Fluke 45: induces voltage while measuring

[Deckert]

Hi guys,

Need a bit of help here with some Fluke 45 weirdness. I bought a Fluke 45 2nd hand and my initial testing showed no issue with it. All ranges measured fine and, indeed, it is still very accurate compared to my recently calibrated Brymen 867.

However, after using it on the bench for a while (and specifically measuring voltage to the gate of a FET), I noticed that the Fluke 45 was actually introducing fairly significant voltages into the circuit. To test it, I hooked the Fluke up to my Philips PM2525 bench meter and set the Philips to fast acquisition mode.

It soon became very clear that, when the Fluke 45 was switching between low ranges, it would output a spike voltage of up to 500mV! It only does this when switching between the 300mV scale and the 3V scale. Selecting a higher scale (e.g. 30V and up), you hear a relay click and all this phantom voltage generation stops happening.

It also looks like this output of voltage is somehow capacitive (i.e. it has a discharge curve and eventually settles down to 0.00mV reading). But that's likely just due to the 20Mohm input resistance of the PM2525. If I don't add that 20Mohm "load" the Fluke 45 takes a lot longer to settle down to zero.

When just measuring a normal voltage source from a PSU set to, say, 150mV, the Fluke reads that fine without any issue. Shorting the input to the Fluke 45 immediately brings it to zero as well.

Here's pic I took immediately after manually switching from the 3V scale to the 300mV scale. This was in manual ranging mode, but the meter does the same thing when it is auto-ranging down on its own. All other scales work perfectly (current, frequency, resistance, etc). It only does this in DC range (when switched to AC, there is no voltage spike being output).



Looking at the front-end, this has to be coming from the components before the analogue processor, correct? Would appreciate if anybody has come across something similar or can offer any insights.



--deckert

[kcbrown]

Unless I'm missing something (a distinct possibility -- I'm quite new to this), the only two capacitors that could possibly put a voltage onto the voltage sense terminal are C1 and C48.  Anything else is buried in the analog processor.

Is that your take on this as well?   It looks like C1 gets switched in only when you're in AC mode, and otherwise is left floating, so I don't see how it could be responsible for what you're seeing here.  That leaves C48 (the 100pf capacitor between pins 23 and 25 on the analog processor).  But those are the only things I can see in the circuit that might qualify.  Anything else will be a lot more subtle than what these neophyte eyes can see.

Might C48 be bad?

Very strange.

It would be interesting to know if other Fluke 45 units behave this way.

[Deckert]

Hi,

Agreed, it seems the only caps that can affect this are C1 and C48. And C1 does not seem to be the culprit. C48 is only 100pF - hmm, will need to calculate the RC time to see if C48 is holding the charge.

- Can anybody confirm whether this behaviour is generic in the Fluke 45 (I somehow doubt it).

At this stage I'm holding thumbs that the voltage being induced is coming from somewhere externally, but it's looking more likely that it may be coming from the analog processor chip 

I guess the crowd-sourced differential needs to be augmented with some actual inspection - i.e. I will have to open up the meter and poke around a bit.

--deckert

[Alex Nikitin]

Most likely it is just a charge injection from the internal switches in the analog processor, not a "voltage" as such, but charging the input filter capacitor when switching. 100pF x 0.25V = 25pC, a somewhat larger value than usual in a good CMOS switch but possible. And it is dissipated slowly due to a very high input impedance of the meter at lower (300mV, 3V) ranges. 30V and higher have 10M input impedance and C48 is out of the circuit, so this small charge is dissipated quickly and you wouldn't notice it.

Cheers

Alex

[Deckert]

Hi,

That's just the thing - I do notice it. Since when is it acceptable for a multimeter to turn on a FET of the circuit you're measuring?

However, I do take your point about the charge injection - could well point to an analogue switch not turning off when it should. I'd still like to verify if this is a general issue with the '45 or only an issue with my unit.

--deckert

[2N3055]

Charge injection shouldn't be so large... I would open it and check for leakage, maybe from Ohms current source... There is a relay that switches it in and out...
Also you would need to check if board is clean and such...

[G0HZU]

I've got a very old and scruffy Fluke 45 here and if I toggle up/dwn through the DC ranges (after power on) with no probes on the input the biggest DC offset I see on the display is about 0.4mV. It shows this when going down into the 3V range (displays 0.0003V) and also it is there on the next range down (the 300mV range) on the default M speed setting. It takes a second or two to disappear on the 300mV range.

However, I'd expect that there will be some fast transients that the meter can't capture. I've not looked using a scope with a high impedance scope probe. Your Fluke meter seems to be able to capture/display 259mV which is very different to mine. I see 0.4mV DC typical in the display with very little spread despite numerous range changes. The worst I can see is about 1mV if I go up/dwn very quickly.

Note: If I put some long probes on it then it does show 200mV when going down into the 3V range and it is slow to decay away unless I actually probe something.

[sorenkir]

I have plugged the input of my Fluke 45 to my scope.
Attached below the screen captures when toggling between some ranges.
Same phenomenon on Solartron 7065 & HP 34401A.
Nothing on a Fluke 8842A and Keithley 199.
Michel.

[Deckert]

Thank you for the tests guys.

@sorenkir: I note your spikes are pretty darn short. I hooked mine up to the 'scope, but note the voltage levels as well as the long timebase. You can see the decay times on mine is a lot longer. The scope picks up even higher spikes then I did with the second multimeter. This is on10X probe (so 10Mohm input resistance). The 'scope is floating - it's on an isolation transformer, so nothing on the Fluke 45 is being coupled to ground.



--deckert

[Deckert]

Note: If I put some long probes on it then it does show 200mV when going down into the 3V range and it is slow to decay away unless I actually probe something.

G0HZU, What happens if you measure the output on those probes with a different multimeter while cycling between 3V and 300mV range? Do you read any voltages?

--deckert

[seronday]

I have access to 2 Fluke 45 meters at present for a short time.
Both of these meters exibit the same effect when changing ranges on DC volts.

It seems that the actual transient voltage level that appears on the Fluke 45 meter leads when changing ranges, is somehow related to the level of stray AC volts being picked up by the leads.
When the Fluke is on the 30V DC range this can be seen on an oscilloscope connected to the meter leads with a 10X probe.

See the attached DSO screen prints.

All of these show the range switching sequence as follows:-
Start on 30V DC ,to 3V DC ,to  300mV DC ,to  3V DC , to 300mV DC

Two of the screen prints are with the DSO 10X probe connected to the meter lead probe tips.
The difference is with the Fluke input leads in a different position on the bench.
The other is with the meter leads removed and the DSO 10X probe connected directly to the Fluke 45 input terminals.
This one shows the lowest voltage levels.

[sorenkir]

I have tried with Fluke45 directly connected to a fully isolated Fluke Scopemeter 123: the pulse is there and even shorter!
Michel.

[Deckert]

Wow, seronday, that's some pretty interesting results you have there. Matches mine fairly closely. I can confirm the stray pickup on the 30V scale and also substantially reduced effect when probing directly on the multimeter input.

I guess that proves, fairly conclusively, that it is not an issue with my specific unit, but rather a more general behaviour of the Fluke 45. It's as if the meter is amplifying stray voltage pickup somehow and it's affecting (possibly) the charge-pump of the dual-slope convertor?

I took the meter out of my lab into a backroom where there is almost no electrical noise close by and indeed, it does seem to zero much better and there is almost no voltage being output by the meter (as measured on my Brymen 867). I wonder what exactly the mechanism is behind this behaviour.

Will just have to live with it.

--deckert

[lemon]

Very strange!

I have two Fluke45 and I looked about it, but I couldn't find same behavior.
At the worse case, I see almost or <1mV when the relay click is appeared.