Multipexing a 3458A using a modified 3497A?

[rhb]

How do other people log multiple references contemporaneously over long periods? 

After some study it appears I can modify a 3497A to multiplex 100 channels to the rear panel connectors.  The change is remarkably simple.  Remove the rear panel connector protection circuit, short the internal DMM inputs and connect the DMM Inguard board connector to the rear panel connector and the3458A to the 3497A.  Then just send the proper GPIB commands to the 3497A and 3458A. 


I've attached the relevant schematic segment.  There are several ways to reroute the signals. E1 is a 400 V MOV.

Has anyone done this?  It's a 10x expansion of the 44492A 10 port card for the 3457A.  I'm setting up to use those to measure 3x used LM399s, 3x used LM317s, 3x AD584s and the PSU rail for which I am assembling  permanent test fixture.

[Kosmic]

There's already some threads about multiplexer/scanner:

https://www.eevblog.com/forum/projects/diy-low-thermal-emf-switchscanner-for-comparisons-of-voltage-and-resistor-stand/msg610755/#msg610755
https://www.eevblog.com/forum/metrology/scannermultiplexers-for-voltage-references/
https://www.eevblog.com/forum/metrology/teardown-data-proof-160a/msg977205/#msg977205
https://www.eevblog.com/forum/metrology/low-thermal-emf-scanner-and-ordinary-non-latching-relays/msg1064900/#msg1064900

It's a rabbit hole in itself.

[Dr. Frank]

How do other people log multiple references contemporaneously over long periods? 

..

I try to achieve a good transfer uncertainty for my about 12 references, by manually changing the cable connection within about 30minutes.
The e.m.f. error seems to be neglectable, it's a few 100nV at most, which gives 0.01 .. 0.05ppm error for my configuration.

The temperature stability of my 3458A (~ 0.45ppm/K) yields more severe changes, because the room temperature rises by 0.2 .. 1°C during my manual measurements, due to heating by the fluorescent lamp, and my own 100W dissipation. I've monitored the very same temperature rise during the measurements in a lab of a buddy - volt-nuts.

So I also thought about using the scanner cards of my 3488A, or inside my 34970A, which might give 1..2µV e.m.f., but will not heat the lab.

I think, these aspects might be helpful in addition to the threads by Kosmic.

Btw.: I'm using cheap CAT 5 cables for my scanner cards.. they have 4 drilled pairs of pure copper, the wire isolation is very high, so can be used for resistance measurements and they are shielded twice.

Frank

[Kosmic]

On my side I'm planning to start by using a 3488a I already have at hands.

But eventually, I would like to try building something better (So many project  ). I really like doktor pyta's design. Re-Using RJ45 connectors and cables is pretty clever.

https://www.eevblog.com/forum/projects/diy-low-thermal-emf-switchscanner-for-comparisons-of-voltage-and-resistor-stand/msg1366769/#msg1366769

[voltsandjolts]

There is a version of 34970A without the DMM, just a remote controlled switch unit. OPT 001 NO DMM.

You could design a low thermal switch card for the 34970A based on this 20 channel card. At least it gives a starting point with board outline and firmware.

[Kosmic]

There is a version of 34970A without the DMM, just a remote controlled switch unit. OPT 001 NO DMM.

You could design a low thermal switch card for the 34970A based on this 20 channel card. At least it gives a starting point with board outline and firmware.

Designing a new relay card is also a good idea. You save on firmware, supply, enclosure etc ...

the 3488a/3499a (cards are compatible between the 2) would probably a better starting point though. There's plenty on ebay and they are really cheap.

[rhb]

Thanks.  I assumed there were, but was 3497A specific in my queries. 

I'm now very curious as to why there are no mentions of the 3497A.  I'd originally planned to build one out of a 44421/2A card, but the found the connector blocks which led to finding a 3497A to hack.

My original plan was to mount a single 44421/2A in a GPIB extender case with an Arduino controller.  I have 5 spare relay boards without terminal blocks and 3 extenders for cases which would neatly take 22 banana jacks

nV level measurements are a rabbit warren as are any high precision measurements.  Try millionths of an inch flatness.
 
The current project on the bench is mounting 3x LM399s, and 3x  LM317s, all used with 3x new AD584JHs to scan with my 3457A/44492A.  The enclosure will be sealed with all wires shielded to the common PS ground at one end only.  It will have two *large* thermal masses on the outside and then a large heavily insulated enclosure such that the thermal time constant will be very long to keep the EMFs as stable as possible.

I did the physical mounting last night.  Next task is to supply power and wire the references and power rail to the 44492A connector block.

The 3497A has a NO DMM option but it's not clear I can route the relay matrix by removing the DMM.  It might be just move a jumper and remove the DMM, but all I have is a scan, so no way to search the documentation.

Have Fun!
Reg

[alm]

I'm now very curious as to why there are no mentions of the 3497A.  I'd originally planned to build one out of a 44421/2A card, but the found the connector blocks which led to finding a 3497A to hack.

I guess because the 3488/3499 are more modern and ubiquitous. Same with the Keithley 700/7000 series.

The 3497A has a NO DMM option but it's not clear I can route the relay matrix by removing the DMM.  It might be just move a jumper and remove the DMM, but all I have is a scan, so no way to search the documentation.

There's software, like Adobe Acrobat but I'm sure there are also free or cheaper solutions, that can run optical character recognition (OCR) on a scanned PDF and make the PDF searchable by invisibly embedding the text. This does require the quality of the scan to be good enough for the OCR to work.

[Kosmic]

Thanks.  I assumed there were, but was 3497A specific in my queries. 

I'm now very curious as to why there are no mentions of the 3497A.  I'd originally planned to build one out of a 44421/2A card, but the found the connector blocks which led to finding a 3497A to hack.

The 3497A is rather old, uncommon and doesn't reach the specs for metrology usage. As it get older, you have more chances the relay would be bad.

On top of that, you can find more modern HP alternative that doesn't require a hack to be use as a multiplexer. Those alternatives are readily available and cheap.

[rhb]

Same relays as my 44492A.  Is a 3457A/44492A 10 port scanner not metrology grade?  That's news to me.

These things lived for years doing stuff like strain gauges in a structural engineering lab.   Not many people want 100 channels of I/O.  I'm considering using a 16501A or two for the DUTs.

Have Fun!
Reg

And I've got spares for all the boards. ;-)

[Kosmic]

Is a 3457A/44492A 10 port scanner not metrology grade?  That's news to me.

The 3457a is a 6 1/2 multimeter, I don't think it is no. Don't even have high input impedance when measuring 10V  (30V range at 10M\$\Omega\$)

As for the 44492A card I did not find the specs but thermal offset is probably in the lower uV range (not nV).

[Kosmic]

These things lived for years doing stuff like strain gauges in a structural engineering lab.

The life span of those relays is evaluated in number of actuation. Normally, the older they are the more they have been actuated.

[Kosmic]

I don't want to sound too negative. It's probably going to work fine for your application (measuring LM317, AD584 etc..).

I'm simply trying to explain why people don't use the 3497A in conjunction with the 3458a (or any other 8 1/2 meter).

[rhb]

Is a 3457A/44492A 10 port scanner not metrology grade?  That's news to me.

The 3457a is a 6 1/2 multimeter, I don't think it is no. Don't even have high input impedance when measuring 10V  (30V range at 10M\$\Omega\$)

As for the 44492A card I did not find the specs but thermal offset is probably in the lower uV range (not nV).

The 3457A reads 7.5 digits via GPIB which is exactly what you would do if you were evaluating reference aging 30 years ago at a national lab or OEM.

"Years" of use in an engineering lab might involve turning it on 20 times for a few days.  HP has a long standing reputation for the best metrological grade relays ever made.  Boards with bad relays are spare parts.

In the end, you pay your money and takes your chances.  I got the 3497A *with* 5 relay cards very cheap (price of relay cards alone cheap).  I bought a cheap, very old  3458A parts unit.  If the A3 board is good, major score, if it's bad I'll have  a cal'd instrument for around $5k.

I have a hole card most don't. I am skilled at multichannel data analysis and can quantify and correct for such errors.

My particular interest is in the physics of the "aging process".  At 100 channels LTZ1000s are outside my budget.  Something in the $10 per unit range would be more likely.  After successful resuls with sub $1 references.

Have Fun!
Reg

[Kosmic]

There's no specs for the high res mode of the 3457a not sure I would call that a 7.5digits meter.

As for the 44492A card, the service manual of the 3457a mention 3uV thermal offset. Which is similar to most 3488/3499 relay cards.

Anyway, enough of this, have fun Reg!