Ruf welcomed all present and opened the meeting. Mujunen was assigned to take notes for the minutes.
The following was approved as the agenda for this meeting:
Ruf opened the discussion about EMU-related issues. Burgess told us that a Mark IV formatter manual has been ordered from Jim Levine and that there are a few unresolved issues with formatter firmware which are being investigated by Arno Freihold (MPIfR/Bonn).
Burgess continued to explain that a company called ``Spin Physics'' has offered a new kind of ``triple cap'' headstack which apparently avoids the need for a vacuum change when switching between thin and thick tape. Ruf commented that Haystack has already approved these headstacks for VLBI use but that we are waiting for a detailed report on ``triple cap'' performance by Hans Hinteregger (Haystack). Burgess revealed that Jodrell has ordered one with a delivery time of about 60 days. Foley asked if we can afford these new headstacks and Ruf and Burgess answered that the price is about USD 6000.
(Editor's note: Orfei emailed me the title of Hans' memo:
``Spacing loss vs speed in stepped and triple cap heads'',
24-Jun-1997, Mark IV memo #252. The JIVE library has copies of
all memos and if you wish you may ask for a copy from Sandra Mellema,
``mellema@jive.nfra.nl''.)
Mujunen asked if ``triple cap'' headstacks were tested for good operation at 320ips and 18Mbits/s/track, as it is relatively uncertain if thin tape can be used at all at 5 inches of water and 320ips without it starting to ``fly''. The implications of this are that when we want to start recording 18Mbits/s/track (that is, at 320ips) we either must switch over completely to thin tape (and use standard 10in vacuum) or switch over to using ``triple cap'' headstacks (with 10in vacuum for both thick and thin tape). Porcas commented that the VLBA and the Socorro correlator uses thin tape exclusively and that we need to use vacuum switching only as a short-term solution.
Perea asked how it is possible to use thick tape at all at a station which will be primarily using thin tape. JPL has instructed the Madrid DSN station not to use thick tape ever. Ruf commented that the decision to use mixed thin/thick tape in the EVN was made in the March 1997 EVN Directors' meeting and he also added that we know that vacuum switching both increases head wear and decreases data quality. At the same he noted that we have no chance to completely switch over to using thin tape exclusively because of at least the following reasons:
Foley asked if DSN/Madrid will be operational and available in EVN September 1997 session. Perea asked where he can get advice and tapes for September session. Ruf enumerated that for information about the ``triple cap'' headstack he should contact Burgess, for information about thin tape upgrade Wunderlich, and for tapes Ruf. Alef commented that this tape problem is similar to that of EVN stations getting tapes from a variety of sources such as geodesy groups, NRAO, Bonn, Haystack etc. Porcas estimated that DSN/Madrid would get thin tapes directly from NRAO/AOC/Socorro. Ruf reminded that NRAO expect fast circulation of their tapes and Alef added that the amount of tapes DSN/Madrid is likely to need will not probably pose a problem, especially when compared to a station which would typically need 60 tapes or so in a session.
Foley asked what our timescale will be regarding vacuum switching versus ``triple cap'' headstacks. Burgess answered that this depends greatly on money available. Ruf commented on timescales that the first batch of EVN thin tapes was ordered in November 1996 and we still (as of June 1997) don't have any thin tapes ready for use. Wunderlich added that installing the vacuum switching electronics to a Mark IV drive takes a couple of days per drive. The actual timescale was left open.
Wunderlich described the options of adding vacuum switching. We have three kinds of recorders in operation, Mark IIIA, VLBA, and Mark IV. The VLBA recorders already have the necessary electronics for remotely adjusting the vacuum. For both Mark IIIA and Mark IV, a relay switching between two vacuum level adjust potentiometers is the most straightforward solution.
There are two types of vacuum motors in existence, the old model with a variable transformer (AC model) and the new one with a 0--10V control input for controlling the vacuum. Old ones must be replaced with the new type; apparently only Foley has one in Westerbork and Wunderlich promised to supply him the manufacturer and model of the motor (it doesn't come from Metrum/Honeywell).
Wunderlich continued with showing a schematic of a switcher with a TTL input. Graham made a request for the capability to switch vacuum completely off and Alef and Foley commented that this will require two TTL control bits. Buiter asked if vacuum can be controlled only via software and Himwich and Foley acknowledged that this is the current plan. Selecting the correct vacuum according to tape type (deduced from the tape label) is important to avoid excessive wear of the head stack(s). Wunderlich noted that VLBA station software adjusts vacuum automatically according to the bar-coded tape volume label. Graham added that it would be possible to deduce tape type from a given schedule.
Mujunen asked if any other parameters such as the head calibration parameters need to be changed when changing vacuum. This question was based on head calibration results obtained after Metsähovi thin tape upgrade where change of vacuum (with thick tape) seemed to affect reverse-forward offset tens of microns. Graham noted that 5 inches of water is a relatively low vacuum and even at 160ips tape easily starts to ``fly'' i.e. lose contact with the headstack, not to mention 320ips at 5in. Perea asked if there could be any other parameters than head calibration which may have to be changed simultaneously with vacuum. Wunderlich thought these would be enough and added that changes in reverse-forward offset as a function of vacuum at Noto had been of the order of 20μm.
Ruf asked if the (Mark IV) stations want to build the vacuum switch electronics on their own or if they want them to be built centrally. Wunderlich added that in addition to the vacuum switch electronics installation, the cut-off vacuum sensor must be adjusted by turning a screw on it. Its default setting is 6in and this must be lowered to 3--4in. (This is the same adjustment which was also made to VLBA recorders when their tape path was upgraded for thin tape.)
Alef told that the Bonn correlator is running thin tape at 5 inches of water. Ruf noted the Haystack warning about mixed thin/thick operation (it is easy to accidentally wear a headstack to the wrong contour and get unplayable recordings plus excessive head wear) but assumed that switching between 15in (thick) and 5in (thin) is probably relatively ok.
Ruf asked how we should proceed with vacuum switching electronics. Burgess asked where we should get the needed 1/2 bits. Wunderlich promised to provide an unpopulated PCB and Burgess asked if this board would support the ``vacuum off'' option. Mujunen suggested that a recommendation on which two bits of the Mark IV I/O Board should be used for vacuum switching should be obtained from Dan Smythe (Haystack) and sent to Wunderlich for incorporation into his documents.
Alef realized that all tape operations, even prepassing must occur at correct vacuum after vacuum switching is in place. He asked how to prevent 5in to be used with thick tape. Himwich, Burgess, and Ruf agreed that the FS should decide based on the tape VSN (volume serial number).
Mujunen asked if Wunderlich could collect the memo material and distribute it to Mark IV-upgraded EVN observatories. Ruf suggested that the memo and the blank PCBs should be ready before end of July.
Action: Wunderlich to distribute the vacuum switching memo together with a blank PCB to upgraded EVN observatories before end of July.
Ruf asked how we can handle thin/thick switching with Penny&Giles tape drives (for instance in Torun). Buiter said that the JIVE correlator development group runs thin tape at 160ips with standard (mechanical) tension corresponding to 10in. There are some problems with reverse-forward offset and high parity errors. As the tension is adjusted mechanically in P&G drives, Graham felt that these drives need the ``triple cap'' head more urgently than the other stations.
Conway and Alef commented that EVN Mark IV drives are untested, uncontoured, and operators ``uneducated'' with thin tape and that time and resources must be allocated to this as well. Alef suggested (initially only for Torun) that that they should get one thin tape (presumably from NRAO), record it according to MPIfR/Bonn instructions and then ship it to Bonn for a playback test.
Alef, Foley, and Perea asked if there are operator instructions or a tape handling manual which addresses thin tape handling. Wunderlich noted that handling instructions can be found in Haystack VLBA/Mark IV memos. Perea asked if there is an EVN-specific tape handling manual somewhere and the answer was no. Alef told us that there is a NRAO web page about tape handling http://www.nrao.edu/vlba/html/TAPES/tapes.html. Conway asked if there are any significant differences between handling thick and thin tapes. Alef and Ruf answered that not really, the tape is just more prone to damages and thus requires even more careful cleaning of the tape drive than thick tape. Prepassing and postpassing (see the above URL) are also more important since shipping causes more uneven packing on reels than what it does with thick tape.
Himwich suggested disabling the Mark IIIA/Mark IV drive ``Load/Stop'' button so that a tape cannot be manually loaded with (potentially) wrong vacuum. The FS can refuse to load the tape unless the VSN (tape label) has been entered (or scanned in with a bar code reader) and the FS can thus select proper vacuum for this particular tape. Porcas added that with thin tape we are always talking about high-density recording (56250bpi with Mark III/Mark IV DR format, 56700bpi for VLBA NDR format, see section bitdensity ).
Ruf enumerated EVN-affiliated stations which will need the thin tape upgrade, perhaps in conjunction with VIV (VLBA to Mark IV) project:
Ruf summarized that the ceramic plates have already been shipped to Yebes, they just need the other parts and Leslie Perry (Jodrell) to do the upgrade. Burgess promised Les is available for this installation. Wunderlich asked whether to ship parts directly to Yebes or to Jodrell. Ruf suggested that Les should collect the tools necessary while upgrading Cambridge.
Ruf asked what we should do with the Chinese stations and it was decided that Wunderlich sends email to Shanghai and Urumqi to discuss options.
Porcas asked how we are going to test the upgrades and Ruf suggested that a VLBA-style thin tape be written at every upgraded station and then shipped to Socorro for playback quality analysis. Mujunen added that the stations should immediately switch to 15 inches of water for ``normal'' thick tape operations to let the head recontour to the more pointed shape required by thin tape at 5in. He also asked how much is enough for recontouring the head and Alef answered that exact times/tape amounts are not known. Fuji H621 (without the -I suffix) is know to be abrasive and it is also called ``lapping tape''.
Action: All stations planning to use vacuum switching: adjust your tape recorder now for 15in operation with thick tape.
Action: Perea remebered that he had got an email about suitable recontouring tapes and he promised to email it to Ruf for distribution.
Action: Ruf to ask NRAO for thin tapes for verification of station compatibility with the VLBA correlator. Porcas suggested that this can be actually assigned to Mioduszewski.
Action: All stations with thin tape upgrade: record (RF/IF) noise on this thin record quality verification tape and send it to Socorro for analysis.
Graham described tests he had done with the Effelsberg VLBA recorder at 5in and 160ips. Some tracks resulted in parity errors as high as 1E-1. This resulted in discussion if the quality test should be made at 80ips only. Porcas reminded us that VSOP recordings are to be made at 160ips. Alef noted that the VLBA sets up a fake experiment when they do recording quality checks.
Burgess searched and found a more precise price for the ``triple cap'' headstack by Spin Physics, USD 7627. Alef reminded that especially with thin tapes the humidity in the headstack area of the recorder must be kept below 50%. The recommended value is 20--30%. Himwich noted that less than 5% is not useful either. Foley, Buiter, and Tuccari asked about the Urumqi serious head wear experience despite their about 5% relative humidity.
Tuccari explained their experiences at Noto. They have just changed in a new headstack, their third. Their ambient relative humidity is about 40% and inside the recorder, thanks to the dry nitrogen system it is less than 30%. No such system was installed for the duration of their first headstack which lasted for about 1000h. Their second headstack failed after 800h but in a different manner: one track is missing but the headstack is otherwise in good condition. They have just started to use their third headstack and cannot comment much about it yet.
Burgess wondered if the expected lifetime of a given headstack is 4000--5000h and asked if people with dry air/nitrogen kits are experiencing problems with temperature or head failures. Nobody had anything to comment.
Ruf explained that Arno Freihold (MPIfR/Bonn) has found approximately 16 items in Mark IV formatter firmware which need attention. Arno had described these in an email to Ralph Spencer (Jodrell), explained Burgess and the consensus was that local firmware expertise must be developed and that Arno Freihold and Burgess could perhaps participate in this.
Foley asked if there are any urgent upgrades necessary before the September session. Ruf did not believe this would be likely, although he asked about the time jump problem. Ruf and Porcas wondered if the time jump problem could be specific to Westerbork Mark IV formatter, but Mioduszewski confirmed that all Mark IV formatters exhibit the same time jump problems.
Ruf summarized that there are problems but we cannot do anything to them within such a short timescale as the September session. Ralph Spencer (Jodrell) is working with Arno to find a solution to the firmware problems. Ruf asked if there are any urgent known bugs and this was a natural place for Mioduszewski to present the results of Mark IV ``all-modes'' tests.
Mioduszewski described the results of ``FT1'', a test experiment held in April 1997. Five antennas participated:
HN (Hancock), JV (Jodrell Bank VLBA, sometimes known as Cambridge), NT (Noto)
JB (Jodrell Bank Mark IV), MC (Medicina), WB (Westerbork)
``FT1'' was observed at 1.6GHz and it tested 42 modes. The tapes were correlated at the VLBA correlator.
The results were encouraging, 34 modes were validated with fringes found and cross-power spectra looking ok. There were problems with poor recording, severe interference, and with 1MHz external filters (this bandwidth shows up as ``0'' in logs; it did not fringe at JB). There is also a peculiar delay jump associated with every change in speedup factor.
``FT2'' will be observed at 5GHz to avoid interference on June 30, with:
HN (Hancock), SC (St Croix)
JB (Jodrell), ON (Onsala), WB (Westerbork)
Graham asked if external 1MHz filters are to be used in ``FT2'' and Mioduzsewski said no. She also explained that whenever the speedup factor changes, the Mark IV to VLBA delay jumps. Conway asked if GPS timing data was relied on in correlation and Mioduzsewski said no. Perea asked if GPS timing data is required and Altunin said DSN stations already have the data available. Porcas found the ``FT1'' test results good and Graham asked about the bandpasses of 16MHz Mark IV VC filters. Mioduzsewski replied that they look as bad as before. Graham and Alef wanted to know if 16MHz will be used in ``FT2'' since they have changed some details in the filters at Effelsberg. Mioduzsewski said that ``FT2'' will include 16MHz filters.
Conway asked about relatively poor playback quality performace of stations and whether it is normal. Mioduzsewski replied that unfortunately this is the case. Alef asked why there are no readback tests in schedules in general. Ruf suggested that this item should be moved to tomorrow into section netops . Burgess asked about high parity error rates of Jodrell VLBA and Mioduzsewski told him that whole tracks or BBC IF input data seems to be missing.
Mioduzsewski concluded that ``FT2'' will be almost identical to ``FT1'' (with the exception of external 1MHz filters). Altunin asked if DSN stations should/could participate in this test and Conway noted that this test tries to mainly verify the correct operation of the Mark IV formatter design and firmware in all modes, not particularly the performance of any given antenna.
Porcas asked when ``FT2'' will be correlated and Mioduzsewski expected it to happen within one month of observing. Ruf asked if this test is enough, i.e. is this really an ``all-modes'' test, but nobody actually answered this question. Gurvits asked how to notify users of the EVN about the new Mark IV modes available and Porcas suggested there is a file about this in Socorro. Gurvits asked where PIs can find SCHED setup files and Porcas assumed Huib van Langevelde (JIVE) and Craig Walker (NRAO) have been / are designing these. Mujunen asked how we can ensure that pointers to these resources can be found in the EVN proposal guide.
Action: Porcas to ensure that pointers to EVN station Mark IV capability status table and SCHED setup files for Mark IV observing can be found in the proposal guide.
Conway noted that only 4--7 new Mark IV modes would be enough, i.e. satisfy the needs of most users. Alef suggested even a lower number, 2--3, but cancelled this after realizing that spectral line observations may benefit from a couple of additional modes.
Ruf asked what was felt the most important problem in the Mark IV formatter design to be corrected, according to the ``all-modes'' test observations. Mioduzsewski thought that delay jumps is the most serious problem. Conway noted that for geodesy observations the setups don't change during a given experiment and thus there is less pressure on the geodesy side to have the problem corrected at all.
Burgess introduced (on behalf of Ralph Spencer, Jodrell) the EVN VLBA to Mark IV upgrade project. The team will include Ralph Spencer as the project leader, Arno Freihold (MPIfR/Bonn) and Dan Smythe (Haystack) working on the Mark IV formatter, Burgess to work on FS (together with Himwich), Ritakari to work on read/write electronics, Huib van Langevelde (JIVE) to work on scheduling and FS support, and Foley to work on formatter/FS issues.
Ruf asked about read/write electronics required to upgrade a VLBA recorder to Mark IV. Ritakari replied that to increase the bit rate from 9Mbits/s/track to 18Mbits/s/track the existing Write Driver module must be replaced. However, adding a second headstack will require at least the following items:
Mujunen added that the only headstack of a VLBA recorder is a read/write head and it is fitted in head position #1. In Mark IV recorders the r/w stack is at position #2. Whether this causes a problem or not has to be determined.
A device similar to the Mark III decoder/data buffer will be eventually required since the data buffer and the DQA (data quality analyzer) of the VLBA formatter will disappear along with the formatter swap.
Ruf and Burgess confirmed that the VLBA formatter is really to be removed and replaced with a Mark IV formatter. The precise way of doing this is unclear at the moment, however. Mujunen asked who is designing this and Burgess was unsure. It was an assumption that Haystack must be doing this but nobody knew anything for sure.
Mujunen reminded that only the two highest-speed Mark IV modes 512Mbits/s and 1GBit/s cannot be attained with the VLBA formatter and one headstack. Replacing the formatter and the existing write module (but not adding a second headstack) brings us the 512MBits/s mode, provided that recording quality at 320ips and 18Mbits/s/track can be managed. (Practical correlation of 18Mbits/s/track modes also requires a correlator and a reliable playback channel which can play back at 18Mbits/s/track. Graham noted that Mark IV modes can be correlated at the Bonn correlator in multiple passes and with a slowdown factor of four (18 to 4.5). Alef did not particularly like this idea.) Adding a second headstack adds only the 1GBits/s mode (or if 18Mbits/s/track won't work, then a two-stack 512MBits/s mode).
Ruf suggested a planning / ``kick-off'' meeting to be held in September.
Action: Burgess or Ralph Spencer to organize this ``VIV kick-off'' meeting.
Ritakari commented the upgrade of read/write electronics as follows:
As far as I know, existing Read and Write modules can be used for 9Mbits/s and slower speed recording. The only necessary modification is the addition of a second write module. We have already produced several batches of this module, making some more is relatively straightforward.
Some of the old write-modules use transformers to sample data for decoding. I am not sure if these can be upgraded for 18 Mbit/s operation or if they need to be replaced with new write modules.
Adding another head is more of a problem: We need the headstack, inchworm motor, LVDT transducer and the head electronics, including the hard-to-get headstack connectors.
Some Mark III -style head electronics were left over in the EVN Mark IV upgrade. We should consider the possibility to convert these to Mark IV by the individual observatories.
Ritakari continued to comment on the decoder and data buffer:
These are being developed in Haystack and it is very difficult to estimate the work involved, because no data is available at the moment.
In principle this module should be very simple and inexpensive. Optimal solution would be a IBM PC/AT ISA board with line receivers (connected to Mark IV formatter sampler outputs), synchronisation logic (fits in one seven-dollar MACH 211 PLD chip), and 64 megabytes of 10ns SDRAM (one DIMM module, available at the nearest computer shop) and of course ISA bus interface logic. This is completely adequate to collect 8 megasamples of data on all tracks and the PC main processor would be used for data quality checks and/or correlation.
Ruf explained that we are now facing a transitional period of mixed thin/thick tape operation and requested suggestions as how to minimize the problems associated with this. Foley suggested minimizing the number of changes within a given session but Conway and Porcas felt this was a nightmare for the scheduler. Ruf commented that an ideal arrangement would be separate thin/thick parts within each frequency. Alef wished for automatic vacuum changes and Himwich reminded that every change wears the head a little. Alef suggested using ``triple cap'' heads as the final solution. Ruf suggested scheduling around VSOP observations. Gurvits asked how much thick tape is expected in September and November sessions. Porcas noted that all VSOP observations will be done on thin tape as they are destined to Socorro for correlation. Gurvits added that this will happen to most global observations, too, so that leaves us with EVN-only observations. Porcas commented that the Program Committee will discuss these, but that some multi-epoch proposals will surely appear in both sessions. Gurvits noted that the less we have thick tape observations, the less we will have backlog in Bonn correlator.
Hase asked how many thin tapes can be made available. Porcas stated that they will have to come from NRAO. Conway asked if we have to manually adjust the vacuum in September session. Porcas and Foley claimed this would be dangerous. Ruf asked how certain is the VSOP schedule and Gurvits answered that it depends on the progress of in-orbit checkout. The VSOG has most probably created an optimistic schedule and some experiments will most probably be cancelled, but nobody knows for sure.
Porcas asked if we could preallocate the ``right tapes for right experiments'', properly prepassed (Alef). Ruf asked if the tapes could be prelabeled and if thin tapes purchased by Hase's institute have shown any signs of damages. Altunin reiterated that DSN stations will use only thin tapes and that no operator training can be provided for handling thin/thick switching. Ruf concluded that no real conclusion on thin/thick tape operation can be found since the requirements are conflicting.
The meeting had a lunch break.
After lunch Ruf asked if there was anything to add to thin/thick operation. Porcas asked what needs to be informed to PIs about thin/thick tapes and Ruf noted that all VSOP and global experiments will be recorded on thin tape (and correlated in Socorro) and that all EVN-only experiments will be recorded on thick tape for the transition period and correlated in Bonn. Graham and Alef reminded that at least the DSN stations are thin-only. Mujunen asked who will collect information about thin tape capability and make it available to PIs.
Action: All stations: Send updates of station information
(especially including when thin tape is available on your
station) to Porcas
(porcas@mpifr-bonn.mpg.de).
Alef requested updating of LOs and IF patching information at the same time. Ruf noted that DSN/Madrid must be listed as thin-only in the status table and that soon all other EVN stations can be listed mixed thin/thick.
Action: Porcas to contact Mike Garrett (JIVE) to arrange for station status table updates in JIVE WWW and PI information.
Porcas asked for a deadline for completing testing thin/thick switching and thin tape recording quality test. Ruf suggested before start of September session. Mioduzsewski told us that an engineer will look into the data quality on test tapes at AOC and this will not require VLBA correlator time.
Ruf described how efforts to buy thin tapes for EVN started in 1995 but how regardless of an early start we are still somewhat ``stuck''. A new (slightly cheaper) supplier Quantige(sp?) (former Ampex) has been found and their tape quality is being assessed by Haystack and NRAO. NRAO/AOC (Socorro) has promised to help in repacking future tapes onto glass reels.
Hase noted that the geodesy community needs thin tapes for its CORE program. This would mean about 350 thin tapes, acquired by US Naval Office, NASA, and geodetic institutes all around the world. Hase suggested the possibility to investigate a consolidated buy. Alef told that in the latest teleconference with recorder folks of Haystack and NRAO, Hans Hinteregger (Haystack) told that Haystack is coordinating the purchase. Porcas asked what is the timescale for the geodesy community to start using thin tape. Himwich commented that (due to correlation in Socorro) some stations can become thin-only and that among EVN stations, the participation of Onsala, Noto, and Medicina would be appreciated. NASA is upgrading Fairbanks and Kokee Park in September 1997.
Buiter was concerned about the availability of thin tapes in the future since in the video market there is zero demand for this kind of tape. Ruf replied that this is why we need to buy all the tapes we will eventually need now while this tape type is still available. Buiter asked what are the (supposedly ``extra tight'') tolerance specifications for the glass reels and Ruf assumed reliance on NRAO specifications.
Porcas reiterated the question about geodesy timescales concerning thin tape and Himwich replied that he wasn't sure. The CORE program starts before either JIVE or Haystack Mark IV correlator is in production use, so this necessitates correlation at Socorro and this implies thin tapes. Thin tapes probably won't be introduced in 1997. The geodesy community has acquired some thin tapes and Arthur Niell (Haystack) knows the details. Ruf asked how many tapes there were, 100?, and Himwich answered ``perhaps''. Himwich also stressed that for geodesy-only stations they are not willing to risk mixed-tape operation and that their ``own'' stations will switch over to thin tape once they have to start using it.
Ruf concluded that due to practical necessity EVN has to bear with a (hopefully short) transitional period of mixed thin/thick operation. He also recommended that EVN should take advantage of the NRAO offer to repack new thin tapes.
Action: Ruf contacts Arthur Niell (Haystack) to get details of the thin tape purchase plans of the geodetic community and to discuss the possibility of combining purchases.
Conway asked how many thin tapes will be needed. Ruf explained that roughly a thousand tapes would cover a year of correlator backlog. Alef noted that we must stop observing more than what we can correlate. Porcas estimated that EVN would require typically about 200 tapes for each session.
Altunin reminded us that the ``Caltech'' correlator still exists and that there is a relatively low demand for it. It is a four playback drive correlator and it will be upgraded to thin tape operation in 1998. He expects this ``availability offer'' to be valid at least to the end of 1998.
Ruf asked if Wettzell could join EVN thin tape purchase and Hase suggested that as Ruf contacts Arthur Niell, Hase contacts his coordinators.
Burgess asked about the expected life of a given thin tape and Porcas explained that this far NRAO has lost about 50 of their roughly 1000 tapes.
Wunderlich explained the repacking procedures needed to transfer new thin tapes from their metal shipping reels to ``self-packing'' glass reels. This process started at Bonn roughly in June 1996.
The Acrometal self-packing reels are actually narrower than 1in tape. Empty reels are usually 0.984" (25.0mm) and with a properly packed tape, the tape widens the reel up to 0.996" (25.3mm). A phenomenon called ``flange forcing'' has happened if the tape has forced the flanges of a given reel wider than 1.000" (25.4mm). These changes in reel dimensions are possible because the glass flanges are made of ``security glass'' and they are actually flexible.
The procedure for repacking goes like this:
After these steps approximately 2--3 tapes were removed due to persistent bumps and spokes.
Thin tapes repacked with these procedures were used in VLBA/Effelsberg experiments, together with tapes from NRAO. The Socorro correlator complained that EVN tapes were ``not in perfect condition'' whereas VLBA tapes were ok. The tape reels of Effelsberg VLBA recorder were adjusted to be aligned in the same plane but still in another VLBA/Effelsberg experiment the same degradation happened to EVN thin tapes and not to VLBA thin tapes. Repacking of further EVN tapes was stopped at Bonn.
Wunderlich continued to explain that the reasons for ``forced flanges'' usually lies in the tape path: tape is pushing towards one flange due to a misalignment. The fact that adjusting the Effelsberg recorder didn't help meant that another cause had to be found. The 3M/Scotch tapes EVN got were inspected and measured against Sony tapes which NRAO usually uses. EVN tapes seemed to be approximately 30μm wider and the tape edges were not symmetrically cut, i.e. the cross-section of tape edge showed a non-perpendicular tilt. Six tapes were shipped to Hans Hinteregger (Haystack) for inspection and Wunderlich continues discussions with him.
Editor's note: At the time of the meeting the cause for tape damage was unclear. It was suspected that either the difference in tape edges of 3M vs Sony or any improper handling could be the cause. Later 3M admitted that the tapes were defective and they agreed to take them back and refund the price. (3M doesn't make any new tapes of this type, thus the refund and not a replacement with new tapes.)
Buiter asked what kind of take-up reel was used in Effelsberg recorder and Wunderlich replied that a standard blue non-self-packing one. Buiter continued that JIVE has similar packing problems with P&G drives but that P&G recommends using a self-packing glass reel as the take-up reel. Buiter also asked if the manufacturer (3M) has been contacted and Wunderlich explained that this will be done after Hans Hinteregger has had a chance to inspect damaged tapes. Mujunen asked if unrepacked tapes were also sent for inspection to NRAO and/or Haystack and Wunderlich confirmed this was the case, NRAO got six tapes and Hinteregger one ``good'' and one ``bad''.
Alef asked what are NRAO's experiences with 3M since they also have 3M tapes. Ruf also noted that they keep statistics of failed tapes and that they had said that there ``may be a tendency'' that 3M tapes fail more often. Wunderlich added that the tendency had been observed by ``cut batch by batch'' basis.
Buiter confirmed that there were similar problems with P&G drives in JIVE and that he had heard that George Peck (NRAO) has widened self-packing reel width specifications. Alef commented that Hans Hinteregger (Haystack) considered this wider specification unsafe for tape shipment. He also noted that it is possible that self-packing reels used in Bonn repacking may have been of the wider specification but that this should only influence transporting tapes.
Ruf concluded that tape experts are currently working on the problem and that EVN should consider and prefer NRAO help in future repacking of new tapes.
Alef described the current status of EVN tape logistics as
``improved''. WWW-based ``Track'' is being used by almost
all stations (see red ``TRACK'' button on page
http://www.nrao.edu/vlba/html/VLBA.html, password
requests to ``vlbatape@nrao.edu''). Tape transit
times have shortened and are usually a few days, less than
a week except for Shanghai which takes usually one week.
Also station bar code labelling is improving as
``drudg'' can now generate PostScript bar code labels
and matrix printers with faint ink ribbons have all but
fallen into disuse.
Foley complained that in the past Web-based Track didn't accept all tapes, but Alef ensured that this is not the case anymore. He added that geodetic stations are also using Track. Wiik asked if CMVA (Coordinated mm-VLBI Array) tapes are to be entered into Track and Alef thought there should be no exceptions. He also recommended that:
Action: Himwich to encourage geodetic stations to
use Track, either the Web or the terminal version.
(Contact ``vlbatape@nrao.edu'' for new station
arrangements and passwords.)
Foley asked Mioduzsewski if there were any comments from the VLBA correlator regarding tapes and she answered that tapes arrive usually in time and that ``most stations'' use Track.
Alef stressed the importance to attach bar code station labels to every tape which gets shipped to the correlator. Polatidis commented about getting ``FedEx'' tracking labels ``for free''. Buiter asked if thin tapes should be bound in bundles or cardboard boxes for shipping and Alef and Mioduzsewski agreed that the best way is to ship tapes separately in their own (2in thick foam-filled) plastic transportation boxes. NRAO has been successfully shipping thin tapes individually.
There was some discussion about tape labelling, Himwich asked if anyone is still using ``Epson'' matrix printers for bar code labels and Alef discouraged this. Polatidis complained that since green base labels are reused multiple times (although they originally were designed to be replaced at the correlator when tapes are degaussed), they tend to peel off.
The meeting had a coffee break and decided to handle a couple of items from Saturday's agenda.
Porcas presented the concept of ``target of opportunity'' proposals. These observations are only carried out if/when a precondition, a predicted astronomical event actually happens. The typical delay from a given event to the corresponding observation would be of order of 4 months. A question posed by Porcas was if the EVN can react much faster, would it be possible to react, say, within 4 days?
The consensus of the meeting was that ``TOO'' proposals can be included in the EVN block schedule and that under a relatively short notice observatories can omit them from the session, i.e. not observe them if the corresponding event has not occurred. However, the preparatory work, schedule files etc. must appear according to the usual EVN deadlines for ``TOO'' experiments to be included in sessions.
Action: All stations: list the boundary conditions and timescales for reaction to target of opportunity proposals.
Porcas continued to depict the EVN ``Call for Proposals'' which
gets sent one month before the proposal deadline. It
includes the EVN station status/capability table
(``evn.sts'' in ``astbo1.bo.cnr.it''). It is
the responsibility of stations to send updates whenever
their receiver availability, Tsys (Jy), and recording
system capabilities change. At the moment Porcas
will collect this information at
``porcas@mpifr-bonn.mpg.de'' and make sure it
gets updated in the tables which gets distributed to PIs.
After the PC has selected proposals to be observed and after the block
schedule has been created by the EVN scheduler Rolf
Schwarz (MPIfR/Bonn), the prospective PIs get additional
``PI instructions'' sent to them. This information
includes pointers to scheduling software (PC-SCHED and
SCHED) and setup files. Any changes to station LOs, IF
patching etc should also be sent to Porcas who
will collect this information at
``porcas@mpifr-bonn.mpg.de'' and make sure it
gets updated in FTP servers accessed by PIs.
Burgess asked why so few experiments properly use the ``cover
letter'' anymore and Mujunen and Gunn requested
that at least phase cal on/off and LCP/RCP usage should
appear on the cover letter as PC-SCHED/``.drg''
formats don't support indicating these---eventually the
new VEX schedule format will include pcal and
polarization. Polatidis also complained that
schedules often arrive late at ``astbo1.bo.cnr.it''.
Mujunen asked if there are any circumstances when EVN
stations should look at files found on
``aspen.aoc.nrao.edu'' and Foley, Mioduzsewski,
Graham, and Porcas all agreed that all schedules
intended for EVN stations will appear on ``astbo1''
and that they can deposit the corresponding log files and
GPS data also there.
Mujunen asked about unified IF patching for both Mark IV and VLBA racks. Alef reminded the meeting about the automated IF switchbox which was a topic in the previous TWG. The stations which would benefit most of such a box are those regularly participating in both geodesy and astronomy (Medicina, DSN/Madrid, Onsala, Wettzell), since astronomical dual polarization IF patching is very different from geodesy patching. Perea noted that JPL has created an automated box for IF patching for DSN stations. Hase asked if the box was commercially available and Altunin answered that he is not sure.
Action: Perea to send JPL IF switchbox capabilities documentation to Ruf. Altunin to email possible Web references about the box to Ruf.
Alef presented a preliminary May/June 1997 CAL report. The full report can be obtained at ftp://ftp.nfra.nl/jive/monitor/cal may97.ps.
A worrying problem was that at 6cm half of the stations got LCP and
RCP interchanged. At 1.3cm the calibration sources were
quite weak. Graham noted that the usual CAL team has
left us and this has put the extra burden of creating CAL
schedules on Conway and Polatidis. Gurvits
suggested using another source at 1.3cm as the fringe
search candidate. Alef requested long enough gaps in
the CAL schedules to allow for readback tests to be
inserted by ``drudg''. Himwich added that this
requires only 70 seconds after a given pass.
Altunin asked how often EVN arranges CAL observations and expressed their interest in participating them. Ruf replied that CAL observations are included in every session, that is every 2--3 months. Porcas added that a CAL run usually lasts for a day or slightly less per each observing frequency and that the schedule should be known 6 weeks before the observation.
Action: Ruf to send the block schedule to DSN/Madrid as early as possible indicating especially the CAL observation.
Himwich noted that automated tests in FS/``drudg'' have been
tested to work with Mark III mode C only and that it is
possible that they require further development to be
generally applicable to any VLBA and/or Mark IV mode.
Porcas asked if the stations felt that cover sheet information in the distributed block schedule would be helpful and at least Gunn acknowledged yes.
Action: Gurvits to ask Rolf Schwarz (MPIfR/Bonn) for an ASCII version of the block schedule and to arrange that it gets updated to JIVE Web pages.