DIY Simple DVM Cal Device

[mawyatt]

After retiring last year I purchased an HP34401A off eBay for repair, then repair an Agilent 34401A. Earlier had repaired a couple Tek 2465 scopes, so beginnings of a new to me lab at home

While revisiting and old unique voltage divider circuit (used a flip-flop) and measuring, the result the two DVM differed by a small amount, then realized I needed something to calibrate these 34401A and the two old Flukes (87 and 77) we had laying around. Instead of sending everything out for calibration, decided to get a new Keysight 34465A DVM and use this to help calibrate/check the other instruments.

Anyway, long story short a few custom PCBs were designed and fabricated. Features are listed below.

A) Resistor "On-DVM" PCB.
  1) 4 resistor positions selectable by jumpers for 4 wire measurements.
  2) Each resistor can be a Leaded type, 0603 type, 0805 type or 1206 type, or various combinations (for adjusting value).
  3) 3 Thermistors types to read PCB temperature, leaded, 0603 or 0805. Separate pins which can be read by DVM or another meter.
  4) Directly connects to DVM by Banana plugs, no leads required.
  5) Rotated CCW 90 degrees becomes and good zero Volt & Ohm (2 or 4 wire) reference.

B) "On-DVM" Calibrator PCB,
  1) Functions selectable by jumpers.
  2) Single resistor position, can be Leaded type, 0603, 0805 or 1206, or various combinations.
  3) DC Voltage reference, can be various voltages (2.5, 5, 10 volts with different pre-regulator) and has somewhat standard "footprint" of SOIC 8 pin. Pin 2 Vcc in, Pin 4 And, Pin 6 Vout, Pin 5 Trim. Layout supports various trim techniques for various references, and includes trim pot.
  4) AC Voltage reference, creates accurate 50% duty cycle square waveform of Vref, thus rms value is Vref/2. Supports various crystals & frequencies , HC49 or leaded type. Jumper enables AC waveform and another jumper selects frequency from 14 stage counter (CD4060). Has discrete output buffer for lower impedance.
  5) 3 Thermistors types to read PCB temperature, leaded, 0603 or 0805. Separate pins which can be read by DVM or another meter.
  6) Standard Power 12V connector and also 2 pin battery connector, both reverse protected.

I built the 1st of the Resistor PCBs with 5ppm resistors, 100 ohm 0.05% 1206, 1K Vishay 0.1%, 10K 0.1% and 100K 0.1%. Used a hermetic 10K 1% leaded thermistor. 2nd PCB used all 1206 0.1% 15ppm resistors of 1K, 10K, 100K & 1M and a 0603 10K 1% thermistor.

The Calibrator used a 1206 10K 0.1% 10ppm resistor, a 10K 0603 1% thermistor, a 4.096MHz 20ppm Crystal and a Analog Devices ADR4550 5 volt reference.

All these used gold plated Banana plugs for direct DVM mounting, very convenient

With a nominal room temperature of ~77F they measured about ~82F or ~28C, so the inputs to the DVMs seems to be about ~5F warmer than ambient.. The Keysight 34465 has a nice feature of directly converting 5K and 10K thermistors to temperature, also the trend chart is really nice for observing readings of voltage, resistance and temperature.

All readings are very stable on the 34401As or 34465A after a time to settle in to DVM input temperature. Of course this isn't NIST certified, nor even ultra-precision, but adequate for my needs

Here's some images of the resistor PCBs.

Best,
 
     

[mawyatt]

Don't know what's going but the images are getting mixed up and I can't upload with another post???

Best,

[mawyatt]

Try again.

[mawyatt]

Another try!!

[mawyatt]

Here's the Resistor (2) and DVM Cal connected up to the two 34401A and 34465A DVMs. The top 34465A is measuring AC voltage and shows a SD of 36uv on a 2.498501v average 19,000 measurements. The 34401As are measuring 1M and 100K ohms from 0.1% 15ppm and 0.1% 5ppm devices respectively.

Edit: Blocked the direct path from the Air Conditioner vent blowing over the DVMs and the SD dropped to 8uv on the 2.5V AC measurements and 4uv on the 5V DC measurements.

Best,

[daisizhou]

Where did you buy the malfunctioning 34401 multimeter?

[mawyatt]

Off eBay, both a HP 34401A and a Agilent 34401A.

Best,

[wizard69]

Those are really nice!    Which brings up the question do you intend to sell the boards, completely assembled or not?   I don't have such high precision meters in my possession but such boards would ideal for proofing lesser meters that I do have.   I really like the idea of being able to plug directly into the meter.

By the way very nice banana plugs, do you have a part number for those beautiful gold plated plugs?   

I've been looking at some of the other Voltage reference solutions but this direct plug into the meter just looks to be well thought out.

[mawyatt]

Those are really nice!    Which brings up the question do you intend to sell the boards, completely assembled or not?   I don't have such high precision meters in my possession but such boards would ideal for proofing lesser meters that I do have.   I really like the idea of being able to plug directly into the meter.

By the way very nice banana plugs, do you have a part number for those beautiful gold plated plugs?   

I've been looking at some of the other Voltage reference solutions but this direct plug into the meter just looks to be well thought out.

Sorry for the late reply, I missed this somehow, senior moment

The connectors plugs are from eBay, but it took me a few sets purchased, over 4 months to finally get what I liked, then the eBay supplier shipped the wrong jacks...so back to square one! I'm waiting on another set of jacks, and if they are right then I'll try another reordereding of the plugs. I have a precision based LTZ1000 and LM399 design waiting for these connectors!!

These are so easy and convenient to use, I just plug and swap between the two 34401A and 34465A DVMs to check on how they compare. Not intended to be standards, but hard to beat for ease of use!!

They also support many different components, so for the resistors you can mix and match with 0603, 0805, 1206, 1210 or leaded devices. The reference voltage supports many of the popular ref sources that use a SOIC footprint, including trim components. The Analog Devices ADR4550 I have in the shown PCB is really good, been very stable over the past couple weeks, same goes for the AC waveform (250Hz form 4.096MHz crystal).

PM me if you are interested in the PCBs, think I still have a couple. If enough folks are interested I may reorder more PCBs (and parts).

Best,   

[mawyatt]

Update.

I've finally built up a Precision 10.000V Reference based upon the LM399, the 1st one I got off eBay for $5 wasn’t good, drifted a lot. So I purchased from Mouser, a couple LM399 for $15 each and a single LTZ1000 for ($60). Also designed a couple PCBs to create an ultra-stable but low cost voltage reference. 1st PCB was based on the LM399 and a typical type reference device (ADR01 for example) and included a precision 1.000V and 0.1000V outputs based upon a simple voltage divider and unity gain buffers. Since I didn’t want to spend $ for low TC and accurate Vishay resistors, and also not $ for ultra-precision op-amps I used an OP-07 with a feedback resistive network based around low cost 0805 or 0603 low TC resistors and a trim pot, also created a trim network for the ADR01. For the 1 and 0.1V outputs I didn’t want to use trim pots, and decided to try something instead of simple 2 resistor 10X divider. I used 6 low TC but only 0.1% accurate resistors of the same value to create a 10 to 1 voltage divider (works beautifully!!), then used a new part to me which is a dual op-amp made on a low voltage low cost CMOS process that only supports ~6 volts Vcc. This part is probably a direct copy of an older paper on achieving almost 0 TC with a simple CMOS process utilizing commutating techniques. The part has a bad offset spec (+-20uV) but ~0 TC spec, so I figure I could live with the offset error as long as it didn’t change.

Well got some time and built up the 1st PCB as shown using cheap stereo gold plated speaker jacks (banana type) for low thermal EMF. I used dual gold plated banana jacks with solid copper wire (not plated) test leads, the solid copper wire I got from some stiff CAT5 cable conductors.

This works well!! After initial trimming, the 10 volt output is 10.000000 volts average of ~35,000 samples (yep that’s 1uv resolution on 10 volts!!) and direct displays 10.00000+-1. The 1V output is 0.999973V (SD ~600nv!) and doesn’t change, the 0.1V output is 0.100005V (SD ~ 200nv!) and also doesn’t change. This was over a period when the AC wasn't cycling much, later is started raining and the humidity increased and the AC cycled more often. The LM399 voltage is 7.036181V and very stable. I'm going to let this run for a couple weeks then try a readjustment to 10.00000V if required.

Makes me wonder just how good the LTZ1000 will be!!!

Note the design supports various component sizes, like 0805 and 0603. You can see the LM399 in the thermal plastic white case.

Just went to start assembly of the Precision Reference PCB based upon the LM399 & the LTZ1000 (also has precision AC output), only to find out I forgot to order the special LT1013 Op-Amp that's used with the LTZ1000

So will be awhile before I can get to this assembly.

Best

[mawyatt]

Couple more images.

Best,

[mawyatt]

Here's the unpopulated PCB that includes the LTZ1000 & LM399, in bottom middle and right middle respectively.

Best,

[dgminala]

I juat found this thread and was wondering after you got the LM399 and LTZ1000 voltage reference board(s) built, how it performed after being powered on for several months.  Typically, these references drift quite a bit after first being powered up, then settle down and are remarkably stable after 6-12 months of constantly being powered up.  The longer they are powered up, the more stable they are.
I'm also a volt-nut and am hoping that you could share the schematic and PCB design files for your project.  I'd like to try to build a copy of it and maybe compare results.

Thanks for sharing such an interesting project.
DaveM

[tatel]

+1

[strawberry]

I juat found this thread and was wondering after you got the LM399 and LTZ1000 voltage reference board(s) built, how it performed after being powered on for several months.  Typically, these references drift quite a bit after first being powered up, then settle down and are remarkably stable after 6-12 months of constantly being powered up.  The longer they are powered up, the more stable they are.

34401A drift more than LM399 it self. done couple day test and cant tell difference.
test lead movement/relocation may change result by some ppm. maybe noise or leakage

[mawyatt]

Well they've been running almost 24/7 for over 2 years now and seem to be very stable. We made some 3D boxes and placed the PCBs inside.

In the mean time we've "Collected" a new KS3465A (2 months), another KS3465A (~1yr), light case KS34465A (~2yrs), DMM6500 (~1yr) a SDM6500 (~2yrs) and a HP34401A and AG34401A. All purchased new except 33401As. These were acquired to support a Custom HV (+-100V) 128 Independent Channel AWG with 15~16 bit accurate outputs of Independent Arbitrary Waveforms that could be precisely Bi-Phase modulated. We needed to monitor a number of different voltages simultaneously and used the various DMMs for such, and needed documentation & verification on some of those measurements. The eventual goal will be to roll this custom AWG into a full custom chip design.

Anyway, these references have been used to verify the various DMMs.

The last KS3465A was an accident as we had placed an order early this year and they were BO, so we found a DMM6500 in stock and ordered it and forgot to cancel the KS34465A. It showed up not long ago and we decided to keep it after realizing the very long lead time today.



Here's what the references look like now.  In front is the reference based around the LTZ1000 & LM399, this has 10V and Zener outputs, the one in back is based upon the LM399 and has 10, 1 and 0.1V outputs. Will try and find the schematic and PCB files later.

Edit: Attached are the files for both references, these from long ago and don't remember much, so check very carefully before using!! Altho don't recall any issues, we did add shunt EMI capacitors across outputs as noted on schematics and plan on inserting Ferrite Beads on leads. PCBs support many different type components to be flexible, so lots of choices. These normally operate from +24V, but can handle below/above this, the limiting factor is the +15 volt LM7815 regulator used in both.

Hope this helps.

Best,

[mawyatt]

If you are interested in Precision AC RMS Waveforms, check out the technique utilized here.

https://www.eevblog.com/forum/testgear/ac-rms-dmm-tests/

After replacing the Voltage Reference, and most of the chips, and still needing time to "age", just did a quick measurement on the KS34465A.

DC Ref voltage measures 5.0008645VDC

AC RMS voltage measures 2.5000857VACRMS

If you follow the technique it divides the Ref Voltage in half with a Squarewave. Of course the "one half" assumes an ideal squarewave (zero rise/fall with ideal 50% symmetry), and an infinite Band-Width DVM.

This technique has certainly proved good enough for our use, but as always YMMV!!

Best,