Questions regarding the calibration of a HP 3478A 5 1/2 digit DMM

[EE54]

So I have two questions.
I have a HP 3478A that I picked up locally for 3700000 VND(~160USD). It was built in 09/10/96(DD/MM/YY) according to the date code on the back. I have no idea when it was lasted calibrated so i want to take it to this local cal lab and get it calibrated at least once https://kimlongcom.com(you can change the language to English on the top of the website). I asked for a quote and they responded with a price of 1870000 VND(~81USD) and they said it would take 2 to 3 days due to how many functions my DMM have.
So my 1st question is: is 81USD a reasonable price of a full calibration of my HP3478A?
I am aware that the cal data is stored on a volatile memory powered by a lithium battery. My battery is the Panasonic BR 2/3A, with the number 1722 printed on its side(if that means anything). It measures 3.393V.
So my 2nd question is: do I need to replace the battery, as to not risk losing the precious calibration data?
Any help would be very appreciated.

[niner_007]

How old is the battery? If more than 7-10 years I’d say replace it first. $80 seems cheap for 7.5, but I’m not familiar with the accuracy requirements of 3478A, nor the accuracy specs of the cal lab, just check the datasheets, and make sure the cal lab and their cal equipment is suitable and they aren’t just burning your money for nothing.

[niner_007]

Sorry, though DMM was 7.5, no idea of cal pricing for 5.5

[bd139]

$81 is about same as what I got quoted for mine here so is reasonable.

Make sure you replace the battery first. Note that if you disconnect the battery, cal data is gone. You have to stuff a floating power supply across the RAM in the meter and replace the battery. Make sure your soldering iron doesn't short anything to ground. That happened to mine and I lost all the cal data  . Have a look around for other threads - lots of people have done the battery replacement on here. I am clearly shit at that so don't take my advice on how to do it 

[maxwell3e10]

I recently found this data about Panasonic BR 2/3A battery. If the battery measures close to 3.4V I think the concern about it suddenly failing is overblown. It actually decays quite slowly.


In fact, recent datasheets for BR-2/3A suggest that starting voltage of newer batteries is lower than it was in 1990s. Does anyone has a new one to measure?

[EE54]

How old is the battery? If more than 7-10 years I’d say replace it first. $80 seems cheap for 7.5, but I’m not familiar with the accuracy requirements of 3478A, nor the accuracy specs of the cal lab, just check the datasheets, and make sure the cal lab and their cal equipment is suitable and they aren’t just burning your money for nothing.

The cal lab I'm planning to go to claims to meet ISO/IEC 17025:2005 standard. This is some of the specifications that they stated in their cal lab certificate. Do you think it would be sufficient to calibrate my DMM?
Also is there anything else I should ask the cal lab before sending my DMM in for calibration? I talked to them over the phone before and they stated that they would provide a calibration report showing how much my DMM is off compare to their standard, when it arrived and after adjustment.

[Kleinstein]

Unless you really need a calibrated meter right now, I would first use / test the meter for a few weeks or months, to see if there is a hidden fault. With used meter there is certain chance there could be some problems and these may not be so obvious.

When sending the meter in one may have to think about whether one wants just the calibration or also an adjustment (reset the scale factors), even if still in spec. This may cost extra, but could be free as the adjustment is through software.

[EE54]

Unless you really need a calibrated meter right now, I would first use / test the meter for a few weeks or months, to see if there is a hidden fault. With used meter there is certain chance there could be some problems and these may not be so obvious.

When sending the meter in one may have to think about whether one wants just the calibration or also an adjustment (reset the scale factors), even if still in spec. This may cost extra, but could be free as the adjustment is through software.

I bought the meter closed to a year ago and been using it on and off during that time. It seems to work perfectly fine as far as I can tell(measure thermocouple voltages, sub 1 Ohm resistors using the 2/4 wire ohm mode), is there any particular problems i need to look out for(beside the lithium battery which I would probably replace)?

[Kleinstein]

I don't know of a specific fault that come up with the 3478. So it would be usual tests that can be done without special instruments:
1) check the noise when measuring a short (3 V and 30 mV ranges)
2) check for input bias current in 3 V range  (1-10  M resistor to ground or drift rate with a 1 - 10 nF PP or PS capacitor)
3) Ohms stability with a few good resistors especially at some 1 M or so.

1 year of use should have done quite some tests.

Is the mains RF filter still the old original Schaffer one ? These have a tendency to blow with nasty smoke. Changing the filter would not effect the calibration - but the smoke from a blown filter could.

[bd139]

3478a doesn’t have a Schaffner mains filter. It does have some RIFAs though!

[EE54]

I don't know of a specific fault that come up with the 3478. So it would be usual tests that can be done without special instruments:
1) check the noise when measuring a short (3 V and 30 mV ranges)
2) check for input bias current in 3 V range  (1-10  M resistor to ground or drift rate with a 1 - 10 nF PP or PS capacitor)
3) Ohms stability with a few good resistors especially at some 1 M or so.

1 year of use should have done quite some tests.

Is the mains RF filter still the old original Schaffer one ? These have a tendency to blow with nasty smoke. Changing the filter would not effect the calibration - but the smoke from a blown filter could.

1) I need to short all 4 terminals together(the 2 HI, LOW inputs as wells as the two 4 wire resistance sense terminals), right?
2) Can you elaborate? Do I need to connect the 1 Meg resistor between the Hi input to Low input? If not ten how exactly do I need to connect the 1 Meg resistor?
3) I'm planning to buy a bunch of 1% wirewound and metal film resistors from Digikey as well as a 3V lithium battery that looks exactly like the one inside my DMM. The resistor were chosen to push the DMM as close to full scales as possible without causing it to switch to the next range(1% tolerance accounted for) while having fairly good TCR(because the temperature of my room probably isn't very stable) as well as not costing to much(the whole order came out to 11.95USD). Will these resistors do for this test?
 - 1 Panasonic BR-2/3AE2SP 3V lithium battery https://www.digikey.com/product-detail/en/BR-2%2f3AE2SP/P270-ND/64350/?
    itemSeq=307722140
 - 25 Ohm resistor(±20ppm/°C) https://www.digikey.com/product-detail/en/43F25RE/43F25RE-ND/823232/?itemSeq=307728035
 - 250 Ohm resistor(±20ppm/°C) https://www.digikey.com/product-detail/en/WW3FT250R/WW3FT250RCT-ND/2075183/?
    itemSeq=307728195
 - 2,5k resistor(±30ppm/°C) https://www.digikey.com/product-detail/en/ALSR052K500FE12/ALSR5F-2.5K-ND/257352/?
    itemSeq=307728225
 - 25k resistor(±20ppm/°C) https://www.digikey.com/product-detail/en/WHE25KFET/WHE25KFETCT-ND/3114495/?
    itemSeq=307728257
 - 270k resistor(±50ppm/°C) https://www.digikey.com/product-detail/en/MRS25000C2703FCT00/BC4098CT-ND/7351528/?
    itemSeq=307728529
 - 2.7M resistor(±50ppm/°C) https://www.digikey.com/product-detail/en/MBB02070C2704FCT00/BC4459CT-ND/7351889/?
    itemSeq=307728622
 - 10M resistor(±50ppm/°C) https://www.digikey.com/product-detail/en/MRS25000C1005FCT00/BC3961CT-ND/7351391/?
    itemSeq=307728749

[Kleinstein]

For testing the noise in voltage mode, one just needs short for the 2 voltage terminals. Just a wire is preferred over a classic cable with 4 mm plugs as it caused less thermal EMF.

For the bias check there are two options. This again uses the 2 voltage terminals, with the resistor or capacitor between
A) use a resistor of some 1-10 M and measure the DC voltage in the 30 mV range.  This checks the bias at 0 V.
B) use a low loss (PP or PS type) of some 1-5 nF. The drift rate gives the current as I = C * dU/dt.  This usually needs to send the data to a PC.  The capacitor can be charged to different starting points an this way allows to measure the bias current at other voltages too.
There may be some internal capacitance in the DMM. So with a small cap one may have to add this. With a large cap DA can cause some background.

Normally the current should be less than some 100 pA, ideally more like < 50 pA.

The 2.7 M / 10 M resistors would be definitely good enough to test the bias current.  They are probably also Ok for a stability test to some degree, if the room temperature does not change very much, or the temperature is also read.

[EE54]

For testing the noise in voltage mode, one just needs short for the 2 voltage terminals. Just a wire is preferred over a classic cable with 4 mm plugs as it caused less thermal EMF.

For the bias check there are two options. This again uses the 2 voltage terminals, with the resistor or capacitor between
A) use a resistor of some 1-10 M and measure the DC voltage in the 30 mV range.  This checks the bias at 0 V.
B) use a low loss (PP or PS type) of some 1-5 nF. The drift rate gives the current as I = C * dU/dt.  This usually needs to send the data to a PC.  The capacitor can be charged to different starting points an this way allows to measure the bias current at other voltages too.
There may be some internal capacitance in the DMM. So with a small cap one may have to add this. With a large cap DA can cause some background.

Normally the current should be less than some 100 pA, ideally more like < 50 pA.

The 2.7 M / 10 M resistors would be definitely good enough to test the bias current.  They are probably also Ok for a stability test to some degree, if the room temperature does not change very much, or the temperature is also read.

Will these PP capacitors do?
 - 1000pF(1%):https://www.digikey.com/product-detail/en/PFR5102F100J11L4BULK/399-7694-ND/3465862/?itemSeq=307879445
 - 6800pF(1%):https://www.digikey.com/product-detail/en/FKP2J016801J00ESSD/1928-1281-ND/9370229/?itemSeq=307879582

[Kleinstein]

The capacitors should be OK for the test. There is no need to have accurate capacitance values though (+-10% is sufficient): it is more about the order of magnitude of the input current, not the exact value.

[tggzzz]

When sending the meter in one may have to think about whether one wants just the calibration or also an adjustment (reset the scale factors), even if still in spec.

Understand that difference, and be sure you understand what any cal lab will and won't be doing.

If you do lose the cal constants in the RAM, you will need adjustment.

I successfully replaced the battery in my 3468. Provided you understand what you need to do and avoid, and have practical skills, it isn't difficult. Personally I would prefer not to do that after sending it off to a lab.

[EE54]

When sending the meter in one may have to think about whether one wants just the calibration or also an adjustment (reset the scale factors), even if still in spec.

Understand that difference, and be sure you understand what any cal lab will and won't be doing.

If you do lose the cal constants in the RAM, you will need adjustment.

I successfully replaced the battery in my 3468. Provided you understand what you need to do and avoid, and have practical skills, it isn't difficult. Personally I would prefer not to do that after sending it off to a lab.

Yes, I am planning to replace the battery before calibrating it.
And if I understand correctly, a calibration means the cal lab will compare my DMM against their standards and tell me in the report how much my meter is off compare to their standard and whether my meter is still in its 1 year specs or not. And it does not mean the cal lab will adjust the scale factor of my DMM even if it is out off its 1 year spec. Is that correct?

[iMo]

Yes, "calibration" and "adjusting" are two things for the labs. The "adjustment" of a meter is usually more expensive as well (and there are labs which do not provide the adjustment service at all).

[EE54]

Thanks everyone for your help(especially all the useful advises Kleinsteinfrom . I ended up choosing 4 Polypropylene (PP) film capacitors from the Wima FKP2 and FKP3 series capacitors for the input bias current test. Chose 4 different values just so I can have some variety, aslo because the datasheet of these capacitors claims a very high insulating resistance of 300 GOhm or more.
 - 1000pF: https://www.digikey.com/product-detail/en/FKP2F011001D00JSSD/1928-1256-ND/9370204/?itemSeq=307923801
 - 3300pF: https://www.digikey.com/product-detail/en/FKP3G013303D00JSSD/1928-1315-ND/9370259/?itemSeq=307924128
 - 6800pF: https://www.digikey.com/product-detail/en/FKP2F016801I00JSSD/1928-1261-ND/9370209/?itemSeq=307924013
 - 10000pF: https://www.digikey.com/product-detail/en/FKP3G021004B00JSSD/1928-1317-ND/9370261/?itemSeq=307924068
I'll let everyone know when the parts arrive. Probably going to be a while since I'm trying to have a local part store(who also buys parts from Digikey) to buy my stuff for me. This way I hope to reduce the shipping cost by a bit since shipping to Vietnam is so damn expensive(25 USD for a 14.35 USD order!). It that doesn't work, i'll bite the dust and buy directly from Digikey.

[Conrad Hoffman]

FWIW, I've had a 3478A for many years. I check the cal every now and then and it hasn't moved enough to worry about in a decade. Nor have I touched the battery. I'd just use the thing. If you happen to meet somebody with a 6.5 digit meter of recent cal, just check it against that.

[GigaJoe]

had about 7 of them, very stable, had purchased battery from ebay, times cheaper comparing to unhuman digikey. most my units with unknown cal. time. those who where first 198x-year,  had a large  shift. last one almost dead on ... .  it have a few variations of board, i think 7 revisions,  where first was 2 colors HP logo and case assembled in 2 pieces,  last version, just black logo, one piece of case, and no coils inside for a signal transfer. LCD also has different fonts shape, that one was dead on ... - maybe cal. was 10y ago, and it was a reason of precise measurement.

power filter caps usually became bad,  an indirect indicator: short on 30mv produced 10-30 microvolt noise.  with a new one: 0.2-0.4mkV
with 110V i don't bother even for main filer. When it calibrated drifting i noticed , maybe 2 last digit on voltage,  but much more for resistance (10-15 digits) in 3 years period.

[EE54]

FWIW, I've had a 3478A for many years. I check the cal every now and then and it hasn't moved enough to worry about in a decade. Nor have I touched the battery. I'd just use the thing. If you happen to meet somebody with a 6.5 digit meter of recent cal, just check it against that.

I don't think that's possible. The only guy I know who has 6 1/2 digit meter is the guy who sold me the HP meter in the first place. And he lives pretty far away. And considering his occupation appears to be buying and selling T&M equipment and he seems pretty busy, I don't think he would appreciate an annoying kid wanting to use his 6 1/2 digit meter. So the plan for now is to do some other project while I wait until my local part supplier order from Digikey again. Then I can ask them to include my order into their order. This way I can reduce shipping cost(hopefully). After that I'll perform all the basic test as suggested by Kleinstein and if all goes well, I'll have the meter calibrated by my local cal lab. I want to get my meter calibrated(not immediately but hopping at some point) because I want to use it to do some precision stuff in the future(measure the performance of a 16 bit ADC, maybe even my own 16 DAC...). Another reason is I think I'm slowly becoming a voltnut. I bought the meter the moment I found out someone who live in my city is selling it, I also have 4 LM399 I haven't touch(not sure if I want to).Hopefully by having a nice calibrated 5 1/2 DMM, I can convince myself I don't need to pursue higher accuracy(for now) and hopefully keep myself away from the rabbit hole that is voltnutting and work on something else.

[Kleinstein]

The main point that is checked in the normal calibration of a meter like the 3457 is the accuracy / scale factor for the different ranges. The main point's that are a little prone to drift is the input divider for the 30 V and 300 V ranges and the current sources for the ohms ranges. The calibration would usually not check things like linearity very much (they should check things like +3 and -3 V though). So for things like checking an ADC / DAC, there is no need to have the meter calibrated.

When the meter was bought from someone with a good other meter, there is a chance the calibration was at least checked to some degree. This is not a official calibration and maybe to a lower standard. Still one can have some confidence even in the old meter.