Author Topic: Valhalla 2701b Voltage Standard  (Read 8178 times)

0 Members and 1 Guest are viewing this topic.

Offline henasTopic starter

  • Contributor
  • Posts: 32
Valhalla 2701b Voltage Standard
« on: June 09, 2015, 02:20:22 am »
My first post. Hello to everyone!!

I recently purchased from Ebay voltage standard Valhalla 2701B. I tested it with my HP3457a multimeter. It was slightly out as far as calibration is concerned. No surprise here, calibration sticky show December 1985 as last calibration. I am in the process of getting service manual.

What I found in the meantime it drifts around 1.5uV for every x1 Celsius. No problem here I can easily stabilize the temperature and this is close to specs anyway. More troubling, it drifts 2uV for every volt mains power change. I am in Australia and our 240V mains power supply can easily fluctuate between 255v and 235v and sometimes more; 40uV is quite a drift. I would be interested if anyone on the list has 2701B and could share some experience. I found quite a few units floating around but very little information.

Still without schematic I checked the unit. It appears to have classic linear power supply. The caps tested OK. Any suggestions??

Other option, I could use a device to decrease power fluctuation. I don't have UPS on hand to test it but I am not sure if UPS can stabilize slow voltage fluctuations. Any thoughts??

Any Help Greatly Appreciated
« Last Edit: June 09, 2015, 02:22:21 am by henas »
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #1 on: June 11, 2015, 12:54:58 am »
It appears this subject is not very popular on this forum.
Perhaps there is another forum which deal more specifically with multimeter calibration and voltage standards. I would appreciate if someone could point me in the right direction.
 

Online Vgkid

  • Super Contributor
  • ***
  • Posts: 2710
  • Country: us
Re: Valhalla 2701b Voltage Standard
« Reply #2 on: June 11, 2015, 01:37:58 am »
Have you tried contacting Valhalla, they might have a manual.
Else try the volt-nuts mailling list.
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline eas

  • Frequent Contributor
  • **
  • Posts: 601
  • Country: us
    • Tech Obsessed
Re: Valhalla 2701b Voltage Standard
« Reply #3 on: June 11, 2015, 07:20:14 am »
Thirty seconds and Google brought me to this:  http://exodus.poly.edu/~kurt/manuals/manuals/Other/VALHALLA%202701B%20Operations.pdf

Sounds like the tempco on your unit is out of spec by +0.5ppm.  No spec for line regulation, other than +-10% on the AC input.

Scan of the schematic is a little awkward, but you could probably print it out and piece it back together on a single sheet...

Its not a classic linear power supply. It uses a switch-mode power supply as a pre-regulator that then feeds the series linear regulator. Looks like the internal reference voltage is created by PWM modulation and subsequent filtering of a stable reference voltage.  Seems like there are a few places for things to go wrong...

 

Offline VintageNut

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: 00
Re: Valhalla 2701b Voltage Standard
« Reply #4 on: June 12, 2015, 02:35:59 am »
The drift you see may not necessarily be the 2701B. DMMs can drift plenty.

If the uncertainty of your multimeter is greater than the uncertainty of the calibrator, then you need a much much better multimeter to evaluate the Valhalla.

Rule of thumb is that the calibrating instrument requires 4X better spec than the instrument you are calibrating.

Most 6 1/2 digit DMMs have the entire 6th digit and much of the 5th digit buried in uncertainty. The bottom digits may be stable and linear but are nonetheless uncertain. 

I used a Keithley 2002 work on my Fluke 332A. This $6000 DMM as good as it is, is not 4X better than the 20 ppm 332A. The linearity of the DMM is 0.1ppm which is plenty good enough to linearize the dials of the calibrator but the absolute uncertainty of the calibrator is still not going to be as it can be.
working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #5 on: June 15, 2015, 01:50:39 am »
Thank you for the info. I found and downloaded this manual from Kurt website some time ago but this is not a complete manual so I end up buying one from Artek Manuals. Every circuit and PCB layout is scanned 50% only and the other half is missing. Looks like somebody's promo to sell complete manual.

For testing I am using hp3457a set into statistic mode with integration time set 100 so I have pretty stable 7.5 digits. This thing is definitely drifting with temperature and AC voltage change. I feel it needs calibration too. When I received it one of the dials was frozen and old calibration seal was broken. It is quite possible someone might have an attempt to calibrate it. I have added extra external insulation to LM119AH voltage standard. It would be interesting to hear from another owners about the unit.

Calibration is calling for differential multimeter and voltage standard I don't have. I wonder if this can be done just with my Hp3457a.

I understand my multimeter may not be enough to properly asses/calibrate 2701B but I am more than happy if I can get it close to specs of my Hp3457a. One day maybe I can afford Hp3458a.....
« Last Edit: June 15, 2015, 01:53:53 am by henas »
 

Offline VintageNut

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: 00
Re: Valhalla 2701b Voltage Standard
« Reply #6 on: June 15, 2015, 07:20:29 pm »
Hi

I just downloaded the datasheet for the 3457A. It looks like on the 30V range, measuring 10V, the total uncertainty is 59 ppm. Somebody correct me if I am wrong.

Knowing 59 ppm uncertainty, what you most want to know is offset and linearity. The thing that will kill your ability to use the 3457A is if noise is a large portion of uncertainty. If noise is small, then you can correct for the offset.

Checking linearity will require a Kelvin Varley divider. I Googled and someone on Volt-Nuts says that there is no spec for linearity for the 3457A. You will have to find out for yourself.

You want to only use th 30V range on your 3457A. Try to set the calibrator for every combination of dial settings. Does your 3457A correctly change a digit every time you change that digit on your calibrator?
working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #7 on: June 17, 2015, 12:15:08 am »
Thank you VintageNut for all your help. I had another go yesterday. It appears that someone attempted to tweak it. By going through the ranges I found this instrument is not linear any more. It needs proper calibration. I tried to tweak it with my 4357A but this task is impossible. I need the right tools.

What is recommended. Unfortunately at this point of time I luck experience in metrology and would appreciate some input from more experienced posters. What gear is recommended and perhaps some guidance for prices I should expect on Ebay. I don't want go over the board (I can't afford Hp4358A) but saying this I prefer decent instruments.
 

Offline VintageNut

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: 00
Re: Valhalla 2701b Voltage Standard
« Reply #8 on: June 17, 2015, 03:02:15 pm »
Thank you VintageNut for all your help. I had another go yesterday. It appears that someone attempted to tweak it. By going through the ranges I found this instrument is not linear any more. It needs proper calibration. I tried to tweak it with my 4357A but this task is impossible. I need the right tools.

What is recommended. Unfortunately at this point of time I luck experience in metrology and would appreciate some input from more experienced posters. What gear is recommended and perhaps some guidance for prices I should expect on Ebay. I don't want go over the board (I can't afford Hp4358A) but saying this I prefer decent instruments.

Please describe your procedure that led you to your conclusion.
working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #9 on: June 18, 2015, 08:52:41 am »
All measurement I made with Hp3457a using Math Stat feature. For each sample, to average reading results, I set integration time through NPLC to 100, added 1s delay between measurements and set the measurement counter NRDGS to 100. I compared MEAN, SDEV, LOWER and UPPER between the samples. For some who may not be familiar with Hp3457A; MEAN=averages reading for the sample taken, SDEV=standard deviation, LOWER=minimum value red, UPPER=maximum value red for the sample.

I used autotransformer to vary the mains supply and I came to conclusion 2701b output changes consistently about 2 ppm per volt. While the sample was taken I was constantly monitoring mains voltage and adjusted autotransformer accordingly to maintain 240v. Not perfect but it worked I could consistently decrease SDEV from about 20-30ppm to mid single figure.

For temperature I had temperature sensor inside instrument’s case. The drift was about 1-1.5ppm for each 1 centigrade Celsius.

For linearity check I flicked switches while measuring the output. For instance in 1.2 volt range the error was around 12ppm for 1V set but it gradually increased when decreasing the voltage and was about 500 ppm for 100mV.

Knowing the linearity is off I attempted to calibrate it with my multimeter. First zero offset and secondly linearity in low range. After spending a few hours without getting anywhere I finally came to conclusion I need the right gear for this task.
« Last Edit: June 18, 2015, 10:24:00 am by henas »
 

Offline deadlylover

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: au
Re: Valhalla 2701b Voltage Standard
« Reply #10 on: June 18, 2015, 03:18:35 pm »
While you don't have the gear to calibrate/adjust the 2701b to it's accuracy specifications (not many do though =P), I don't see why you can't at least do the zero and linearity adjustment with your 3457A. Linearity verification is another thing entirely, but I don't see the linearity spec in the manual so let's not worry too much about that, perhaps it's in that 4ppm of range + 2uV accuracy spec.

Though the caps test okay you might as well replace them, even if only pre-emptively if you have any plans to keep the unit. Does it drift 2ppm per AC line volt on all ranges? Lots of things to go whack on the power supply indeed...

I did a quick 2 minute linearity check of the 3457A's 30V range, testing 10V down to 1V (in 1V steps) against an R6581 8.5 digit meter(~-/+0.1ppm linearity IIRC). Just with the front panel on 6.5 digit mode 100NPLC and no math, it tracked nicely with the R6581 -/+ 1 count or so. Keep in mind my voltage source is only 6.5 digit as well, so we expect the last digit to dance a little from rounding error and stability.
 
Not a comprehensive test by any stretch of the imagination, but I hope it gives you a rough idea of the 3457A's linearity, the last digit is pretty useful on 6.5 digit mode (accuracy/uncertainty issues aside). I didn't calibrate my 3457A's linearity yet and I bought it used, dunno how yours is performing though, is it in calibration?
 

Offline VintageNut

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: 00
Re: Valhalla 2701b Voltage Standard
« Reply #11 on: June 18, 2015, 05:21:17 pm »
All measurement I made with Hp3457a using Math Stat feature. For each sample, to average reading results, I set integration time through NPLC to 100, added 1s delay between measurements and set the measurement counter NRDGS to 100. I compared MEAN, SDEV, LOWER and UPPER between the samples. For some who may not be familiar with Hp3457A; MEAN=averages reading for the sample taken, SDEV=standard deviation, LOWER=minimum value red, UPPER=maximum value red for the sample.

I used autotransformer to vary the mains supply and I came to conclusion 2701b output changes consistently about 2 ppm per volt. While the sample was taken I was constantly monitoring mains voltage and adjusted autotransformer accordingly to maintain 240v. Not perfect but it worked I could consistently decrease SDEV from about 20-30ppm to mid single figure.

For temperature I had temperature sensor inside instrument’s case. The drift was about 1-1.5ppm for each 1 centigrade Celsius.

For linearity check I flicked switches while measuring the output. For instance in 1.2 volt range the error was around 12ppm for 1V set but it gradually increased when decreasing the voltage and was about 500 ppm for 100mV.

Knowing the linearity is off I attempted to calibrate it with my multimeter. First zero offset and secondly linearity in low range. After spending a few hours without getting anywhere I finally came to conclusion I need the right gear for this task.

Are you using more than one range on your DMM? If so, you are doing it wrong. I made that mistake when I started to try to adjust my Fluke 332A. You have to use the most linear range, and only that range.

working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #12 on: June 19, 2015, 12:34:59 am »
While you don't have the gear to calibrate/adjust the 2701b to it's accuracy specifications (not many do though =P), I don't see why you can't at least do the zero and linearity adjustment with your 3457A. Linearity verification is another thing entirely, but I don't see the linearity spec in the manual so let's not worry too much about that, perhaps it's in that 4ppm of range + 2uV accuracy spec.


I thought the same but interaction between two pots for linearity and two pots for zero combine with unstable reading from Hp3457A make this task close to impossible but I will take a closer look maybe I do have a problem. As far as 2701b linearity it is clear is way off as my examples for 1.2v range demonstrates.

It is encouraging to hear linearity of your 3457A is so good. I have a couple more simple DC sources which I tested against another 6.5 digit meter and both tested spot on my Hp3457A as well as 0.01% and 0.005% resistors. When I received it about a month ago it had calibration sticky expired last year with CALNUM counter set to 108. Apparently it spent its healthy life in Pratt & Whitney special project division lab in Florida. Not freshly calibrated but I am pretty confident my Hp3457A is as good as it can be.


Are you using more than one range on your DMM? If so, you are doing it wrong. I made that mistake when I started to try to adjust my Fluke 332A. You have to use the most linear range, and only that range.


My example for 1v and 100mV is showing the diference by factor of 41. This looks to me that linearity is way, way off. Mostly I tested it in 30V range.
« Last Edit: June 21, 2015, 10:42:15 am by henas »
 

Offline deadlylover

  • Frequent Contributor
  • **
  • Posts: 314
  • Country: au
Re: Valhalla 2701b Voltage Standard
« Reply #13 on: June 19, 2015, 02:08:49 am »
It does sound like the 2701b is a sick puppy, you shouldn't have to resort to using math/statistics just to get stable readings on it.

I'd definitely look into the power supply some more, just be really careful....1kV+ rails, yikes. I would only set up probing when power is off (and caps discharged!) with hooks and grabbers, you know the drill. See if there is any good spot to probe the low voltage rails with a scope, I guess the most obvious thing is that there's a good chance one of the rails has too much ripple on it from bad caps.

The 3457A is a good box, quite stable now with age, make sure you look into replacing the AC inlet filter if you haven't already...mine blew up on first power on because it lived a cosy life on 100V mains before being exposed to the wrath of our 230V mains.  :-DD
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #14 on: June 19, 2015, 06:27:06 am »
The same me it took about 20 minutes before it blew up but i am used it, any gear from the US off A blows up its filter caps regularly. I can almost smell it in my dreams...
You have only 230V, how about 240V to 250V where I am.
« Last Edit: June 19, 2015, 08:24:22 am by henas »
 

Offline VintageNut

  • Frequent Contributor
  • **
  • Posts: 534
  • Country: 00
Re: Valhalla 2701b Voltage Standard
« Reply #15 on: June 20, 2015, 01:11:33 pm »
Linearity is easy to check. You set the calibrator to the 10V range, 9.999999V out and set your DMM for the 30 V range and turn of auto range on the DMM.

Your reading on the DMM over or under 10V is the offset. It should be a stable reading. Keep that offset in your head.

Grab a decade knob on the calibrator and slowly click from 9 to 0. Your DMM should change that column digit the same amount as you change the calibrator. The ppm offset should remain unchanged.

These calibrators have much much better linearity than ppm offset/uncertainty. That is why you cannot adjust the decade string pots with a 6 1/2 digit DMM.

working instruments :Keithley 260,261,2750,7708, 2000 (calibrated), 2015, 236, 237, 238, 147, 220,  Rigol DG1032  PAR Model 128 Lock-In amplifier, Fluke 332A, Gen Res 4107 KVD, 4107D KVD, Fluke 731B X2 (calibrated), Fluke 5450A (calibrated)
 

Offline Kleinstein

  • Super Contributor
  • ***
  • Posts: 14076
  • Country: de
Re: Valhalla 2701b Voltage Standard
« Reply #16 on: June 20, 2015, 02:31:50 pm »
If the unit shows measurable changes of output with mains voltage, there should be something wrong with that unit. Usually sensitivity to mains voltage should be so good that it's hardly measurable, even in these old days. Usually it's hard to see any effect on a LM199 reference. So check the supply - especially capacitors.

The temperature drift may be due to not so good OPs used, and maybe someone changing the offset pots at some of the amplifiers. From what one can see from the small part of the schematic, there may be some drift in the 1-3 µV/K range - which can be significant at low voltages.
 

Offline henasTopic starter

  • Contributor
  • Posts: 32
Re: Valhalla 2701b Voltage Standard
« Reply #17 on: July 09, 2015, 08:02:36 am »
I am revisiting my own post just to give closure to my thread.

My original problems:
  - Voltage drift with temperature change
  - inability to to calibrate the unit especially Zero offset using Hp3457a multimeter
  - Voltage drift with mains voltage fluctuation

I concluded early, tempco is pretty close to spec and nothing I could do about it though I enclosed lm199 voltage reference and surrounding components with silicon foam similar as suggested in monster thread about building voltage standard. I believe the stability has improved.

Zero offset calibration difficulty were due to presence of fluorescent lamp  and low voltage lights in my workshop. After removing this interference I was able to go through calibration process using just Hp3457a. I am pretty pleased with the results.

Thank you Kleinstein, you pushed me in right direction as far as the last problem.
I did more probing and found reference voltage is changing accordingly with mains fluctuation. The power supply voltages were rock solid. I found floating ground was shifting. This led me to chopper OpAmp 7650. After replacing it, the problem went away.

Thank you all for helping me.


 

Offline eas

  • Frequent Contributor
  • **
  • Posts: 601
  • Country: us
    • Tech Obsessed
Re: Valhalla 2701b Voltage Standard
« Reply #18 on: October 26, 2015, 08:09:00 pm »
I ended up buying one of these off of eBay. The seller listed it For Parts/Repair" and said it wouldn't power up, not a surprise, since the fuse holder was missing.

When I received it, I opened it up for inspection. Mine seems to date to the mid 1980s. I haven't tried powering it up yet because I found some issues during my preliminary inspection.

First, I noticed that the PCB under some of the components has started to discolor from excessive heat. I wonder if anyone else has seen anything similar? The specific components implicated in my unit are some diodes at E22 & E23, a heatsinked 5V VRM (mc7805CK), a big power resistor and a less beefy resistor.

The other issue is there are two components at the back right of the main PCB. I think they are decoupling caps with a connection to the chassis, though the markings confuse me a bit. A lead has torn loose from one of them, and neither of them are securely mounted.



One hypothesis is that they are 0.2uF axial caps rated to 5000v, made in 1974 and 1975, probably bodged in as a repair. My confusion comes from what I take to be the older of the two using the newer uF designation, and the newer using an older 0.2M designation. The other, is that I have no idea what they are. Can someone validate either of my hypothesis or suggest alternatives?

This question would be easier to answer if I had a service manual, but as has been noted, the one available online has the diagrams and parts lists cropped off to a letter sized page.  If anyone has a better scan, or can point at a reliable source to buy a reasonably priced manual, that would be appreciated.
« Last Edit: November 16, 2015, 08:00:25 am by eas »
 

Online SingleBitError

  • Newbie
  • Posts: 4
  • Country: us
Re: Valhalla 2701b Voltage Standard
« Reply #19 on: January 24, 2017, 07:46:25 pm »
eas, did you ever get your 2701B working? I just purchased one from eBay myself.

First, I noticed that the PCB under some of the components has started to discolor from excessive heat. I wonder if anyone else has seen anything similar? The specific components implicated in my unit are some diodes at E22 & E23, a heatsinked 5V VRM (mc7805CK), a big power resistor and a less beefy resistor.

I see the same issues with heat damage to the PC board. The discoloration around the large resistors didn't surprise me but I didn't expect it around the diodes.

The bodged capacitors are interesting. The capacitors on mine appear to be original.

" id="thumb_287365

« Last Edit: January 24, 2017, 07:50:44 pm by SingleBitError »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf