Valhalla 2720GS

[jfphp]

Hello,

I was very happy with  a Valhalla DC standard 2720GS but unfortunately it died in two steps : I lost the differential voltmeter ("DVM repair" as on the one offered now on Ebay) and now "Reference repair". After a lot of tests including the various stacked power supplies, I suspect the Pal deconding chip on the CPU, an EP900. Did anybody read it ? I can read it on a chip programmer but, of course, I down know if it is a good or wrong JEDEC file.
Another question : does anybody know the default password to reset the 2720GS to the factory calibration default state ?
Thanks.

[CaptnYellowShirt]

I don't have any information on the 2720GS.

But I'd love to see some pictures of it! (And so would other people that might be able to help more than I can)

[TiN]

Same here, one of few units using statistic voltage references.

That pal is like simple version of CPLD, so it's just couple logic gates in there probably.
In theory should be possible to replace it with some bodged logic circuit, if you can find out what are input signals, what are outputs and what
function that PAL was supposed to do. These may have security bit, so even if somebody have working 2720GS, they may not be able to read it.

Tear-down hi-res photos would be gold.

[Vgkid]

I would also require pictures.

[jfphp]

http://www.adret-electronique.fr/valhalla_2720/adret_valhalla_2720.html

Some pictures and documentation  before the crash...

[CaptnYellowShirt]

" This system's internal 30 bits A to D converter continually monitors each reference zener. Each reference module contains its own non-volatile memory wich stores calibration and drift rate data."

This is something I've wondered about quite a bit. What does the A/D compare the A to in this configuration?

I mean, at its core an A/D only determines ratios.... "so much more" or "so much less" than V_ref. 

But if you're measuring something that's *the most accurate* already... what's more accurate than that?

My guess is the A/D compares the zeners to one another -- in a bit of a kilogram divergence problem:



So can this make the reference more accurate? Is there some statistical magic or model for zener drift that helps divine the true value of the volt? (e.g. if they all drifted the same amount, how would you detect this?)

[jfphp]

It is exactly the same problem with the Fluke 732A : you need 4 to be sure of the respective drift of one to another (therefore 4 references in the Datron 4910) : you compare each day or week ... 1 v 2, 1 v 3, 1 v4 , 2 v 3... In comparing the various drifts you can isolate the (eventually) faulty one (drift exceeding the mean value) and you can take as reference the mean of the 4 or 3 if one faulty and minimize noise (reduction of noise of root square 6 or . Superiority of the Valhalla is the computerized mesurement of the respective drifts in comparison with the drifts stored in EEPROMs of 6 zeners  or 8 as an option.
In the high stability option the 6 or 8 LM399AH (selected !) are replaced by 3 or 4 LTZ1000A (1 zener per card instead of 2).

[jfphp]

Thank you very much to all people who neither helped nor answered : you have forced Bruno and me to work. We have now reconstruct all the logical equations of the PAL, burned new ones and find the factory reset password (one week work for this very carefully hidden password). Various failures (nullmeter, output board...) were easy to fix. Two 2720GS are now living again...

[Edwin G. Pettis]

To jfphp,

I am curious, are any of the precision wire wound resistors in your 2720Gs yellow?

Sorry no one could help but Valhalla keeps any details they view as proprietary out of the manuals and only a few people at the factory, the engineers who designed this unit, have knowledge of such things.  They also like to charge a lot for their manuals if they are available, which would not have been much help either.  It is a bit strange that Valhalla claims to not have the microcode though, that should have been in the engineering documentation and that should have been kept on file.

[Mickle T.]

In the high stability option the 6 or 8 LM399AH (selected !) are replaced by 3 or 4 LTZ1000A (1 zener per card instead of 2).

Very strange. AFAIK the first of the Valhalla 2720GS calibrators don't have any of it, but only selected 1N827 zeners in the ovens. In the mid-1980s, LTZ1000 did not exist.

P.S. And how about factory reset password?

[Kosmic]

Very strange. AFAIK the first of the Valhalla 2720GS calibrators don't have any of it, but only selected 1N827 zeners in the ovens. In the mid-1980s, LTZ1000 did not exist.

I recently acquired a unit built in 1988. It's equipped with the HSR option  (High Stability) and has 8 x LM399 voltage ref.

@jfphp: I would be really interested if you had any information to share on the reset password and how to reflash the PAL.

Thanks!

[TheSteve]

Very strange. AFAIK the first of the Valhalla 2720GS calibrators don't have any of it, but only selected 1N827 zeners in the ovens. In the mid-1980s, LTZ1000 did not exist.

I recently acquired a unit built in 1988. It's equipped with the HSR option  (High Stability) and has 8 x LM399 voltage ref.

@jfphp: I would be really interested if you had any information to share on the reset password and how to reflash the PAL.

Thanks!

I don't have any info to share but am glad it stayed in Canada. I was local to it and I was really considering it.
If it does need repair please consider starting a thread in the repair section to share the problems and fixes.

[Kosmic]

I don't have any info to share but am glad it stayed in Canada. I was local to it and I was really considering it.
If it does need repair please consider starting a thread in the repair section to share the problems and fixes.

Yup it made the trip from BC in one piece (I'm on the east coast)   

I will definitely start a thread since there is a couple of issues. For one, the fan is not working and I'm pretty sure it's not normal  After a hour it get really hot. Nevertheless after running the internal check all the voltage ref were back online. So that's a good start 

[branadic]

Quite some old thread, but I thought to revive it.

In the high stability option the 6 or 8 LM399AH (selected !) are replaced by 3 or 4 LTZ1000A (1 zener per card instead of 2).

Isn't it more like the 2720GS contains three reference boards with a total of 6x LMx99, while the HSR option has four board with a total of 8x LMx99?

Was someone else successful in replacing the Xircor X2212 chips? A replacement was mentioned here, but without further details.
Same goes for all the rest mentioned like factory reset password or firmware file for the PAL. Already contacted jfphp via his website, but yet without any feedback.

We (me and a friend) have developped a drop in circuit for replacement of the Xircor X2212 on each zener board and a specialy U shaped board for the 4 x X2212 on the output board : the X2212 are a nightmare and die without warning driving the very bad made software to absurd diagnostics or to stall in starting : the still obtainable NOS X2212 are very concerned but there is no general law.
We can supply new burned PAL for the CPU, the front panel board, factory reset password and various spare parts including manuals. We have rebuilt option boards too.
Eventually we can sell some 2720GS : one as is (HV issue, zener issues...), one not meeting the output settling time (> 10s for 1ppm on 13V range and below) and several fully refurbished (all caps, relays new, all X2212 replaced by custom boards, 4 zener boards --8 zeners-- , factory reset password, cal with 2ppm accuracy of the standards used...).

-branadic-

[branadic]

Hello everyone,

I received such a unit today, so I will put everything I find out about it here so everyone can benefit from it. 

When the unit was about to be sold to me the seller tested the unit, which tripped his circuit breaker. He also noticed some smell of burning components inside the unit. Okay, so it was eventually sold as defect, which is fine, right? Let's see what needs to be fixed...

[branadic]

First thing I've noticed was the front panel being pretty yellow. A good looking meter is good motivation to get things fixed. So the first step I've done after The Arrival was using pure ethanol to remove whatever this yellow film was. Could have been either a transparrent but yellowed protection coating or human residue from the last decades? Who knows, but in any case it was removable and is almost gone.

[branadic]

Next I pulled some cards that obviously contained physical large axial capacitors and what do you know? It's almost always a bad electrolytic capacitor, also here. Since the guts of the caps was released only a few days ago vinegar helped for the most part. So next step is a fully recapping of the unit. I also noticed that some of the fuses are blown, maybe a good sign and the rest was well protected? Fingers crossed.
However, after removing the caps I've noticed at least two burn marks creating short circuits between traces, which needs fixing as well. I scraped away the carbon and filled the remaining grooves with some UHU endfest. First fix done.

[branadic]

On the front panel board is some AA 3.6V lithium primary cell, measuring it showed not a single bit of live. So it needs replacement too.

-branadic-

[Kosmic]

Make sure you are really gentle with the high voltage transformer. Mine shorted after being removed / reinstalled. It's not like I dropped it or something. I guess the cables were bent a bit too much 

anyway I still need to source a new one.

[branadic]

Thanks for the hint, I will keep that in mind.
The caps on the power supply board all measure good, nevertheless they will be replaced in any case. The caps on the high voltage power supply seem to be the main issue and were already removed.

-branadic-

[branadic]

Ordered most of the caps, fuses and battery. I guess they will arrive next week. I already replace the 100 µF/25V with 100 µF/50V electrolytics on low voltage and high voltage power supply boards and noticed, that some of them started leaking. So it's good advice to replace all caps in the unit.
Slow but at least some progress, that will hopefully bring back the unit to life. I will take more photos of all boards and link them here.

Someone by accident has the CAL key? It's missing on my unit, but before I order a replacement key switch, maybe someone can help?

-branadic-

[Kosmic]

Someone by accident has the CAL key? It's missing on my unit, but before I order a replacement key switch, maybe someone can help?

-branadic-

I have one. I'll check if I can make a copy.

[Kosmic]

Someone by accident has the CAL key? It's missing on my unit, but before I order a replacement key switch, maybe someone can help?

-branadic-

I have one. I'll check if I can make a copy.

Done! I had to go see a locksmith but that was no problems.

[branadic]

Make sure you are really gentle with the high voltage transformer. Mine shorted after being removed / reinstalled. It's not like I dropped it or something. I guess the cables were bent a bit too much 

anyway I still need to source a new one.

Houston we have a problem. Seems like the dying caps took the HV transformer with them. Anyone has any detailed info about that xformer?
I better check low voltage transformer too.

-branadic-

[coromonadalix]

sv manual for thoses i havent seen it ...

https://xdevs.com/doc/Valhalla/2720GS/
https://xdevs.com/doc/Valhalla/2720GS/Valhalla_Scientific_2720GS_Vol.%202_sm.pdf

Man  thats not funny,   and you need at least some specs for sure, and or find an defective unit hoping the xformer is still good ??

page 6  show the connector pinout of the xformers  ... searching for the voltages

p34 p35 shows  some of them  not sure if we all of them

2x16v ac ? with an ct
2x 115v ac ?
2x 1100v ac ??

that give us 7 wires ??  possibly 8   because we faintly see 115vac CT  going to the 16vac CT  ??