4

The problem

I am trying to measure quite precisely a voltage around 7.2V on a line of a PCB with a voltmeter. In this voltage range, my voltmeter only displays 3 digits. So the reading is something like 7.26V. However, I would like to have 4 digits of resolution.

Idea to solve the problem

To increase the resolution, I connected two voltmeters in series to have the voltage they each measure drop at half the range and benefit from the 4 digit resolutions. When doing so, the first voltmeter is stabilizing at 3.626V. The second one starts its reading, stabilize shortly at 3.626V but soon after restart its reading and repeat this sequence indefinitely.

The question

Is it possible to consider the total voltage to be twice 3.626V, so 7.252V? Is there a risk that the voltmeters are interfering each other in their reading, and I cannot use this result?

Schematic

How I connected the voltmeter on my circuit.

Schematic of the connection of the voltmeters

Tulmandil
  • 51
  • 4
  • 1
    No, this is not going to work. Buy a better voltmeter. – Chris Stratton Oct 27 '20 at 17:48
  • I agree. By the way, for which purpose would you need four digits of precision? Your BT1 and your R1, how closely do you control their temperature? – Marcus Müller Oct 27 '20 at 18:36
  • A better approach to combining two meters is to average the results. This way you reduce error and gain precision. – user1850479 Oct 27 '20 at 21:09
  • 2
    @user1850479 but if quantization is rough enough, you're not actually gaining anything unless both show something different. This works if Tulmandil could dither - but for that they'd need some source of noise... – Marcus Müller Oct 27 '20 at 22:05
  • @MarcusMüller I would like to have 4 digits of precision to be able to better compare this 7.2V with another one at about 4.8V. So my measurements are more a relative measurement than an absolute one. If both measurements are off by the same amount our % amount in the same direction, I'm quite ok with it. However, repeatability is not a parameter I have found for my voltmeters. – Tulmandil Oct 28 '20 at 09:02
  • 1
    @Tulmandil "Repeatability not found" + "only 3 digits" + need exact relative measurements: get one, better, voltmeter and maybe switch its input between the two measurement sources with an external relay. Two measurement devices can have different systematic error – one and the same will have the same error. – Marcus Müller Oct 28 '20 at 09:51

1 Answers1

2

schematic

simulate this circuit – Schematic created using CircuitLab

Figure 1. Two digital meters with nominal 10 M input impedance but second one is 10% below nominal.

Since your meters are in series the same current flows through both. The impedances of the meters will differ slightly (although yours are very close) and the voltage read by each will be proportional to their resistances. In my illustration I have assumed that the meters are the 199.9 type which have a full-scale reading of 199.9 mV.

Is it possible to consider the total voltage to be twice 3.626V, so 7.252V? Is there a risk that the voltmeters are interfering each other in their reading, and I cannot use this result?

You would need to understand the inner workings of the meters a bit more. They shouldn't interfere but if the second meter is switching its auto-ranging resistors in and out then it might. If you can turn off auto-range it might help.

Be aware of your meters' capabilities. Look up their specifications to work out the % error (of full scale - not your reading) and how many digits the reading can be out by.

Transistor
  • 168,990
  • 12
  • 186
  • 385
  • I disabled auto-ranging, and strangely it was not helping. With a third multimeter, I was able to pinpoint that it is the first multimeter that causes the change in the reading of the second. I'll check if the cumulative error of my 2 voltmeters make this connection useful or not. – Tulmandil Oct 27 '20 at 18:08
  • 2
    Um, but you're still now having twice the measurement error to account for this. I doubt you're actually gaining any meaningful amount of digits - maybe you get half a digit of additional resolution, but how's that going to help you? – Marcus Müller Oct 27 '20 at 18:35
  • 2
    @MarcusMüller, I agree. I don't think it's a good measurement technique. – Transistor Oct 27 '20 at 18:37
  • I computed the cumulated error. With only 1 voltmeter at 7.3V, I have about 50mV of error. With two voltmeters at 3.7V, I have 40mV of total error. So I gain a bit of theoretical accuracy, but my 1mV resolution is quite irrelevant compared to the global error. – Tulmandil Oct 27 '20 at 18:47
  • 1
    Your absolute voltage measurement may not be precise but the extra digit may be useful in relative measurements of voltages within 10mV of each other. – td127 Oct 27 '20 at 21:29
  • @td127 I try to compare two measurements, so even if the absolute error is quite high, is the repeatability of the voltmeter good? This is not a parameter I have found for my voltmeter. That would allow me to use my measurements despite the absolute error. – Tulmandil Oct 28 '20 at 08:58
  • 1
    I would think even if the meter isn’t absolutely correct and/or has some small nonlinearity error it would at least be monotonic: a larger voltage will give a larger reading. So I think your idea is valid, that two meters reading 4 digits gives you more resolution than one meter reading 3. Repeatablility should be good for measurements made with the same setup in the same timeframe. Maybe not trust measurements from yesterday, as temperatures and amount of coffee you’ve consumed can affect results. – td127 Oct 28 '20 at 16:48