Wavetek/Fluke 7004 system

[branadic]

Hi folks,

I managed to score a 7004 system, that recently arrived. With the help of TiN aka xdevs it was possible to get it shipped from US to Germany.
The system is obviously not in its original setup, but put together by multiple different modules collected over time. At least one unit was with the original system as the seller told me. The unit contains a 7000T for averaged output, a simplified version of the 7000N, with the front PCB assembly and the side PCB assembly almost unpopulated, but missing the main PCB with the power supply (DC/DC) and of couse the nullmeter with its display.
The system further contains one Wavetek and one Fluke module, both with a discrete oscillator for the DC/DC converter, while the other two contain a DC/DC based on LT1533.

Last calibration was performed on 1. Sep. 2020, while TiN measured it on 26. Nov. 2020 during its time at xdevs. Since the system was shipped cold and might have seen some very low temperatures during shipment, we can't say for sure if the xdevs volt was successfully transfered to my lab, even though conditioning was performed twice on each module to restore the original output voltage, but by the looks of it the averaged output agrees to about 1.5ppm with xdevs.

Date	AVG-output	cell1	cell2	cell3	cell4	Comment
01.Sep. 2020	9.999 964 2V	9.999 986 2V	9.999 992 9V	9.999 885 4V	9.999 992 5V
26.Nov. 2020	9.999 9603V	9.999 980 7V	9.999 991 8V	9.999 877 6V	9.999 991 2V	xdevs
12.Dec. 2020	9.999 975 6V	9.999 990 2V	10.000 009 0V	9.999 895 9V	10.000 007 9V	K3458A positive
12.Dec. 2020	-9.999 977 7V	-9.999 992 1V	-10.000 010 8V	-9.999 898 0V	-10.000 010 1V	K3458A negative
12.Dec. 2020	9.999 974 2V	9.999 987 8V	10.000 006 8V	9.999 893 6V	10.000 006 0V	R6581D positive
12.Dec. 2020	-9.999 979 2V	-9.999 993 6V	-10.000 012 0V	-9.999 899 3V	-10.000 011 9V	R6581D negative

Attached is a measurement performed with R6581D and K3458A on the averaged output over some hours.

-branadic-

[branadic]

With the help of some hidden figures some bits and pieces to service the units have already been ordered and arrived this week. So after some further measurements and comparisons the system will be maintained and some parts will be replaced. Thanks to them at this point.

-branadic-

[bingo600]

Damm  ... Drooling
Lucky you

[HighVoltage]

I was probably sleeping when you managed to score a 7004 system, 
Congratulations on great system !

Against what reference did you calibrate your K3458A and the R6581D?

[Dr. Frank]

It's still based on the recent Metrology Meeting 2019, comparison to M71 October  2019, and on our private MM June 2020 exchange.

I think branadic also had an official calibration recently, and me an unofficial one by ab-precision, to about 1ppm.

Frank

[branadic]

Fortunately you have been sleeping
Now some people from Metrology Meeting will have the chance to participate as the system will be split. The three empty slots will be assembled with some much more simple LTZ references, so that I still have something to average with.

K3458A was calibrated in 2019 with an uncertainty of 2.5ppm at Keysight Böblingen and will be calibrated in 2021 prior to MM2021 again. An additional measurement point for it was gained at small scale MM2020 in comparison to Frank's 3458A, while we also calibrated R6581D to his 3458A and his references. So pretty much how your doin without a fortune.

Unfortunately I had no chance to compare readings to M71, which could have helped a lot.

I for my part maintain 3x LTZ references based on the design by Andreas, that have been untouched and unmodified since 2019 prior MM2019 (t.c. compensation) and provide some unadjusted 10V, while another 9x LTZ references based on a modified design by Frank were continuiously modified (t.c. compensation, 10V adjustment) and improved, thus are out of any calculation by now.

-branadic-

[The Soulman]

Now some people from Metrology Meeting will have the chance to participate as the system will be split. The three empty slots will be assembled with some much more simple LTZ references, so that I still have something to average with.

Do I understand correctly that you are selling three modules?
Congrats btw. 

[Dr. Frank]

Well, one of these will be mine

Meanwhile, by the aid of our small CAL Club, it's possible to make very good drift predictions of our reference groups to <1ppm/year. A big THX to Adrian in this context.

I will check the calibration value of this M7000 provided by TiN against the baseline of my references , as soon as I get my unit.
Another big THX to Illya!
Frank

[branadic]

Service done on cell2 and cell4, that are now back into the mainframe and perform conditioning again. 
Will see, if there is any change in readings once finished, but I don't expect it anyway.

-branadic-

[martinr33]

That's a super nice setup. I guess you might be answering the question of whether the Fluke or Datron design is the better reference.

These units are a bit difficult to measure with "just" a 3458a.

A couple of thoughts:
   - You might try measuring just the difference between the outputs using the 100mV range (or a nanovolt meter, if you have one). Reversing the leads will deal with any offset at that level.
   - What would be nice would be if you can then reproduce the deltas against TiNs measurements.

If all is well, you should get deltas very close to TiN's measurements. This approach largely eliminates any drift in the meter. You'll still be dealing with noise, but a long average can deal with that.

A nanovolt meter will give you very clean deltas. A nanovolt meter is pretty much essential when dealing with references of this class.

[branadic]

Or you can compare them to a set of references with multiple points over the years and compare each reading to them. That's at least what I did to see if and how they have moved, using the 3458A as a "comparator" with its good transfer spec only.

-branadic-

[Dr. Frank]

That's a super nice setup. I guess you might be answering the question of whether the Fluke or Datron design is the better reference.

These units are a bit difficult to measure with "just" a 3458a.

A couple of thoughts:
   - You might try measuring just the difference between the outputs using the 100mV range (or a nanovolt meter, if you have one). Reversing the leads will deal with any offset at that level.
   - What would be nice would be if you can then reproduce the deltas against TiNs measurements.

If all is well, you should get deltas very close to TiN's measurements. This approach largely eliminates any drift in the meter. You'll still be dealing with noise, but a long average can deal with that.

A nanovolt meter will give you very clean deltas. A nanovolt meter is pretty much essential when dealing with references of this class.

Hi Martin,
you're making some quite interesting statements, which fit to our current discussion in the LTZ1000 thread, about what's the 'best' layout, and what possible practical influence the different solutions might have.
First, there is no essential difference between the FLUKE and the DATRON / Wavetek  devices, concerning the basic circuit of the reference board.
Maybe a few small updates / Engineering Changes have been done, especially on the power supply, but nothing that might give 'better' stability parameters of any kind.
Wait a moment, the LTZ1000A was used in the more recent devices, at the same nominal oven temperature. But that does not make any difference in performance, only 5mA less supply current.

As branadic already wrote, the 3458A can very well be used to make measurements on these references.

In transfer mode, you can achieve about 0.1 + 0.1 ppm uncertainty in relative comparisons. Ratio mode gives similar results.
Direct difference comparison is as well possible, with 10nV resolution, but tricky due to mains loops of the DUTs, same for a nV meter.
The M7000s have a sophisticated and highly isolated power supply to mitigate such effects, and of course they feature the battery mode.
A 3458A got another advantage over a nV DMM, that you can compare two very different voltages like 10V vs. 7.15V.

You can also determine the relative noise performance, that means you measure the quadratic sum of the reference under test plus or versus the LTZ1000A reference inside the 3458A.
If you would do this measurement with a nV meter, you would as well only measure such a quadratic sum between two DUTs, which gives exactly the same result as the case before.
Only an absolute noise measurement with an AC coupled low noise amplifier would give the true noise performance.

If you already have a group of long time monitored and characterized references, you can as well measure or estimate the absolute drift rate of your new DUT, by means of a 3458A as a comparator in either of the 3 mentioned modes. 

A nV meter would not give a big advantage in any case.
There's a general systematic reason behind, and that is the high noise level of zener based references, on the order of 100..200nV_rms.
Therefore you won't get 'cleaner' difference signals, even if a nV meter usually has less noise than a 3458A.
See DMM noise comparison thread by TiN!!

Due to the special characteristic of zener noise, you also can't average this noise out (see discussion in other thread)..
Only if your other reference would be a JJ array, you might perhaps get lower noise, only that of your DUT, i.e. about 1/sqrt(2) only.


Due to that noise source, any zener reference, let it be LTZ or LTFLU based, are limited in uncertainty or stability to a bit less than about 0.1ppm or 1µV.
That's also the limit to link the (zero) uncertainty of a JJ array to the real analogue world.. a reference with much lesser noise, or a different kind of noise is required.. see branadics thread about a discrete bandgap reference, and the chinese zener diode experiments, I don't remember its designator right now.. was it 2W22?

All other possible effects like thermal e.m.f., amplifier noise, T.C. effects and so on, which theoretically might be 'improved' by different circuits, or layouts, and so on, are mostly hidden below this noise limit.

I always recommend to do a very critical reality check on such measurements, and of course on theoretical excursions about what's possible, or useful.
A 3458A or similar DMM is a good starting point to get a realistic experience.


Frank

[branadic]

Hi all,

I recently received a scan with the cross-section of the W/F7000 single cell enclosure from TheSteve. I managed to import the picture to a CAD program and retraced it.
I still need someone to check, whether the dimensions I could extract from the picture are about right or if they need some further tweaking, but I guess there might be an interest in this enclosure anyways, so I share the CAD file here.

-branadic-

[notfaded1]

So branadic you're going to replace three of the modules with your own LTZ1000A references?  I assume this is why you are copying the case as well?  Is this because you're selling the other three or something?  Anyhow... pretty neat stuff!  Just curious. 

Bill

[branadic]

The F7004 was meant to be split from the very beginning on, even though I was meanwhile uncertain about that decision. I will keep one unit for myself. So the rest of the mainframe wants to be filled with something until it is needed at our Metrology Meetings, when all units come back together.
The enclosure I was looking for and redrawing is a single W/F700x module enclosure that would be nice to have once my unit is in transport for calibration or comparison with other voltnuts. That's it.

-branadic-

[Bill158]

I agree with martinr33.  A HP/Agilent/Keysight or Keithley uVolt meter is the better/preferred way to intercompare two 10 Volt Reference Standards.  I show two examples of the use of a uV meter.  One is on page 40 of the Fluke 732C manual where a Keithley 2182A Nanovoltmeter is used and the other is from an advertising brochure from Supracon's Josephson Voltage Standard where a HP/Agilent/Keysight Nano Volt meter is used.  Fluke is now a master of maintaining the Volt.  Fluke does a lot more in maintaining the Volt for more small/medium sized cal labs than NIST via the 732A/B/C.  I have tried using the H/A/K 3458A at the 10 volt range to inter-compare my Voltage References and it just doesn't work as well as using the 3458A as a null meter on the 100 mv range between two References.  I use 100 PLC and then average 40 readings using the stats on the 3458A.  I have done this method for 15 years and I am able to track their changes and predict their drift.  I have also tried using a K182 and it works just as well.  I use the NBS Technical Note 430 (NBS430.pdf) as a guide to do this.  Yes NTN 430 was for standard cells  (1.8xxxxx / 1.9xxxxx volts) but it works just as well at 10.0000000 volts.  Also the Datron/Fluke 7000 included a uV meter to do inter-comparisons between the 7000 modules.  But no matter what method you use, a number of reading samples is required to reduce the Zener noise.
Bill

https://download.flukecal.com/pub/literature/732c734comeng0000.pdf
https://download.flukecal.com/pub/literature/7000____umeng0100.pdf

[Dr. Frank]

..  A HP/Agilent/Keysight or Keithley uVolt meter is the better/preferred way to intercompare two 10 Volt Reference Standards.
..
 I have tried using the H/A/K 3458A at the 10 volt range to inter-compare my Voltage References and it just doesn't work as well as using the 3458A as a null meter on the 100 mv range between two References.  I use 100 PLC and then average 40 readings using the stats on the 3458A.  I have done this method for 15 years and I am able to track their changes and predict their drift.  I have also tried using a K182 and it works just as well.

.. But no matter what method you use, a number of reading samples is required to reduce the Zener noise.
Bill

Hi Bill,
this really will work with appropriately designed, well isolated / shielded mains supplies, like in the 7000 or 732B/C instruments.

Please, would you mind publishing some examples of your measurements, absolute vs. differential method, to have a possibility to compare your statements with my findings?

I have tried that using quite simple linear PSUs, no success as expected: https://www.eevblog.com/forum/metrology/power-supply-for-voltage-references/msg3378228/#msg3378228

My absolute measurement on the M7000 vs. 3458A is stable within +/- 0.1 ppm over 17h, so can you please provide a differential measurement with even better stability, similar to what SUPRACON claims, like +/- 100nV = +/- 0.01ppm ?

I'd also be interested in the statistical method, how to achieve these numbers.

Thanks.

I've attached the equivalent NPLC 100, 40 points average absolute measurement, i.e. the black line in the diagram.
Anyhow, the mid term instabilities of the LTZ references limit the uncertainty or reproducibility to 0.1ppm.

Frank

[Bill158]

Dr. Frank:
I see a basic problem here.  We are discussing apples and oranges here, in terms of time periods of measurements.  I am thinking of making one group of measurements between each reference pair of a bank of Voltage References (Vrefs) once a week over 15 years.  See NBS Note 430.  You are thinking of making measurements for a much shorter period of time, in your example 17 hours.  The 3458A simply cannot be stable enough, at 10 volts, over 15 years, or even a week, to give me the results I need to predict the drifts of the Voltage References.  I agree that comparing two Vrefs in a short period of time, i.e. 5 minutes, you will get excellent results and I have done this.  But try to do this again 7 days from the initial comparison and you will not be able to achieve stable enough results to get meaningful data, let alone over 15 years.  So the only way to remove the long term drifts of the 3458A is to use it as a Milli-volt meter on the 100 mV range.  Then do 100 PLC, and take the mean of 40 measurements, reverse the leads and do another 40 measurements and then add the two readings, reversing the sign of the second reading, and divide by 2.  I still get a lot of zener noise but over time the drifts can be determined.  I didn't have a uV meter at the beginning of these measurements.
The 3458A has only 100 nv resolution, on the 100 mV range, but that is .01 ppm of the 10 volt references and with averaging is good enough for my measurements.  The K2182 has a 10 mV range and a resolution of 1 nV, the HP/A/K 34440A has a 1 mV range and a resolution of 0.1 nV.  The K182 has a 3 mV range and a resolution of 1 nV.
I would imagine that with the SupraCon JVS they supply the 34420A because of the 0.01 ppm spec of the JVS or something like that.  I have no idea how the JVS is recommended to be used.  Again I would imagine that you would compare the JVS to the Vref under test, once a day, over a period of a week or more before reporting the absolute value of the output voltage.  Again the 3458A simply cannot get the required results using the 10 volt range if it was being used as a comparison meter between the Vref and the JVS.  You can see from the JVS specs that you could expect to give a value to the Vref of around 0.01 ppm, depending upon the zener noise.  Attached is a paper on the use of the SupraCon JVS with a F732A or F734A.  This is discussing a one time calibration of the Vref, but to predict drift of the Vref this could be done over a week or more.
Bill

[Dr. Frank]

Dr. Frank:
I see a basic problem here.  We are discussing apples and oranges here, in terms of time periods of measurements.  I am thinking of making one group of measurements between each reference pair of a bank of Voltage References (Vrefs) once a week over 15 years.  See NBS Note 430.  You are thinking of making measurements for a much shorter period of time, in your example 17 hours.

Hi Bill,
that's not the case. Exactly like you, I also make regular comparisons (every 2 or 4 weeks) of my bank of meanwhile 11 references, the 3458A itself is used as a transfer instrument only.
The 17h measurement has nothing to do with that, it only demonstrates the stability and noise of the 3458A relative to a DUT, but also that over a longer period of time there might occur some irregular  bumps.
For 9 of these DUT I monitor their (more stable)  raw zener output of 7.08 .. 7.16V. Then I have two 10V references.
Therefore the differential method is not suitable, so I have to make absolute measurements, using the extraordinary linearity of the 3458A and its good short term temperature stability, and afterwards calculating a gain correction.
I use 2 x 1min (16 samples) for each DUT, with reversing the polarity.
It's to be discussed, if 2x40 samples, like you do, really give a better uncertainty.
I think, that both methods are sufficient to achieve below 0.2ppm transfer accuracy, which is also sufficient for long term monitoring.

The 3458A simply cannot be stable enough, at 10 volts, over 15 years, or even a week, to give me the results I need to predict the drifts of the Voltage References.  I agree that comparing two Vrefs in a short period of time, i.e. 5 minutes, you will get excellent results and I have done this.  But try to do this again 7 days from the initial comparison and you will not be able to achieve stable enough results to get meaningful data, let alone over 15 years.
So the only way to remove the long term drifts of the 3458A is to use it as a Milli-volt meter on the 100 mV range.  Then do 100 PLC, and take the mean of 40 measurements, reverse the leads and do another 40 measurements and then add the two readings, reversing the sign of the second reading, and divide by 2.  I still get a lot of zener noise but over time the drifts can be determined.  I didn't have a uV meter at the beginning of these measurements.

As explained above, I do not use the 3458A (any more) as a long term reference.
Anyhow, doing regular ACAL every day, and keeping the R.T. stable to within +/- 1°C, you will even get excellent stability over 1 week with the 3458A. My 3458A drifted only 1.7ppm over 7 years in DCV, because it is running only when used, and I have lowered its oven temperature to 65..70°C, i.e. adapted it to metrology requirements.
The LTZ 1000 references all drift between -0.3 .. -1.0ppm/year because they run 24/7 as usual.
The extrapolated mean value of this group meanwhile seems to be predictable to about 0.2ppm/year.. and that is now my actual volt representation.

The 3458A has only 100 nv resolution, on the 100 mV range, but that is .01 ppm of the 10 volt references and with averaging is good enough for my measurements.

Sorry no, it's got 10nV resolution in the 1V and in the 100mV ranges. The noise floor of nV meters is lower.

The K2182 has a 10 mV range and a resolution of 1 nV, the HP/A/K 34440A has a 1 mV range and a resolution of 0.1 nV.  The K182 has a 3 mV range and a resolution of 1 nV.
I would imagine that with the SupraCon JVS they supply the 34420A because of the 0.01 ppm spec of the JVS or something like that.  I have no idea how the JVS is recommended to be used.  Again I would imagine that you would compare the JVS to the Vref under test, once a day, over a period of a week or more before reporting the absolute value of the output voltage.

Well, and here again I wonder if that really makes sense to have 10nV, even 1nV resolution with accordingly low noise instruments.
Any zener based reference shows an order of magnitude higher short term noise, at least these irregular jumps will ruin the wedding.
I would estimate from my latest measurements that you can determine the mean voltage of such references to no better than about 200nV, or 0.02ppm.
And that's easily accessible with a  3458A - in differential mode.
Changes by T.C. of < 0.03ppm/°C as specified for practically all voltage references will make such measurements even more questionable regarding their uncertainty.

Again the 3458A simply cannot get the required results using the 10 volt range if it was being used as a comparison meter between the Vref and the JVS.  You can see from the JVS specs that you could expect to give a value to the Vref of around 0.01 ppm, depending upon the zener noise.  Attached is a paper on the use of the SupraCon JVS with a F732A or F734A.  This is discussing a one time calibration of the Vref, but to predict drift of the Vref this could be done over a week or more.
Bill

Well, the Supracon specification only shows some error bars, but no longterm differential measurement. There simply do not exist any LTFLU or LTZ based reference  which would have this performance. Probably the averaged signal of 8 or 10 modules (734C, 7010N) might nearly reach this level.

Would be interesting to see one of your differential measurements, or such a long term noise measurement over 1h of your system.

I have done over night such a differential measurement, using the battery powered M7000 versus a mains powered LTZ 1000, 10V reference, 3458A as a null instrument.
Yes, that's really better compared to the absolute comparison method.
1h StD is about 40% less, and 'your' 40 point averaged peak to peak noise is 50% only, about 1µV_pp over 17h.
Take another factor of 1/ sqrt(2) down, to get the noise performance of a single zener, and you get the physical limit of about 800nV.

If you make measurements over 1 or 40 minutes only, the measured StD and pp noise might be even lower, but that'll not give you the real uncertainty for zeners.

Frank

Rem.:
absolute vs. differential method over 17h or 8h
pp noise, 40 point average is  1.9 µV vs. 1.0µV
StD over 1h is 220nV vs. 190nV

added picture of complete setup

[Bill158]

Frank:

Sorry no, it's got 10nV resolution in the 1V and in the 100mV ranges. The noise floor of nV meters is lower.

Mia culpa.  You are correct.  It is 10 nV resolution on those ranges.

I have also lowered the resistor on the A9 board so that my LTZ 1000A runs at 65 - 70 C.  I think this mod was from you if I can remember correctly.  This does a LOT towards making the 3458A very stable.

I have not ever added an IEEE488 interface to my computers.  One basic reason is that my computers are far away from my Vref setup.  Another is that I don't have a nV switching matrix.  I simply use the F(x) keys on the 3458a with the appropriate commands to make the one time measurement of a pair of Vrefs, record the mean reading manually and then reverse the wires and do the same procedure again.  I then go through all of the 5 x 732As, per NBS 430, recording all the readings.  So I get a total of 6 averaged readings which I then enter into my Excel spreadsheet.  This usually takes about 1 hour every week.  I do other tasks while the readings are being taken.  My measurement F key is programed "DELAY 60;MATH 14;NRDGS 40;TRIG4;TRIG;".  Before I begin this I have a F key programed "RESET;NPLC 100;NDIG 8;" to setup the 3458A in the beginning one time.

Would be interesting to see one of your differential measurements, or such a long term noise measurement over 1h of your system.

Sorry, no IEEE488 to do this kind of a measurement.

For 9 of these DUT I monitor their (more stable)  raw zener output of 7.08 .. 7.16V. Then I have two 10V references.
Therefore the differential method is not suitable, so I have to make absolute measurements, using the extraordinary linearity of the 3458A and its good short term temperature stability, and afterwards calculating a gain correction.

I only have 10 volt Vrefs ( 5 x 732A ) and a 732B to measure so that makes my task simpler.  The one 732A in the upper right hand corner of the bank is simply too unstable to be used in my measurements.  I do keep track of it's output but separately.  My measurement differential cable is based upon the FLUKE #5440A-7002 low thermal test leads.  The FLUKE cables are basic RG58U with special low thermal plugs on the ends which are unobtainium for us.  So I use gold plated solid copper plugs on the ends.  These are made by VIBORG VB-401G and are banana type intended for audio use.  Spade lugs are just way to hard to use when changing the leads around for the measurements. 

I then take this "raw" data and apply formulas in Excel.  Then I use this to create various charts to display the manipulated data.  I also then have a spreadsheet to track the drift of each unit relative to one 732A which I have determined to be ultra-stable (I think).  My thinking and various charts are sometimes confusing to me and I am not sure they make any sense or reason.  But so far I think all of this works.

I got the 732B in 2015 so that I could send it into FLUKE for a calibration.  I did this once and am going to send it in again to see if my reasoning is correct or if I am simply way off with what I am doing.  One of these days real soon now.  Between shipping and FLUKE charges for cal this adds up to around US$1,000, so for a hobbyist it is a little steep.

See the pictures posted below of my setup.

Bill

[Dr. Frank]

Hi Bill,
thanks for the presentation of your lab, that's really impressive, especially that you have 'only' 10 x 10V refs.

Do I spot that correctly, that your case board really bends heavily under the burden of two 5440 and the 5200 with the 5105 HV amplifier on top?
That has to be enforced, I think..

We're doing very similar things, obviously, e.g. I also keep track of the references and their drift prediction by an xls file, where I also sometimes have to consider, how I have done it .. so not to lose the overview. .
I found out, that my 5442A is the Golden Device in my setup, as it shows a slight upwards drift over years only, absolute value being less than these -0.3ppm/yr.

Get yourself a cheapo 82357B adapter, they are working pretty well, and they have a 3m long cable.. at the moment you're practically 'blind' about what's going on regarding your live data, if you only use the statistical summaries.

There should be enough python scripts around to dig quickly into that subject. (I'm still using Turbo Pascal on an old W98 PC with an AT bus GPIB card  )

regards Frank

[alm]

See the pictures posted below of my setup.

I notice the Fluke 731B and 730A on top. Do you track those, and do you feel they contribute anything to maintaining the volt in addition to the 732A/732Bs?

[notfaded1]

I'm not sure if that's lens barrel distortion in the picture or the shelf is really sagging under all that weight!

Bill

[Bill158]

I'm not sure if that's lens barrel distortion in the picture or the shelf is really sagging under all that weight!

Yes it is really sagging under all of that weight!  There is about 1/2 inch of sag in the middle.  BUT it has been there for at least 8 years so I am not too worried about it.  The shelf is supported by 4 each 4"x4" posts at the corners of the ends.  Then there are 2 each 2"x4"s running between the ends at the front and back of the shelf, running under the shelf, and the 4" width is vertical just like a roof joist would be.  I use to have another 4"x4" post in the middle at the back but I took it out because it was interfering with being able to have full use of the middle rear of the bench surface.  I have had this bench for 50 years so I don't think anything will collapse.

I notice the Fluke 731B and 730A on top. Do you track those, and do you feel they contribute anything to maintaining the volt in addition to the 732A/732Bs?

Yes I do track both of those also on a weekly basis.  The only reason I do this now is that the F730A was the first real Vref I every had.  The 731B came in around 1996.  Both of these are amazingly stable given they are NOT in an oven.  Both of these have been under 1 ppm of drift for the last 15 years or more.  I use them as a kind of "sanity" check more that anything else.

Do I spot that correctly, that your case board really bends heavily under the burden of two 5440 and the 5200 with the 5105 HV amplifier on top?

I have thought about some way to add back support for the center of the shelf but I don't want to lose the use of the center of the bench.  Sometimes the equipment I am working on can be pushed over back of the bench surface because the feet are not at the very back of the instrument or the instrument has no feet.  One thought I had was to add a couple of supports for the shelf but about 1/3 of the distance from either end, for the rear 2"x4" and then maybe used a 1"x1" support.  The 1"x1" would be under a compression load so there is little to worry about it bending.

I had also thought about moving the F5200A with the 5215A amp on top over to the left side of the shelf.  But then there is cable lengths to consider.  The F5215A has a special cable/connector permanently attached to it to help prevent it from becoming a "widow maker" because of the ability of it to put out 1,200 V RMS AC.  This cable also provides a Kelvin connection for HI and LO inside the connector.  The connector is also "shielded" by a shell, again to prevent the operator from killing themselves.  I have thought about changing this cable to something I would construct but I have to be careful about what the insulation breakdown would be on this cable.  The LO side of the cable is at ground.  So the HI side has to handle up to about +/- 1,700 V DC.

Get yourself a cheapo 82357B adapter, they are working pretty well, and they have a 3m long cable.. at the moment you're practically 'blind' about what's going on regarding your live data, if you only use the statistical summaries.

I have thought about this and I may do this when I get "a round toit".  Just too many things to do and only so much time.  I am 80 so I don't have the energy that I use to have when I was 30.  My days are a lot shorter because of this.

BTW my name is Bill and I AM a TEA in the worse kind of way.  I have a lot more equipment that isn't on the bench!  Including at least 7 Tektronix scopes.  Show me something I don't have but I might need someday and I will buy it, get it working and calibrated and then set it aside for some future day when I might use it.  I really need to check myself into a clinic for TEAs and get treatment!

Bill

[Sprock]

Hi all

what a pleasure - this was a Christmas surprise.
Now need some more Digits 
A big thank to al the people who are involved
to get a dream true.

don´t know why picture does not show up right ! sorry