The Museum of HP Calculators

HP Forum Archive 13

[ Return to Index | Top of Index ]

Looking for suggestion on calibration of a HP-3421A
Message #1 Posted by Renato on 7 July 2003, 3:54 p.m.

I got this nice HP-3421A. Both the SLA and Lithium 
batteries were bad. After replacing them, the newborn
requests breastfeeding - oops - calibration.
Does anyone have experience with that ? Service manual
suggests using a 6 digit multimeter and a voltage
reference. As i do not have neither of those, what do you
suggest (appart from packing it and sending it to you as a
gift) ? This one will go to a good home (or hospítal).
Re: Looking for suggestion on calibration of a HP-3421A
Message #2 Posted by Lawrence on 7 July 2003, 7:50 p.m.,
in response to message #1 by Renato

Renato, Agilent Technologies now handles the HP line of instruments. You can contact them at 1-800-403-0801. I had them check the calibration on my 3421A. The cost is $275 which includes shipping to and from Agilent. I received excellent service and documentation of the accuracy of my unit.

Please bear in mind that if the calibration is due to broken equipment (as opposed to loss of calibration memory due to bad batteries) Agilent may no longer have the obsolute parts.

Good luck.

Re: Looking for suggestion on calibration of a HP-3421A
Message #3 Posted by Renato on 8 July 2003, 10:05 a.m.,
in response to message #2 by Lawrence

Thank you for your reply. $275 is reasonable for a 
professional use unit. Mine is for a hobbist, and $275 would 
be too much. Also, in this case, calibration is not a
burden, but part of the fun.
Thanks, anyway.It is good to know that Agilent is still 
offering service to this old calculator related HP 
Re: Looking for suggestion on calibration of a HP-3421A
Message #4 Posted by James on 7 July 2003, 9:01 p.m.,
in response to message #1 by Renato

It depends on what level of accuracy you are after.

For starters, and if your needs are modest, for a few dollars you can use resistors to divide battery voltage to the voltages required for the reference. Depending on the batteries and resistors used, you might get down to a few percent accuracy.

For more accuracy, you will need access to a voltmeter of (a bit more than) your desired accuracy. You can then use a potentiometer to set the reference voltage for calibration to the accuracy of the voltmeter. You can probably borrow or rent a voltmeter.

If you want the maximum possible accuracy, it may be best to have a calibration facility (eg. HP, but there may be someone cheaper that is close to you) do it for you.

(I haven't gone through the 3421A calibration procedure, so there may be something else required...)

Good luck!

Re: Looking for suggestion on calibration of a HP-3421A
Message #5 Posted by Renato on 8 July 2003, 10:21 a.m.,
in response to message #4 by James

I took a quick look at 3421A Service manual. It seems that
three itens are required:
1. A HP-IL controller (Hp-85,HP-41) with proper software
2. A multi-scale voltage reference
3. A 6 digit multimeter

I'm doing this for fun. Part of the job is choosing the best precision for a hobbist budget. Current candidates for each required component are: 1. HP-41CV with 3421a control module 2. Battery in a temperature controlled situation (Luiz suggestion). Another possibility is special designed eletronics 3. Another HP-3421A (old calibration, unchecked)

Exchanging ideas and discussing possibilities is also part of the fun.

Looking for reference labs around here is a good idea.

Thanks, Renato

Re: Looking for suggestion on calibration of a HP-3421A
Message #6 Posted by Christoph Klug on 9 July 2003, 2:42 a.m.,
in response to message #5 by Renato

Dear Renato,

may be you find a low price kit for creating a voltage referece from your local electronic shop. You only need a precision reference voltage IC and some precision resistances.

For reference voltmeter you can take a second HP3421 or the HP3468 multimeter.

Actual I have tow HP3421 available, and one needs calibration - like your system :-)

Therefore your results are interesting for my own calibration procedure. Like you I will use HP41 plus DAC module...

Best wishes from Germany - Christoph Klug

Re: Looking for suggestion on calibration of a HP-3421A
Message #7 Posted by James on 9 July 2003, 3:49 a.m.,
in response to message #5 by Renato

1) HP-IL controller. Yes you need one (or HP-IB) to calibrate the 3421A. Of course you would need one to use the 3421A anyway, so this is not a great burden. However, you don't need the 3421A specific module. I suspect that any HP-IL interface module (41, 71, 75, 85) will be sufficient, along with the usual ability to send commands.

2) Voltage reference. Since the calibration can be done fairly quickly (I hope), the reference does not need good long or medium term stability. I suspect it does need to be clean DC (no AC ripple), so a battery is almost certainly the best choice for a hobbiest. Since the reference would normally be set/checked with the reference voltmeter (3) the accuracy and precision of the voltage reference is likely irrelevant, so long as it is stable over the course of a few minutes.

3) Reference Voltmeter. Since this is what the 3421A is calibrated against, it will be the limiting factor for the accuracy of the 3421A. A $50 3.5 digit DMM is likely good enough for a hobbiest application, but if a 6 digit meter is available, so much the better.

If no reference voltmeter is available, then we do have an interesting intellectual problem...

a) if relative-only measurements are all that is desired, then I think one could make do with a poor reference (eg. new alkaline batteries and 1% resistors). In this case your 3421A could tell the difference between X volts and X + 0.01 volts, but you wouldn't be sure if X was 0.900 or 1.100. Also, X would jump when the 3421A switched ranges. This would be confusing, and perhaps rather ugly.

b) Christoph's suggestion of using a precision reference IC is a good idea if absolute readings are required. This might give 0.1% calibration accuracy. However, I fear the task of building a reference voltage box around the IC is not practical for someone who does not have a voltmeter (due to the variety of voltages required, and the difficulty of maintaining the precision through to the output). Perhaps Christoph will build such a box to add to his product line... CK could be the new HP!

Other ideas?

I thought about something but I wonder if it works
Message #8 Posted by Vieira, Luiz C. (Brazil) on 9 July 2003, 11:01 a.m.,
in response to message #7 by James

Hi, folks;

I had something in mind that I proposed to Renato through an e-mail. It would involve a voltage reference and I'll change it a bit in this post. Please, if there is any lack of reasoning, I'd be glad to know. I do not want this to be a pattern, I want to know if it works. I did not test the procedure, it's a first proposal. Analog-related guys (Norm? Ellis? Others?), support me, please!

Let's take a new, high-capacity fully charged NiCad or NiMH battery and make sure it is in a 0ºC environment, what can be granted by putting the battery inside an insulated plastic bag immersed in melting ice. With the use of a thermometer, the temperature inside the bag may be confirmed, and tracking it's voltage till it does not change for about half an hour will make sure it is stable. Under these circumstances, this battery is measured with the use of a precision instrument and reading value is written down. Then this same battery is sent to someone who wants to calibrate another instrument. The "receiver" will charge it again (let's consider days between sending and delivering) and measure it under the same circumstances: 0ºC (melting ice) environment, so the reading value is adjusted to match the one read with first precision instrument.

Three facts must be taken into account to avoid battery voltage drifting:

- atmospheric pressure; if the sender and receiver locations have considerable highness difference, then it must be checked if battery voltage will vary as well; I think three-digit precision instruments would not be affected, but I'm not sure if five- or six-digit instruments will not.
- recharging method; electronic "smart" chargers of the same type and brand would be strictly necessary in both places OR sending the recharging equipment with the battery so the receiver will charge it in the same way;
- maximum voltage drop after being recharged; I think the first charges will not cause maximum voltage to drop, but after many recharging cycles, battery voltage must be checked again in a precision instrument. If the battery has already been charged many times, then fully charging it, measuring under these circumstances and sending it for a one-shot calibration must be safe as for reference, but if it is recharged again will probably give a lower voltage in its terminals.

I'd like to be able to test these procedures at home. Even if the 3421A is not calibrated it will serve to confirm my proposal: fully charge a NiCad or NiMH, measure its voltage under these circumstances, wait for a few days to simulate mail postage (average in local mail is about two days, and long-distance posting usually goes for two weeks), fully charge it again and measure its voltage. In this case, if there is a difference when reading with the same 3421A unit, this is the difference that must be taken into account when sending the battery to someone else.

In my first proposal, I asked Renato to use a regular, fresh new alkaline battery if the mail time is small (two days), but I thought about a way to use rechargeable batteries and keep precicion, and I think the approaches I mention above may be of use. The 0ºC temperature reference is necessary to reduce the number of external "variables".

Any other ideas to increase precision?

Hope this is of any use.

Best regards.

Luiz (Brazil)

Edited: 9 July 2003, 11:07 a.m.

Re: I thought about something but I wonder if it works
Message #9 Posted by Dave Shaffer on 9 July 2003, 11:43 a.m.,
in response to message #8 by Vieira, Luiz C. (Brazil)

Why not combine Christoph's and Luiz's ideas:

Somebody build a little precision voltage circuit, based on "a precision reference voltage IC and some precision resistances." This could be (rechargable?) battery powered. WIth a little effort, the temperature coefficient of this unit could be determined. Then, it is mailed around to whoever needs the reference standard.

How accurate could such a unit be?

Re: I thought about something but I wonder if it works
Message #10 Posted by Ellis Easley on 9 July 2003, 6:11 p.m.,
in response to message #8 by Vieira, Luiz C. (Brazil)

I don't know if NiCads have a repeatable output voltage. Just from watching them charge and discharge, their voltage seems to vary quite a bit (in a fairly narrow range, like 1.2 - 1.5V). There is a standard cell that includes mercury for one of the electrodes, I'm not sure what the other one is, the output voltage is about 1.08V I think. I regret forgetting to bid on one on Ebay a couple of years ago. It went for a pretty low bid, <$20 I think. Like most precision instruments, it was probably only as good as its last calibration. My first meter was a Heathkit VTVM I got in high school, it came with a "D" cell (for the ohms range) that also served as a calibration standard: the meter face had a hand-marked spot on it that I was to use to calibrate the meter to the cell on a certain voltage range, after the unit was assembled (and presumably, before the cell was used for ohms measurements).

I have found that very fresh, unused alkaline cells usually measure 1.580 volts, + or - maybe .006 volts. If you believe me and my meters, you should be able to use this information to achieve approximately 0.4% accuracy.

I have a Ballantine RMS voltmeter that measures down to 1 mV full scale. It is basically a high gain AC tube amplifier followed by a ladder circuit with a bunch of crystal diodes that convert AC voltage to an approximation of the RMS equivalent DC voltage, then that voltage is measured by a DC voltmeter. It includes a built-in calibrater that generates a precise AC voltage, analogous to an IC voltage reference. It uses the negative-resistance characteristic of an incandescent lamp to regulate a sample of the AC line voltage, not unlike the way the HP200A oscillator (and its descendants) use that characteristic to regulate the output amplitude.

3421a ranges
Message #11 Posted by Renato on 9 July 2003, 8:07 p.m.,
in response to message #5 by Renato

This is a very interesting discussion.
From 3421a specs, these are the measurement ranges:
DCV: 0.3V, 3V, 30V, 300V
ACV: 3V, 30V
2/4wire ohm: 300,3K,30K,300K,3M,30M

From 3421a Service manual Recommended accuracy for Voltmeter: +/-.004%(30V,300V)m,+/-.005%(.3V,3V) For DC Voltage standard: Short term stability: better than .0002% For AC calibrator: Accuracy:.04%,stability +/-.02% Resistance standard: +/-.0005% for 100R,1KR,+/-.001 for 10KR,100KR,.01% for 1MR,10MR

Goal is to get as close as possible to those figures, while keeping the hobbist budget.

Re: Looking for suggestion on calibration of a HP-3421A
Message #12 Posted by James on 10 July 2003, 4:35 a.m.,
in response to message #11 by Renato

From the 3421A service manual, it seems the desired calibration inputs are:

0.3, 3, 30, and 300 VDC

3 VAC rms @ 100 Hz

300, 3K, 30K, 300K, 3M, and 30M Ohms, or 100, 1K, 10K, 100K, 1M, and 10M Ohms (it appears that either the 1x or 3x value from each decade can be used)

There is also temperature calibration using a thermistor.

Presumably the 0.3 and 3 VDC can be generated with batteries and a simple voltage divider (perhaps preset by someone with a high accuracy unit). Perhaps after calibrating these ranges, the 3421A itself can be used (via a 100x divider) to adjust a 30 / 300 VDC source made from rectifying and filtering the AC line (dangerous... use appropriate care, do not try this without supervision, etc.), which can then be used to calibrate the upper ranges.

Generating the 100 Hz AC signal might be easier if it were done as a square wave, but the voltage would need to be adjusted to compensate for the waveshape if the 3421A is not "true RMS". Or, perhaps we can discover if the frequency is not a significant factor, allowing 50/60 Hz to be used. However, the accuracy of the power line sinusoid may be a big problem in some areas/countries, so making a generator independent of the power line waveform seems a better choice. Regardless, it seems that something more than a battery and a handful of resistors is needed to do the AC calibration.

The resistance standards seem to be the easiest; DigiKey having 1% resistors (eg. 311-100HCT-ND) around $1 for 10 pieces of one value, with some 0.1% types available (eg. RR12P100BCT-ND) at $7 for 10 pieces of one value. So you might get 10 each of 100, 10K, and 1M; put them in series to make 1K, 100K, and 10M, and pay only $3 for 1%, or $20 for 0.1%.

[ Return to Index | Top of Index ]

Go back to the main exhibit hall