at which each gun is enabled to be defined. This allows the operator to define a soft start sequence that conforms with almost any regulatory requirement, for example those requiring increases in energy of 6dB per step. You select which mode is used on the 'options' tab of the 'host parameters' settings since you will not normally want to mix types during a contract.. More detailed information is available in a pop up help window which can be opened from the soft start dialog. Sequenced mode is not available with restricted licences.

There is a new soft start log mechanism. Log files are created for each soft start sequence, these are named by reference to the start time of the soft start sequence since although they usually are related to a specific line the line name may or may not be known at the time the soft-start takes place. The logs contain information on when guns are enabled and the volume fired at each shot when the enabled guns change. Both absolute time and elapsed time (since the start of the sequence) are recorded. The raw log information is stored in text files (in /opt/gunlink/var/soft_start_logs), these are human-readable but you would not normally do so, they are kept in case you need access to the raw data. You can convert them to printable pdf log files using new options on the database tab of display (although they are not database files). The created print files go into the same directory as other log 'pdf' files. You can enable automatic printing of soft start logs via the controls menu as for other log types. The 'raw' data files are managed automatically, you can control how many recent ones are kept from the options section of host parameters, older files are deleted automatically. Once logs have been 'printed' to 'pdf' files they must be archived or deleted manually as for other log files. There is no facility to create soft-start print files from the GunLink Web interface.

Sequenced Firing Mode. A new internal fire mode is added to the existing single array and flip-flop modes. The new mode is selected from the shot control menu as the previous modes. What it does is controlled by a separate menu option on the shot control menu. This allows you to define a firing sequence of up to 16 shots. For each shot in the sequence you can select which arrays will be fired. The configuration window also gives you a duplicate of the gun array membership tables from the 'gun parameters' tab but with the ability to define arrays beyond the 6 normally visible on the gun parameters tab. If you use arrays higher than 6 this is the only way to modify or clear them.

Configuring this mode suitably will enable use of triple, quad and other source sequences as well as more complex ones. There is no option to select the relationship between the sequence number and the shotpoint number as the options are too complex. The first entry in the sequence table will be used when the shot point number is exactly divisible by the sequence length. Remember like flip-flop mode this is purely an internal option unless you have a dumb navigation system, when in external mode the array firing is entirely under the control of the navigation system which selects which array to fire before each shot.

'Bubble' Test - A system for firing guns one at a time to perform what are commonly refered to as 'BubbleTests'. The tests would normally involve recording the near-field hydrophone signals in a series of SEG-D files (which can also be merged with the streamer system data as normal) or possibly as streamer aux. traces. In addition GunLink can process the data to derive the bubble frequency. Using the 'Modified Rayleigh-Wilis' formula it computes a theoretical value of bubble frequency using the measured values of depth and pressure plus the gun volume. Discrepancies between the two values indicate errors in one of the measured parameters or incorrect operation of the gun. More detailed information is available in a pop up help window which can be opened from the control dialog - which is on the 'shot control' menu.

SEG-D Trace header extensions Added the opion to record Sercel SEAL format trace header extensions in SEG-D files. The first main advantage is that if merging this data with SEAL data and the merging system doesn't modify the headers then all traces will have the same size trace headers which although not a requirement of the SEG-D format may reduce processing issues. Secondly they contain an explicit sample count field which may aid processing from disk files. Most of th eother data fields are not used as they are not relevant to gun data. See the revised SEG-D documentation file for details.

nodeAD - Don't poll GL4000 TEM fan controller if it exists while doing firmware updates. Doing so casues self-program of Timing Receiver to fail unrecoverably if it currently has firmware version 1.4.5 or earlier but recent enough to support fans.

nodeAD - The Gun Fire Modules have two temperature sensors, the reading of the first one was being reported twice instead of the second reading from the area of the charging circuit.

nodeAD - Issue regular warnings if an internally pressurised IWM (GL3000) or TEM (GL4000) leaks. Change panel status and hardware status displays so a unit that was deliberately not pressurised is not flagged as a fault.

nodeAD - Add fan failure detection for GL4000 TEMs with internal fans.

pecom - GCS90 header. Force gun 'delay' and 'fired time' data fields to zero for misfires so the faulty gun can be identified. Previously (and for all GunLink versions ever released) the values were held over from the previous good shot.

pecom - GCS90 header. If no main manifold transducer is configured compute header value from string transducers where some guns fired rather than all strings as before. Avoids a misleading low value if flip-flopping.

pecom - If the operator or navigation system sets a new line name while on-line the system should start a new line with the new name. But this (deliberately) didn't happen if the current line name was 'off-line'. This prevented you creating the correct line name if you started a line without the correct name from navigation for some reason.

pecom - Do full SOL processing at first external shot if navigation went online while system was not in external. Previously just went to 'online' but no new logs were opened. This caused incorrect logging if you were (accidentally) not in external at start of line. Note if you need to disable the guns on the run-in just before the start of line, for a noise record for example, it is better to turn them all off rather than switch from external to disable. Similarly do EOL on nav command if online even if not in external in case the operator switches to disable before the last shot has ended.

tcucom - Fix. Aim point time was calculated (from fire time) as one minute late if fire time was before whole hour and aim time after hour. Mainly affects shot time reported in GCS90 header which is 'aim point time' and the time recorded in the general header of SEG-D files. Although aim-time is also recorded in the database all the web tools report fire time. This bug has also probably been present in all GunLink versions.
 
Try to clarify TEM over-temperature warning messages for GL4000, fix over-temperature messages being issued every second by mistake.
Fix Julian day display in TCU GPS menu, was one too big. This is not the same bug as the SEG-D file header issue.

writeSegd - Fix. The 'Julian Day' (actually the day in the current year, not since pre-historic times) in the SEG-D files general header was off by one. So January 1st was recorded as 0 rather than the 1 called for by the SEG standard. This bug will have been presen it all previous versions of gunlink.

gunlink-panel - Fix. GL4000 TEM fan control when more than one TEM with a fan present in configuration. Previously always set controls for the first TEM you ever tried to control regardless of node selected.

gunlink-panel - Remove ability to set manual Gl4000 TEM fan speed as new fans are not speed controllable.

output_stats - Fix. Line names with spaces in them caused line log creation to fail. Changed so created log files have spaces replaced with underscores. The line name reported within the logs is the original with spaces. Add support for soft-start logs.

hwStatus - Fix. The serial number and firmware version fields were swapped in the hardware status display for node depth and pressure cards.

TEM Water temperature sensor - If these are not installed (or broken) it will no longer appear as an error condition on the node status or hardware status displays. A new release of TEM will not have these transducers fitted due to lack of room on the faceplate.

Additional Voltage monitors - The new LX800 processor based GL2000 and GL4000 controller boards have monitoring for additional Voltage rails at 1.2V and 3.0V in addition to the previous 3.3V, 5V, 24V and on the GL2000 plus/minus 12V. Requires latest version of timing receiver firmware.

New features - See previous section.

Line File selectors - File selection boxes on the database tab are now sorted by line so all the available print files for a line will appear together in the list.

USB Serial port devices - Fixed the available /dev names for USB to serial adaptors in 'host params'. Now /dev/ttyUSB0 etc.. Note the current kernel does not include the required module for the FTDI series of chips (ftdi_sio.ko) which is the only tested adaptor at present. However the new R3 standard hardware should have enough serial ports for all normal uses.

Rear entry Strings - See previous section.

Bus Monitor - Add new 'Cable View' window option. Shows graphical indications of either error counts or error rates organised to reflect the physical layout of devices along the array cables on distributed systems. Mainly aimed at GL3000/4000 but works for any system type.

Line Message logs - Searching in 'ALL Lines' was broken, it always returned zero results although searching within a single line with the same search pattern worked. A recent update to the database package was causing the database query to fail rather than make an assumption and carry on as before.

Line Log viewer - Add viewing and deletion of soft start logs. You can only view and delete these files, as with the array and configuration report files these have to be created using the database tab of the display program.

Overdue maintenance warnings - Fixed. These were prone to problems if users manually reset shot counters for individual devices (which you should not do!). Also made some significant speed enhancements to the way this page retrieves data from the DB.

Printing style sheets - Reduced font sizes of serial numbers and removed the line selection dialog when printing the configuration viewer pages. This allows larger configurations to be printed sucessfully but may still not allow printing of the largest configs.. It's not a general solution to printing of web pages which is not really a GunLink specific issue. There may also still be significant differences between browsers in terms of the control they provide over scaling for printing. There is a related trivial change in the gunlink-web package to call up the new style sheet.

TRecv1p4p9 - Support for monitoring 1.2V and 3.0V rails on new LX800 based chassis and TEM controller boards. Fixed incorrect +-12v readings on GL2000/3000. Added voltage filtering. Fix bug switching to internal timing if the only working timing link failed or simultaneous errors occured on both links. Caused a burst of usually 9 concecutive 'lost scan' errors and also bus communication errors. These could also sometimes be seen when the node program started up or was restarted. Fixed bug causing the seconds of node time to jump to an odd number (the day of the month) when they should be zero.

This is a priority update, install as soon as possible.

Log rotation - Fix problem where although the node logs got rotated regularly the node logger was not correctly notified and continued writing to the old file.

PXE Boot support - Added support files to boot PXE boot clients as well as Etherboot. Allows new GL2000 controller boards to boot using their in-built PXE code without needing a Compact Flash card with Etherboot..

New LX800 processor support - Boot script files support detection of new node controller cards based on ETX format modules with LX800 processor.

PXE Boot support - Added second version of node kernel image for PXE boot clients as well as original version for Etherboot clients. Allows new GL2000 controller boards to boot using their in-built PXE code without needing a Compact Flash card with Etherboot. The two kernels are identical, only the file format is different. The initrd image contains modified scripts for detection of the new boards.

New LX800 processor support - Boot script files support detection of new node controller cards based on ETX format modules with LX800 processor.

PXE Boot support - The dhcpd.conf file format is enhanced to support PXE as well as Etherboot clients. Allows new GL2000 controller boards to boot using their in-built PXE code without needing a Compact Flash card with Etherboot. The file format has also been rearranged to move common parameters to header sections rather than repeating them in each client's configuration.

New LX800 processor support - Boot script files support detection of new node controller cards based on ETX format modules with LX800 processor.

3.0.4 Operators manual update.

19/1/2010

Database Update 3.7 - This update is solely to cope with migrated databases from version 2.5.

Database Partition Removal Script - A new script has been added for a one-off use to move the database onto the /var partition and remove the separate partition created by the 3.0.0 and 3.0.2 installation disks. The partition size was too small (3GB) for extended use..

21/06/2009

nodeAD - Ensure the node number is visible in the bonding log messages. Improved user error messages when timing receiver doesn't detect node type, although this should only arise if the timing receiver is dead, or there is something seriously wrong with it's interface to the gate array. Changed some error messages from 'can' to 'node' for consistency.

ad module - Improved user error messages when timing receiver doesn't detect node type (as above).

dandp - "Failed to see fresh dandp data ..." error message. The way the code checks the age of dandp data to decide whether or not to wait for fresh shot data has been changed. There was a corner case if the shot was a few mS before a new second and the host clock was slightly behind GPS time when it would wait although an update had just happened and then time out. Although this resulted in the error message the data in the header was the same as if the problem had not been present and was the most up to date uncorrupted (by shot firing) available.

SEGD - Navigation Line Name - Fixes for Gunlink not setting the correct name for lines based on what is sent from Navigation.

gunlink-panel applet - New 'all node' status icon and display. The icon is five small 'nodes' and appears between the disk space one and the individual nodes. Clicking it produces a table that shows all the status data available on the individual node displays in one large table.

Display Program Redraw - Fixed redraw of transducers in active Display program if you change configuration.

Display Program Redraw - Force redraw if you change host prefs settings that require different menu entries or tab options. For example enabling or disabling far-field.

Dongle Serial Number - Is now shown in the About box on the File menu. This is an internal random number in the dongle and you may need to report it if you have licence file issues.

Comms Error (or "No Data") Default Colour - This status colour (used for signatures) has always been "Orange" but the system Orange is very close to yellow. It has been slightly altered, still orange, but more distinctive.

TEM Fan Status Info - More info is now available on the Hardware Status Tab if a GL4000 TEM has the new power board with intelligent fan control.

Under / Over Monitor support - Improved default window layout if monitors are one above the other. Note you need to manually edit the xorg.conf file to achieve this configuration by changing 'LeftOf' to 'Above'.

Hydrophone Overlay on Single Sensor Frame - The hydrophone overlay would previously lose synchronisation with the sensor signature when zoomed, this has been fixed.

Device Information Box - Fixed the internal links of cluster devices such as the link to a transducer from a GFSM and vice versa.

Clickable Cursor - The cursor will now change to show a clickable status when hovering over table rows that can be selected such as device fault history events.

Maintenance Config - A position column has been added to the device tables on this page. For guns this shows where on the array it is, or in the case of solenoids, which gun they are attached to.

Gate Array Test Module - Bypass tests of internal gate array logic used to interface the 'missing' comms controller on GL4000 boards. Floating pins on the boards could cause these tests to fail randomly, testing of the comms controller itself was already bypassed.

Bus Monitor - Added the capability to log messages to file. Prevent program crashing if you try to change settings while host programs are stopped. Now never attempts to disconnect and reconnect to the message pool, this was no longer necessary and may be cause of problems with messages not being logged seen on some vessels.

Monitor - Underlying code changes to the virtual terminal usage (no noticeable changes to the operator)

Gunlink Firstrun - X monitor detection did not work properly using Philips monitors combined with VGA extenders, resulting in incorrect resolutions. Display options are now fixed in firstrun to be either single screen at 1280x1024, dual screen 1280x1024 or a flexible layout that can be tried with any alternative kind of monitor setup such as 30" monitors or mismatching sizes for development setups. These options currently only apply to Nvidia graphics adapters.

Gunlink Firstrun - Host facilities such as the web server, database server and SAMBA are now correctly deactivated for a Remote Display PC.

Archive Restorer - Fixed a bug where the program would hang on restoring the database.

Shared Libraries - Base code changes for TEM fans (no noticeable changes to the operator).

Database Management Scripts - Minor rewrites to return the correct exit code for different situations, also fixed some bugs in unlikely cases. Also cleaned up the error reporting and removed obsolete code.

PDF Log File Creation Script - Changed the database host to localhost to avoid ambiguity.

FireModule1p4p12 - Fix overcurrent warning messages and phantom overcurrent warning at start-up. Only required for new 'reversible' GFSMs.

Trecv1p4p5 - Support additional messages from new GL4000 TEM power boards with fan controller.

Gpm1p4p4 - Minor changes to reduce possible bus errors with very long drop cables and the new GISMA cable/connectors which has improved frequency response.

Split Surface Reflection Coefficients - There are two reflection co-efficients used internally. One when using the near-field signatures to calculate notional source signatures, and a second separate one when using the notional source signatures to calculate a far-field signature. Previously the operator could only change the second of these in 'host prefs', the one used for notional source calculation was hard coded. If you have source estimation enabled check the new figures match your requirements.

The surface reflection coefficient used for notional source calculation should be adjusted to match the sea conditions. Using too good a value - closer to -1 - will remove too much ghost signal and direct arrivals from other guns. Using too poor a value - closer to zero - will not remove enough. The reflection coefficient used for creating the far-field should ideally also be adjusted to suit the sea conditions if the generated signature is to be correct for data processing. However if you are using the far-field QC settings for peak-peak and bubble ratios to compare with a modelled or library signature then these tests may (correctly) fail if you choose a value lower than that assumed when the modelled signature was created.

Further improvements in processing speed - Processing load is proportional to the square of the number of guns fired on a shot, the latest improvements could be 20% on very large arrays, more realistically you may see another 10% over the improvements in release 3.0.0.

Improved ghost and interaction removal - Algorithm optimisation has improved the conversion of near-field to notional source signatures.

21/04/2009