Schilizzi welcomed all present and opened the meeting.
The following was approved as the agenda of this meeting:
The results of the previous EMU meeting, an email teleconference held around 20.07.1994 were presented by Schilizzi based on notes and memos by Spencer.
Bååth had investigated possible phase cal solutions and come into the conclusion that an analog add-on system will not suffice and measurement after sampling is eventually required. This cannot be fitted into MarkIV formatter, however (see formatter ).
Buiter had found out that the back diodes (BD7) for the total power detector in the new video converter filter boards were readily available. He hadn't ordered them, though, because the newest Alan Rogers design doesn't use them (see about filters ).
Burgess had studied test procedures of DAR equipment and investigated field system alternatives (NASA FS, Darfs.)
Foley concluded that a comprehensive list of MarkIII, VLBA, and MarkIV modes is very difficult to assemble because nobody really knows all the variations that can be devised. There was a comment about a listing of VLBA modes by Alan Whitney (at NRAO), and another listing by Brian Anderson (at Jodrell).
Action: Foley to contact these two sources to include this information in the mode list.
Mujunen had been at Jodrell in May for a week. During this time Linux was installed on one machine and NASA FS 8.x source code was transferred to it. A listing of issues encountered during reccompilation of this system under Linux can be found at ftp://bigbang.hut.fi/pub/emu/fs8RemakeOnLinux.txt .
Parry had measured new video converter filter boards and the results are presented later in the section about filters .
Ruf had investigated the possibility of getting ``second-hand'' precision plates for tape recorders. This is presented in further detail in the section about tape path upgrade .
Schilizzi asked individual teams to present verbally the current status of the various EMU subprojects.
Holland told that the new design that replaces back diodes (see back diodes ) with Analog Devices chips had been received from Alan Rogers at Haystack. Prototype boards should be ready in a week and materials for 160 boards had been acquired. Holland further assumed that 8 weeks would be sufficient for putting the boards together.
Holland continued that he and Parry had measured the performance of new 4 MHz filters. Phase changes seemed linear---the results were circulated in the meeting. The cost of new filter boards was estimated at 87 ECUs each (a saving of 111 ECUs compared to estimates.) A low-frequency network analyzer will be rented for testing the new filter boards.
Buiter reported that new precision chip capacitors for SSB boards had been ordered and received and he handed the capacitors to Holland. Buiter also reminded that previously it had been decided that every station changes capacitors and installs new filter boards by themselves. Holland estimated that new boards would be ready for installation in about 10 weeks. This could enable installation during January 1995. Graham reminded that there is a VLBI session in February 1995 that shouldn't be compromised by ongoing filter changes.
Ritakari warned the meeting about the possibility that changing the capacitors alone might not be sufficient to achieve desired performance out of the modified SSB mixer boards. Graham suggested that this can be checked by frequency sweeping and Ritakari confirmed that by sweeping slowly around a number of different center frequencies SSB mixer operation can be verified. Ritakari stated that changing only capacitors in SSB mixers to attain 8MHz and 16MHz bands apparently eliminates two of the narrowest bands.
Garrett suggested that each station should upgrade one BBC first to check its performance and possibly have it up and running in a session. Graham disliked this idea and recommended upgrading all BBCs at once but not before February 1995 session and not before the 18cm lines experiment. The meeting estimated the duration of this upgrade to be around two weeks per station.
Porcas raised the issue about registering the completion of BBC
upgrades at the correlators (Bonn: mpicor@mpifr-bonn.mpg.de
and Socorro: vlbacorr@aoc.nrao.edu
). This would enable
correlator operators to invalidate assumptions about station behavior
as hardware had been changed.
Graham noted that those stations which do not currently have 4MHz filter wouldn't have them after the upgrade either. Three stations confirmed that they have them and Alef suspected that he would be surprised if everybody didn't have them already.
Action: Hardip to find out whether all Mark III stations (including affiliated telescopes such as Urumqi and Simeiz) already have 4MHz filters or not.
Holland noted that additional filter PC boards have been made that can be populated later. Schilizzi reminded that no funding had been allocated for Simeiz, Urumqi etc.
Action: All stations, after a completed BBC upgrade, please send email to Spencer, Alef, and van Langevelde. (See section participants for an address list.) If your station participates geodetic experiments, please remember to inform Haystack correlator, too.
While searching for optimum timing for BBC upgrade Burgess summarized that assuming that filter boards are delivered fully tested it should take about 1--2 weeks to install them, so it would make sense to start installation at least about two weeks before next session. As the first session in 1995 will start on 08.02.1994 and the second is expected in May, Schilizzi suggested that starting the BBC upgrade around 15.01.1995 at the stations would allow for about three weeks for the installation to successfully complete. Nilsson added that stations participating in geodetic experiments have additional time constraints.
Action: All geodetic stations, find out a suitable three-week time slot for the upgrade.
(Editor's note: When I told about the timing plans to our VLBI friend Kaj Wiik at Metsähovi, he suggested that it might be wise to upgrade exactly one BBC at Jodrell first and then to use it in a session, preferably a CAL one. If something unexpected is discovered, the boards can still be modified before shipping them to the stations for installation.)
Ruf had contacted Jon Romney at Haystack about the new precision plates but Romney had declined to answer.
Ruf explained that the only problem remaining with tape path upgrade is the high cost (about $8000 each) and long delivery time of original Metrum precision plates. In principle new precision plated could be manufactured in Europe, perhaps using old standard spare plates or standard plates that were left over when NRAO upgraded all their VLBA recorders as starting points. Even one spare plate would be sufficient if precision plates were built one at a time and each dismantled old plate was used as raw material for next precision plate.
Workshops need complete drawings of items to be manufactured, however---and this is not the case with the precision plate design. Drawings are apparently Metrum proprietary and recreating them is a time-consuming task for which Ruf wasn't sure time could be found at MPIfR.
Alef noted that there might have been a reason behind NRAO decision to completely replace the plates of their twenty-or-so recorders.
Action: Ruf to find out why this occurred. George Peck at NRAO and also van Langevelde need to be contacted.
Ruf estimated that the number of upgrades needed is around 10--20 and that MPIfR could possibly fabricate about 7--11 of them if detailed drawings were available. He asked for help in making these drawings. Porcas warned about copying.
Action: Ruf to contact Spencer about making drawings.
Alef asked whether old plates really are upgradable. Ruf told that only ceramic and copper inserts are different.
Foley asked if Metrum could be persuaded into upgrading old plates for us. Ruf and Graham doubted this.
Action: Ruf and van Langevelde should ask Metrum and Hans Hinteregger, George Peck, and/or Peter Bolis at NRAO
Schilizzi suggested an action item to check for available mechanical workshop time.
Action: All observatories please check the availability of mechanical workshop facilities in a timescale of one year and report this to Ruf.
Foley asked whether the tape path upgrade will take care of new damper rollers and idle roller sleeves that were introduced in TWG meeting and additionally Alef wanted to know whether EVN VLBA recorders are to be included in the tape path upgrade.
Schilizzi noted that damper rollers ($2000 each) were not included in the original cost estimates and Spencer has to be informed about this. Fortunately only few dampers are needed because apparently for example all VLBA and some MkIIIA recorders already have the new dampers et al. He also mentioned that the inclusion of VLBA recorders was not initially intended but that these might be eventually included. Ruf added that he had had the impression in previous EMU meetings that EVN VLBA recorders would all eventually receive the thin tape upgrade.
Alef and Graham wanted to verify the reason behind damper roller change and it was assumed that tracking would improve.
Ruf stated that it would take two months for Metrum to deliver new damper rollers and Blaschke verified that their installation was a simple matter: three screws and 10 minutes were required for installation plus a calibration tape check.
Action: Ruf orders idler roller sleeves and damper rollers for those stations that have listed him the items they don't have in their recorders. Mujunen suggested that Action: All stations tell Ruf what you need before 14.10.1994 so Ruf can order items early enough.
Spencer had received a message from Haystack that adding an output after sampler for phase calibration electronics wouldn't be possible in formatter. Burgess verified that 8MHz is the maximum sampling speed that can be attained with current decoder hardware and that it probably cannot be easily modified for higher speeds. Bregman had originally devised a new high-speed interface to the current data buffer to relieve the current bottleneck of 9600 baud serial link. Buiter and Foley reminded that the decoder is not needed for phase calibration only but for ``read-back'' purposes as well. Schilizzi asked if we could use VLBA hardware design. Graham probably no.
What is needed can be summarized as follows:
Graham acknowledged that modifying the buffer at 16 MHz is not recommendable. He added that fast reading of buffer into the field system is not actually needed except for getting phase cal readings more often than some minutes apart. Readability checks can be performed with the current slow rate quite well.
Graham added that it should be possible to check recordings by playing back at half speed. Perhaps it would be tolerable if bypass mode and read-after-write cannot be used at 8 and 16 MHz.
Porcas mentioned real time fringe tests: performing them by using for example Internet ``ftp'' to transfer data buffers would actually benefit from >9600 baud transfer rates.
Mujunen, Porcas, and van Langevelde questioned the necessity of frequent phase cal information. PIs frequently have to drop phase cal information provided by Bonn correlator are substitute information deduced from strong calibrator sources. Unless phase cal signal is fed near feed at receiver and special electronics in VLBA style is used, the phase cal readings provided by stations might end up ignored by PIs.
Graham suggested looking into simple modifications to increase current data buffer input rate from 4 -> 8 MHz. Buiter raised his concerns that the design might refuse to work at 8 MHz.
Action: Spencer should do something.
Freiholt had received an email message from Jim Levine (at Haystack) on 03.10.1994 stating that a prototype should be ready before the end of this year. The message politely turned down EVN offers to help documenting and testing the prototype---financial support would be welcomed.
Action: Spencer should continue discussions with Alan Whitney at Haystack whether anything can be done to speed up formatter development and whether EVN help can help during testing and documenting stages.
Freiholt continued that he had been able to find at least two companies capable of and interested in building the formatters, but complete manufacturing documents are a strict prerequisite for them.
Ritakari used the opportunity to remind the meeting that manufacturers of magnetic media (commercial hard disk drives etc.) have since long abandoned the ordinary MFM-NRZM encoding that is used in MkIIIA, VLBA, and MkIV. By simply using RLL encoding instead of MFM a 25% increase in bit density is achieved. Furthermore, current decoding chips utilize PRML predictive decoding techniques to reconstruct bit streams out of less-than-ideal raw data. Looking a little into detail in modern commercial techniques of writing and reading of magnetic media could well result in two-threefold increase of bit density on tape.
Action: Ritakari to prepare a proposal based on his ideas.
Mujunen and Ritakari presented at the meeting:
All eletronic components had been ordered and received. AEC/NIM boxes were being made. All PCB board designs except for the read board were completed and PCBs had been ordered from a manufacturer. Head interface boards (nearest to the actual head) will be produced with original Haystack PCB films although their pad dimensions are not ideal.
Mujunen asked the five station representatives where these new four AEC/NIM modules could be put in their recorders. The responses and the Effelsberg recorder showed that significant variations and thus mechanical problems can be expected.
Action: The five stations receiving the upgrade, please send photographs of your MarkIIIA recorder to Metsähovi. After this Metsähovi will try to resolve a common solution. Although additional 19" AEC/NIM subracks are not included in the R/W electronics contract, Metsähovi will make inquiries about where these subracks could be obtained.
Next Chapter, Previous Chapter
Table of contents of this chapter, General table of contents
Top of the document, Beginning of this Chapter