Sensible approach to HP3456A check/calibration?

[Wallace Gasiewicz]

You should be able to evaluate and. if necessary. adjust the 3456 with the resistors you ordered.
Using the lower voltage and lower ohm ranges (SAME ADJUSTS) I would use copper wire connects directly to the posts without any adaptors to BNC. 
Get some decent lower ohm resistors, The best you can get perhaps on ebay. The bigger the resistor in watts the better. Even .1% resistors will be good enough for hobby use.Telephone wire (BELL WIRE)_ is typically used. this is just copper, no plating.   Single strand.,comes in multiple wire cables for wiring a house for telephone Rig up some way to attach the resistors to the wire.
If you can get axial lead resistors you can attach directly to the 3456 posts.  And just attach the other two posts in the four wire configuration with clean copper wire.

The 3456 is a really good unit and much better than what 99% of hobbyists  need.

[tatel]

The 1 M resistor could be good enough for comparing the ranges. One may not even need it that good with a few more repeats. 0.1% accuracy is of limited use as the test with 100 K and from the 100 K range should give a much better value for the scale factor.
For a direct long term stability test 1 M and even more 10 M is not that good  anyway. It would depend a lot on cleanness and maybe leakage.

The resistors for the 3456 performance test / calibration are a different thing - they are for the absolute value to check or do the adjustment. Especially for the 1 M and 10 M range it is difficult to get very accurate long term stability. The relative comparison from 1 range to the next is relatively easy in comparison and is less demanding on the resistors.

This did go under my radar somehow. It will have to wait until next month. 1 MOhm one is still an order of magnitude worse than what the manual calls for. 10 Mohm one fits that bill. Good enough?
Caddock USF340, 1 MOhm, 0.01%, 5 ppm Thick film. Digikey number USF340-1.00M-T-5PPM-ND
Caddock USF340, 10.0M, 0.01%, 5PPM Thick film. Digikey number USF340-10.0M-T-5PPM-ND

And this 1 GOhm could also fit the bill? I got it right first time.
Ohmite MaxiMox, 1 GOhm, 1%, 50 ppm Thick film. Digikey number MOX2FE-1000M-ND

I can't find any better in Digikey's stock. I would gladly take advice about any of these three resistors. If not that, what could I buy in €20-30 range?

[tatel]

You should be able to evaluate and. if necessary. adjust the 3456 with the resistors you ordered.
Using the lower voltage and lower ohm ranges (SAME ADJUSTS) I would use copper wire connects directly to the posts without any adaptors to BNC.
Get some decent lower ohm resistors, The best you can get perhaps on ebay. The bigger the resistor in watts the better. Even .1% resistors will be good enough for hobby use.Telephone wire (BELL WIRE)_ is typically used. this is just copper, no plating.   Single strand.,comes in multiple wire cables for wiring a house for telephone Rig up some way to attach the resistors to the wire.
If you can get axial lead resistors you can attach directly to the 3456 posts.  And just attach the other two posts in the four wire configuration with clean copper wire.

The 3456 is a really good unit and much better than what 99% of hobbyists  need.

I got Vishay Z-foil series, 0.01%, "0.2 ppm" for 100R, 1K and 10K. 100K is about the same, but "1 ppm". Radial. I was thinking about milling some thick scrap aluminum board to get these resistors cased in. I have that aluminum lying around in the workshop. I don't really feel comfortable about having these resistors tumbling around... or bending the legs, although it can be carefully done with some pliers. But I think the less I fiddle with these, the better. That means I need some binding posts. I would clip the 4-wire thing to the binding post, or have some short length of copper from an Ethernet cable, I guess it would be about the same than telephone cable, right?

Would gladly take any advice about "good enough" but no "so crazy expensive" binding posts. Something in the line of these resistors, if possible. But if "crazy expensive" is really needed, then it's needed, and I will buy it. It will have to wait a couple of months, perhaps three, that's it.

Also about casing in these resistors into the DIY aluminum boxes. I.E, would it be good/bad having  resistor bodies in contact with the aluminum? Should i put some isolating stuff instead? Long or short legs? And so on. Any links would be greatly appreciated.

This is the (of course chinese) stuff I got to make the BNC-banana adapters. I know these BNC connectors are really too cheap to use them with €30-40 resistors. I bought 1 mm thick board because that's what DER EE-5000 seems to need. I also have "usual thick" board lying around.

https://aliexpress.com/item/1005005610197484.html?
https://aliexpress.com/item/33021792391.html?
https://aliexpress.com/item/1005004083068192.html?

I agree, 3456 is overkill for a guy like me, talk about going after some trawler with a battleship. But I can be confident that it will take a long time for me to outgrow it...

[Wallace Gasiewicz]

You should use only copper to copper connections 
With different metals you get some very small volts produced that will interfere, It will disturb the last few digits.
This really applies to the LOW ranges, you cannot see microvolts an the higher ranges.It applies to Both the Lower  Volt and the Lower OHM readings.
Even a solder joint causes discrepancies. Maybe especially a solder joint.This is real and something I learned in calibrating my 3456. 

But how close do you really need to be?

[tatel]

Well, at the very least, I can guarantee you I will make my measurements with these resistors before and after casing them into boxes. To ask for advice just to reject it after received would be really a stupid thing to do.

In addition, I understand Seebeck effect is a thing so your advice makes a lot of sense.

I see casing resistors into boxes is common (bad?) practice, though. So there is something to be learned here. Good to try both ways, then.

I guess final question should rather be "How close to spec a hobbyist could reach with minimal expenses?. I think, unless those resistors could be measured in a known good, currently calibrated device, that hobbyist will never know it for sure. Same for voltage references, not to mention higher voltage ranges.

[Silicium81]

Good link for the HP3456A: https://xdevs.com/fix/hp3456a/ (for me, opens only with 'microsoft edge' 
See the add temp sensor (with external BNC)  aproximatly middle page 'Temperature sensor'

[iMo]

Good link for the HP3456A: https://xdevs.com/fix/hp3456a/ (for me, opens only with 'microsoft edge' 
See the add temp sensor (with external BNC)  aproximatly middle page 'Temperature sensor'

I installed the LM35 into my 34401A years back (glued to the aluminum shield over the ADC, powered from outguard's 5V and its output wired on a free pin on the RS232 connector). Thus I can get nice TC sweeps of my DMM and do the TC compensation of the results..

[Wallace Gasiewicz]

About putting the resistors in a box. 
This affects the temp around the resistor. 
Most resistor "Standards" are in some sort of medium to keep temp as stable as possible.  Many are in oil baths. 
The resistor boxes are, of course, in boxes. So any attempt to keep temp stable is quite appropriate. I guess you have to decide if the box would maintain stability or cause more heating than leaving them out in the open. 

 Something that is a bit similar that I did: 
I rebuilt an old Lab Grade Power supply. I used a high quality Zener but no effort at temp compensation. The thing works quite well. However I discovered that if I even blew on the diode the voltage of the PS would change. I do not remember exactly how much but it was a few millivolts not just microvolts ( I think). Blowing hot air from a hot air gun at the board produced greater change. 
I think the temp stability of any Zener is very much worse than your resistors however. 
I am just relating this story to elucidate the importance of temp stability.  I would think an enclosure would make stability greater than no enclosure, but that would also depend on how much heat the resistor creates itself.  Perhaps someone with more knowledge could help.Perhaps try hooking up your lowest ohm resistor and experiment with some hot air or cool air.
I really do not know the appropriateness of a metal box vs a wood or plastic,    There is all sorts of RF these days that emanates from our computers and WiFI. It can induce currents in our sensing wires, I don't know what this RF would do to the Aluminum box, it seems that the resistor boxes are mostly non metallic.

[tatel]

About putting the resistors in a box. 
This affects the temp around the resistor. 
Most resistor "Standards" are in some sort of medium to keep temp as stable as possible.  Many are in oil baths. 
The resistor boxes are, of course, in boxes. So any attempt to keep temp stable is quite appropriate. I guess you have to decide if the box would maintain stability or cause more heating than leaving them out in the open. 

 Something that is a bit similar that I did: 
I rebuilt an old Lab Grade Power supply. I used a high quality Zener but no effort at temp compensation. The thing works quite well. However I discovered that if I even blew on the diode the voltage of the PS would change. I do not remember exactly how much but it was a few millivolts not just microvolts ( I think). Blowing hot air from a hot air gun at the board produced greater change. 
I think the temp stability of any Zener is very much worse than your resistors however. 
I am just relating this story to elucidate the importance of temp stability.  I would think an enclosure would make stability greater than no enclosure, but that would also depend on how much heat the resistor creates itself.  Perhaps someone with more knowledge could help.Perhaps try hooking up your lowest ohm resistor and experiment with some hot air or cool air.
I really do not know the appropriateness of a metal box vs a wood or plastic,    There is all sorts of RF these days that emanates from our computers and WiFI. It can induce currents in our sensing wires, I don't know what this RF would do to the Aluminum box, it seems that the resistor boxes are mostly non metallic.

At a glance it looks that most (all?) designs leave a good bit of space into the box. Hence, I guess these boxes are mainly to avoid air drafts and resistor heating would have a lesser effect on stability?

OTOH, a guy could mill some fins on the outside of the box and a tight slide fit for the resistor inside... but that would be more of a heatsink? I don't know if that would be useful, perhaps it would make a nice experiment to do with a worse quality radial resistor.

I agree that a plastic box would quite probably be enough, easier and cheaper.

[Conrad Hoffman]

Rule #1 for most resistor metrology is you don't dissipate any more power than absolutely necessary to make the measurement.

[Kleinstein]

A box makes some sense. It protects the resistor and can isolate the resistor from mechanical stress via the 4 wires and gives a defined points to split in drive and sense.
If wanted and for a accurate resistor it may make sense to have at least a crude temperaure sensor in the box - it does not have to be something special, a thin film PT1000 or 10 K NTC or possibly just a diode can be good enough.

For just a check on 1 DMM one does not really need a full set of accurate resistors. It is hard to get some better than 0.01 %. One would need 1 resistor with a known values - here the exchange with somewhat with a good meter could be a way. A measured resistor may be a part that is still sensible to send per mail.

Other ranges could be checked by measuring a resistor in different ranges and if really want an extra 3S : 3 P experiment. This does not need super stable resistors - just short time stability.

[tatel]

Rule #1 for most resistor metrology is you don't dissipate any more power than absolutely necessary to make the measurement.

Yep, that makes sense.

[tatel]

About Seebeck effect: I understand gold doesn't corrode and has like just 0.3V/ºC Seebeck coefficient with copper.

Then you put in some solder to make, well, anything? Wouldn't that defeat the intended purpose of gold plating?

I have read a paper from fluke (Watch Out for Those Thermoelectric Voltages!). There appears a little table of materials but the gold-tin couple wasn't there.

A guy would think that tin plating would be more sensible if tin is going to be in the game anyway. There is something here that I still need to learn, it seems.

[Kleinstein]

The other materials only matter if there is a significant thermal gradient and temperature difference in the other material. Especially a gold plating is usually thin and thus gets very little temperature difference. Also the solder bridge is thin and there is ideally not much heat flow and thus not much thermal gradients.
The purpose of gold plating is to get good contacts and in some cases easy soldering. In most cases the solder will mix with the gold and thus no more gold layer after soldering - maybe a residue of the nickle layer that is often below the gold. Chances are that ENIG gold is more nickle than gold.


One does not need the Seebeck coefficients for all material pairs. It is enough to have them relative to a ref. material (usually platinum if choosen for data collections, but copper would also do).

With not much power there are little temperature differences and thus little thermal EMF. If in doubt one could exchange the polarity of the resistor and measure in both directions. This would than average out most (the average part) of the error from thermal EMF.

[TUMEMBER]

About putting the resistors in a box. 
This affects the temp around the resistor. 
Most resistor "Standards" are in some sort of medium to keep temp as stable as possible.  Many are in oil baths. 
The resistor boxes are, of course, in boxes. So any attempt to keep temp stable is quite appropriate. I guess you have to decide if the box would maintain stability or cause more heating than leaving them out in the open. 

 Something that is a bit similar that I did: 
I rebuilt an old Lab Grade Power supply. I used a high quality Zener but no effort at temp compensation. The thing works quite well. However I discovered that if I even blew on the diode the voltage of the PS would change. I do not remember exactly how much but it was a few millivolts not just microvolts ( I think). Blowing hot air from a hot air gun at the board produced greater change. 
I think the temp stability of any Zener is very much worse than your resistors however. 
I am just relating this story to elucidate the importance of temp stability.  I would think an enclosure would make stability greater than no enclosure, but that would also depend on how much heat the resistor creates itself.  Perhaps someone with more knowledge could help.Perhaps try hooking up your lowest ohm resistor and experiment with some hot air or cool air.
I really do not know the appropriateness of a metal box vs a wood or plastic,    There is all sorts of RF these days that emanates from our computers and WiFI. It can induce currents in our sensing wires, I don't know what this RF would do to the Aluminum box, it seems that the resistor boxes are mostly non metallic.

At a glance it looks that most (all?) designs leave a good bit of space into the box. Hence, I guess these boxes are mainly to avoid air drafts and resistor heating would have a lesser effect on stability?

OTOH, a guy could mill some fins on the outside of the box and a tight slide fit for the resistor inside... but that would be more of a heatsink? I don't know if that would be useful, perhaps it would make a nice experiment to do with a worse quality radial resistor.

I agree that a plastic box would quite probably be enough, easier and cheaper.

An interesting idea to improve resistance to environmental conditions
https://www.eevblog.com/forum/metrology/teardown-standard-resistors/msg5154711/#msg5154711
branadic solved the problem in a low-cost way

[Wallace Gasiewicz]

Silicone  Fluid is used in transformers these days instead of oil because it is not flammable.There are Silicone Products that are made for heat transfer. Silicone pads for mounting transistors and ICs are one common use, but these come in sheets. There is also Silicone white heat sink paste that I use frequently.  I wonder if there is a silicone caulk that hardens that is easily available that one could "Pot" the resistors into? I think this has been discussed but I don't remember the conclusion.
You would not need anything as water tight as Branadic's rather ingenious solution.  A big straw would work. Or a clear plastic tube from the hardware store.

[tatel]

An interesting idea to improve resistance to environmental conditions
https://www.eevblog.com/forum/metrology/teardown-standard-resistors/msg5154711/#msg5154711
branadic solved the problem in a low-cost way

Alas, these things doesn't seem to be available on eBay anymore, and it looks they will remain unavailable unless Russia wins WW3. At that point I don't think anyone would care about it. It looks these are for axial resistors anyway.

Dr. Frank recommendation is to embed the resistor in an aluminum block https://www.eevblog.com/forum/metrology/t-c-measurements-on-precision-resistors/msg464413/#msg464413

While he does it to have a temperature sensor near the resistor, it is obvious there's no problem about the resistor becoming too hot; au contraire this heatsinks the resistor, transmitting that heat to the box.

For resistors like mine (Vishay VPG, Z and S series), which are not in metal cans but epoxied,  the utility of heatsinking will be less. Epoxy will not transmit the heat as easily as a mineral oil bath and metal can. But it will help anyway. Z-series datasheet says max spread TCR is 0.2 +/-0.6 ppm/ºK (100R-600K), but PCR is 5 ppm/ºK.

So, having it heatsinked would be a Good Thing (TM), I think.

BTW, it's said that epoxy body is the reason these resistors have quite a yearly drift? It seems it absorbs humidity, which changes its behavior. In that case, having the whole resistor as sealed as possible, would also help. I got a couple ideas about putting them into a box.

Not hoping that resistor will become as good as these metal canned ones, but if I can get some windfalls after doing something to avoid mechanical stresses, why not?

Anyway I'll get measurements with just the naked resistors, too. To have, say, a control group to compare against.

[tatel]

" temperature sensor that goes to a BNC conector implanted"

really ?   
I have some feeling you don't need to do anything with this unit .... including adjustment ....
it would be sufficient at beginning, then you figure out ,  i would say cool down inpatient to fix it (my initial advises under assumption - it scrap yard salvage unit)

Well, memory didn't served me well.... There are two BNC connectors implanted on the back panel, and they are wired up to the stick that changes from/to front/back panel. Serial number prefix is 2201A

[guenthert]

Wut?  No pic of 'the stick that changes from/to front/back panel'?

[tatel]

It supposedly was on the second attachment on previous post... let's try again, and this time with a closer view too

[Wallace Gasiewicz]

I know that I am changing the topic a bit. But since there are very knowledgeable people watching, I will attempt a question about one of my 3456 units.
The unit will not align on the low ( 100mV ) range it reads about 150 counts too low at max adjustment. I believe this is about 15 uV.  The input switching has been repaired by me and a few transistors replaced.  At this point the signal as measured by a good 3456, is a few counts lower at the output of the switching circuit. This is measured at J19.  When I measure at the end of the shielded wire that connects to the amplifier section , the signal decreases to about 150 counts low. basically the reading that I see on the display of the problem unit, This is a solid wire , but measuring at one end or the other does show a difference.  Replacing the wire does not change the readings. 

Since the manual states that problems with just the 100 mV range are in the amplifier section. I wonder if anyone has any suggestions. I have done a lot of measuring and cannot find a particular problem.
I am using a recently calibrated (Dec 2023) General Resistance DAS 45 as the volts source. I have the cal sheet.
One unit, newer with soft buttons, calibrated (adjusted) easily.  This unit will not adjust on the low range, although the readout is stable and consistently low.

[guenthert]

    Ah, thanks, so the status of the front/rear switch is made available through those custom BNC connectors in the rear.  For some reason, I thought that a means to automate the switching (like for the 3457A), e.g. using a relay, would have been deployed.

[tatel]

FWIW, 3456A has been on since February, 16. There are some preiminary measurements with non-calibrated resistors.

100R, 1k, 10K are Vishay Z-201, now called Y1453. These are the epoxy body ones, 0,01% tolerance and supposedly .2 ppm TCR.

100K is Vishay S102K, now called Y0062, also epoxied, 0.01% tolerance and supposedly 1 ppm TCR

1M is RESI MMFR5215B1M00V9, 0.1% tolerance, 5ppm TCR

10M is Ohmite MOX-750231005FE, 1%, 25 ppm TCR

1G is Ohmite MOX-2-121007FE 1% 50 ppm

Measurements taken at 6 digits, 100 NPLC, Auto ZeroOn, autorange. 20ºC

            |     100R      |         1K       |       10K          |     100K           |        1M           |       10M          |          1G         |
-------------------------------------------------------------------------------------------------------------------------------------------------------
2W      | 100.4031    | 1000.703     | 10.00050+3   | 100.0023+3    | 1000.039+3    | 9.98653+6      | 1033.732+6    |
-------------------------------------------------------------------------------------------------------------------------------------------------------   
2WOC | 100.3818    | 1000.671     | 10.00046+3   | 100.0021+3    |        OL           |       OL            |        OL            |
-------------------------------------------------------------------------------------------------------------------------------------------------------
4W      | 99.9975      | 999.947       | 10.00014+3   | 100.0018+3    | 1000.038+3    | 9.98674+6      | 1032.535        |
------------------------------------------------------------------------------------------------------------------------------------------------------
4WOC | 99.9995      | 999.947       | 10.00013+3   | 100.0018+3    |        OL           |        OL            |        OL           |

5 Volts DC from DMMCheck:  4.99947-4.99955
5 Volts AC    "             "            5.00083-5.00095

Without having these resistors calibrated, I don't know if it's in spec or not, but at least it seems it's not jumping like a goat.   
       

[tatel]

I was going to do Ohms Performance test but I'm not sure I'm getting it rigth. Please see attachmente. As I understand it, the red marked steps would be done in autorange, then the green ones would be done in manual 100K range?

Last thing, I have noticed some hum coming from 3456A, weak enough to be noticed only at night. Is that normal?

[guenthert]

[..]
5 Volts DC from DMMCheck:  4.99947-4.99955
5 Volts AC    "             "            5.00083-5.00095

Without having these resistors calibrated, I don't know if it's in spec or not, but at least it seems it's not jumping like a goat.   
       

    Same for the DMMCheck.  It comes with a paper listing its values before shipping (as measured by the maker's DMM, but iirc no uncertainty is given).  Its poor connectors aren't helping.  It's ok for what it was meant, but calibrating a 6 1/2 digit VOM is outside its comfort zone.