Fluke 8050A calibration

[bob91343]

I have one of these that seems accurate except on the 20V dc range.  As it happens, there seems to be no adjustment for that.

I have searched the diagram and there is no pot for the 20V range.  The calibration procedure skips that range.

Not that it's terribly far off, maybe around 0.2% at most, but I'd like to trim it.

Does anyone know about this?

[med6753]

First make sure your supply voltages are in spec. There is interaction between R11 (2V cal) and R5 (200V cal) and you may have to adjust for best compromise. Or you can play with the U5 jumpers on table 4-10 on page 4-11 of the manual.

Above all, how accurate is your source voltages?

[bob91343]

Allthe voltage ranges appear to be in spec except the 20 V range.

However I will open it again and measure the voltages.  There do not seem to be any precise voltages, at least from my schematic perusal.  And no adjustments.

Can you explain what this is all about?  If I can't adjust the power supply I don't know how to correct the error if any.  And I will have to look for the tolerances on the supply voltages.

[med6753]

The power supply voltages should be within +/- 10%. If not, investigate the cause.

By "source voltages" I am referring to the standards you are using to calibrate it. What is it? Is it a traceable standard? If not then your source voltages are suspect. 

If the 2V range and the 200V range are properly set then the 20V range will automatically fall into place. If not then there is the possibly, although unlikely, that one of the precision resistor networks (U1 - U5) has drifted.

[Fungus]

I have searched the diagram and there is no pot for the 20V range.  The calibration procedure skips that range.

Not that it's terribly far off, maybe around 0.2% at most, but I'd like to trim it.

Does anyone know about this?

You need to calibrate it on the the 200V range, the 20V is derived from that with precision resistors.

Feed it a voltage while it's on the 20V range and twiddle R5. The reading on screen should change.


For me the 20V range is the most useful so I'd "Lego" two 9V batteries together and use the 18V output to match it against my best meter while the 8050A is on the 20V range... 

[NoisyBoy]

I agree with med6753 that having a calibration standard with wide enough of a calibration voltage range would be a good start.  The 8050a DC calibration step 3 and 4 requires 190V and 1000V DC source, a stable calibration grade source at these voltages are not common in most labs.  Low end bench supplies are simply not precise or stable enough for most calibration work.

It would be beneficial to share with others on how your unit was calibrated, and with what equipment.  When you said other ranges other than the 20V were within spec, what equipment did you use to verify them, and what are these test points. 

If I were you, I would do the proper calibration first, then test each range at its respective test point, before trying to troubleshoot a specific range.

Good luck. 

[bob91343]

I am verifying the unit with an HP 3456A which is almost overkill for accuracy.  I like the idea of miscalibrating the 200V range to bring the 20V range in; most of my measurements are on the 20V range.  I will also look into a possible resistor out of tolerance to make the two ranges exactly 10 to 1.

First I have to clean up the workbench; there is a spectrum analyzer all apart that has an intermittent.

[NoisyBoy]

That's exactly what I do for handheld DMM that has a non-linear reading and only 1-2 adjustment pots.  Adjust it based on the range you use most rather than the entire range.

[bob91343]

Can someone point me to the resistor divider that makes the 200V and 20V ranges in the proper 10:1 ratio?  If I find it I can trim the offending one and have what I want.

[Martian Tech]

You can find the schematics online.  Looks like it is not a voltage divider, but the gain of the buffer stage is either x1 or x10.  A 2.2K resistor in U34 (which appears to be a custom part with multiple precision resistors in a SIP package) - pins 4 & 5 - is switched into the feedback loop or not by the CMOS switches U19C/U19D based on a control signal from the CPU. And just to make things more confusing, the RMS converter has settable gain as well for AC signals.

[Fungus]

Can someone point me to the resistor divider that makes the 200V and 20V ranges in the proper 10:1 ratio?  If I find it I can trim the offending one and have what I want.

Look in the manual, it has all the schematics, etc:

http://manuals.chudov.com/Fluke-8050A.pdf

[bob91343]

It appears that the ac ranges do not suffer the problem, just the dc.  So maybe I would be better off leaving well enough alone.

A man with a watch knows what time it is.  A man with two watches is never sure.  A man with  20 or so voltmeters (me) scratches his head in confusion.

Anyway, while this unit has many nice features, it also has down sides.  No autorange and no readout illumination.  That's offset by the nice conductance ranges for very high resistances.  I think it also calculates dB and some other stuff.

[NoisyBoy]

I feel your pain with DMM alternative truth, it was bad.

I cured it by sending four bench meters back to Keysight each year for calibration.  I calibrate the less precise bench meters and handheld DMMs myself with a Krohn-Hite calibrator, now the pain is gone and every meter agrees with each other :-) 

You should at least get one of your meter through the NIST calibration so you know what the truth is.  Perhaps the 3456A is a good candidate for it.

[Fungus]

It appears that the ac ranges do not suffer the problem, just the dc.

It's a different circuit. 

So maybe I would be better off leaving well enough alone.

There's no way you can break anything*. Those things are designed for people to go in and twiddle manually - it's how they did it at the 8050A factory.

(and those people were working in a hurry...)

All you need is a stable voltage source that's somewhere near the top end of the range you're adjusting and a better meter for comparison. 

(*) Famous last words.

[bob91343]

matiantech has convinced me not to meddle in this thing with a soldering iron.  I will do the compromise tweak and stop agonizing.  (yeah)

I also see some drift in readings for the first few minutes.  I presume that's endemic to the unit.

[med6753]

As I've stated before you can play with the resistor values as stated in table 4-10. Then do a complete DCV re-cal and see where it's at. If it's unacceptable you can reverse it.

Some drift up to a half hour is normal for any 4.5 digit DMM.

[Martian Tech]

Assuming I'm reading the schematic correctly, here's what I don't understand: If it's accurate on the 200mV scale, it should be accurate on the 20V scale as well.  They both use the x10 buffer gain, and on the 20V and 200V scales, the input signal is divided by 100 by U1(1,2), R5, U1(3,4).  So on the 200mV and 20V scales, 0-200mV is presented to the buffer and multiplied by 10 to present 0-2V to the integrator.  On the 2V and 200V scales, 0-2V is presented to the buffer and multiplied by 1 to again present 0-2V to the integrator.

If the divide by 100 is inaccurate, both the 20V and 200V ranges should be off.  If the x10 multiplier is off, both the 200mV and 20V ranges should be off.

Could it be something as simple as oxidation on the switch (S7) contacts in the 20V range?

[bob91343]

Well I never checked the 200 mV or 200 V ranges so that's my next step.  It may be that a full adjustment will fix everything.

[med6753]

Well I never checked the 200 mV or 200 V ranges so that's my next step.  It may be that a full adjustment will fix everything.

Well, that would explain your issue and you should have stated right from the get go that you did not attempt to calibrate to other ranges. Follow the DCV calibration procedure as stated in the manual in the order given. But obviously you may not have access to an accurate 1000V source so that range can be left alone.

[Fungus]

But obviously you may not have access to an accurate 1000V source so that range can be left alone.

All you need is enough 9V batteries... 

[bob91343]

They are not adjustable.  Well the combination is adjustable in 9V increments.  To get 200 Volts one needs about 22 of them.  I think they cost around $3 so that makes the source a bit pricey.

I have planty of sources of voltage in that range anyway.  If there is one thing I have plenty of, it's power supplies.

[Fungus]

I think they cost around $3 so that makes the source a bit pricey.

Here's 40 of them for under $20, shipped from the USA:  https://www.aliexpress.com/item/32324041879.html

[CDaniel]

The calibration procedure was made to be standard , but if you know what are you doing any voltage close enough in that range  is good when you have another calibrated voltmeter .

[coromonadalix]

Maybe tons of 9v cells from a dollar store ??

[Fungus]

The calibration procedure was made to be standard , but if you know what are you doing any voltage close enough in that range  is good when you have another calibrated voltmeter .

Yep. If you have a second meter with better precision than the one you're calibrating then there's absolutely no need to generate a precision voltage, just a voltage somewhere near the top of the range being calibrated. The OP says he has a 6.5 digit HP 3456A, that's easily good enough to lick a Fluke 8050A into shape.

The only important thing is to calibrate somewhere near the top of the range being calibrated. In theory you can calibrate a 20V range with a 1.5V battery but I'd be much happier working with 18V from two 9V batteries. It will be much easier to do and the error will be 12x smaller (18/1.5=12).