How does a modern digital multimeter "sense" electricity?

[stitch]

In the early days of electronics, a galvanometer was used to sense electric current.  The galvanometer used magnets and a coil attached to a needle to do this.  Can anyone explain to me what part of a modern digital multimeter replaces the coil and magnets of the galvanometer? Asked another way, what electrical phenomena are exploited by modern DMM's to sense current?

[TimNJ]

I'm pretty sure it's as simple as ohm's law. I believe a DMM measures the voltage drop across a known resistance, also known as a current shunt.

[Wytnucls]

The heart of the modern DMM is the integrating ADC:
Read this article:
http://en.wikipedia.org/wiki/Integrating_ADC

[stitch]

I'm pretty sure it's as simple as ohm's law. I believe a DMM measures the voltage drop across a known resistance, also known as a current shunt.

Yes, but how does the DMM measure the voltage drop?

[FrankenPC]

I'm pretty sure it's as simple as ohm's law. I believe a DMM measures the voltage drop across a known resistance, also known as a current shunt.

Yes, but how does the DMM measure the voltage drop?

That's a good question.  I don't know how modern ADC's work.  I THINK I know how simple ones work.  They take in a voltage and pass it by a resistor ladder.  As each divider is passed, it triggers a pulse that goes through a counter which ultimately puts out a byte (0-255) depending on how far the voltage passed through the ladder. 

TI has a white paper that describes doing something similar using capacitors:
http://www.ti.com/lit/ds/symlink/tlv1548-q1.pdf

There's also successive approximation and sigma-delta which look pretty sophisticated.

[IanB]

Yes, but how does the DMM measure the voltage drop?

If it's a digital voltmeter (DVM) it's done digitally.

The fundamental element is a comparator, which is a circuit built out of transistors that compares two voltages A and B. If A is greater than B then the comparator produces an output, otherwise it doesn't. To make sure the circuit draws very little current (which would disturb the reading) the transistors are field effect transistors which have a very high input impedance.

A single comparator can produce a single on/off state, which means that in binary terms it is a "1-bit" converter.

However, various clever circuit arrangements can use many comparators to produce a more precise binary output with 8 or 12 bits of precision.

These multi-bit samples can then be filtered and scaled and otherwise processed to produce a reading on the display.

[FrankenPC]

Yes, but how does the DMM measure the voltage drop?

If it's a digital voltmeter (DVM) it's done digitally.

The fundamental element is a comparator, which is a circuit built out of transistors that compares two voltages A and B. If A is greater than B then the comparator produces an output, otherwise it doesn't. To make sure the circuit draws very little current (which would disturb the reading) the transistors are field effect transistors which have a very high input impedance.

A single comparator can produce a single on/off state, which means that in binary terms it is a "1-bit" converter.

However, various clever circuit arrangements can use many comparators to produce a more precise binary output with 8 or 12 bits of precision.

These multi-bit samples can then be filtered and scaled and otherwise processed to produce a reading on the display.

So, if a ADC has 8 bit resolution, you're saying there's 256 comparators in that chip?  Or does it do something tricky like see if the voltage is half the reference, then half that, then half that, etc...?

[IanB]

So, if a ADC has 8 bit resolution, you're saying there's 256 comparators in that chip?  Or does it do something tricky like see if the voltage is half the reference, then half that, then half that, etc...?

There is not necessarily a need for many comparators. You can trade off the number of comparators for operational speed.

In an application like a multimeter you only need a slow sample rate. In this case, the integrating ADC referenced in the post above can in its simplest form use only one comparator. Instead of many comparators, it instead uses a timer to measure a length of time proportional to the input voltage. Of course the timer now has to be a digital timer or the measurement problem has just been transferred from one place to another.

[T3sl4co1l]

By counting time and using the fundamental capacitor equation I = C * dV/dt, holding I, dV or dt constant in various ways (see the earlier link about slope conversion).

Tim

[jeremy]

So, if a ADC has 8 bit resolution, you're saying there's 256 comparators in that chip?  Or does it do something tricky like see if the voltage is half the reference, then half that, then half that, etc...?

The 256 comparator version is what is known as a flash ADC (not to be confused with flash memory). They are used nowadays in really really fast ADCs (like digital oscilloscope fast) but are very expensive due to their size and usually lower resolutions.

The most common one I see is a successive approximation register ADC. These represent a good tradeoff between speed, cost and resolution. Most microcontrollers use a SAR ADC.

The best resolution ADCs are usually delta-sigma ADCs. These are usually super accurate and >=16 bits but slow and expensive.

Of course there are other special versions as well as hybrids of these three, but these are the most common ones I have seen.

Most measurement equipment like multimeters use either SAR or delta-sigma ADCs. My beloved agilent 34405A uses a kind of delta-sigma method I believe.

[Hideki]

That's actually not the case for most multimeters. Look at what Wytnucls posted about the Integrating ADC -- dual-slope or something more fancy.

[jeremy]

Ah interesting. I'm pretty sure that most of my multimeters are delta-sigma, but perhaps that is just luck!

[stitch]

Thanks for all of your responses to my topic question.  From your responses, I think I have been able to find an answer to my extreeeeeeemly fundamental question:  "Can anyone explain to me what part of a modern digital multimeter replaces the coil and magnets of the galvanometer?"

I'm pretty sure now that the answer is … a resistor and a capacitor.  Who knew?
Apparently a modern DMM (DVM) makes use of the RC Time Constant.  By counting and comparing the time it takes to charge a capacitor, the DMM (DVM)  calculates voltage.  But here's what's really interesting …

The theory behind the coil and magnets of the galvanometer (electromagnetic induction theory) was developed by Joseph Henry and Michael Faraday around the year 1830.  The invention of the capacitor (Leyden Jar) was made around the year 1745 by Ewald Georg von Kleist and Pieter van Musschenbroek.  The modern DMM (DVM) uses an invention that is almost 100 years older than the antique galvanometer!

[Wytnucls]

It won't work without a voltage reference (Clark Cell 1873)
1.434 volts

http://en.wikipedia.org/wiki/Clark_cell

[pickle9000]

A good question judging by the answers.

[Richard Crowley]

As Wytnucls said, the meter uses some internal reference to COMPARE the voltage under measurement with the internal reference.  In the old-school d'Arsonval, the "reference" was a very small, internal "hair-spring".  In modern solid-state, digital meters, a very crude process known as "approximation" is done.

From the reference voltage they use resistors to create all the intermediate voltages: 0.1, 0.2, 0.3 etc....
Then they ask, over and over again, is the measured voltage = 0.1?, is it = 0.2, is it = 0.3, is it = 0.4, is it = 0.5, etc. etc.  Then, when they find a match, it is displayed on the readout.  NOTE: This is an extremely truncated and over-simplified explanation, but I think it is what stitch is asking.

As Jeremy explained, there is a different kind of analog to digital converter ("flash converter") where they compare the input voltage SIMULTANEOUSLY from 0.0V all the way up to to 2.56V  That method is much faster, but it takes more power and has lower resolution.  Now in some applications (very fast signals like video, etc.) that trade-off is worth the extra power and lower resolution.  But typically it is not suitable for a proper test meter.

That is how modern digital multi-meters measure VOLTAGE.  They measure CURRENT simply by using Ohm's Law and measuring the voltage drop across a resistor (actually a very low-resistance "shunt"). And they measure RESISTANCE, again by Ohm's Law, where they apply a current across the resistor under test, and they measure the voltage across the unknown resistor.  So you see that no matter what you are measuring (volts, ohms, amps, etc.) it comes down to some voltage that is converted from analog to digital (A to D) and shown on the display.

[stitch]

Actually my question was not about how to measure electricity, but rather how to firstly sense it so that it can subsequently be measured.  In my topic question, I used the example of the galvanometer which employed "magnets and a coil attached to a needle".  I probably should not have mentioned the needle (the measuring part), because the sensing is done with the coil and magnets (the coil magically moves when current is passed through it). Yes, there was a needle attached to the coil to measure the movement of the coil, but it was the magical movement of the coil which revealed to human observers that something was happening - regardless of the magnitude of what was happening. That's the sensing part that I was talking about … you know, the kernel, the phenomenon, the very first step.

[T3sl4co1l]

Well, when an electron and a proton love each other very much....

Neglecting the underlying physics though, it's just Ohm's law through a bunch of switched resistors, of the order 1-10M for the outward most ones.

[Richard Crowley]

Actually my question was not about how to measure electricity, but rather how to firstly sense it so that it can subsequently be measured.

That seems like an odd question. Clearly, the old d'Arsonville/Weston meter movement uses simple electro-magnetic effect to "measure" the current through the coil.

But if you are asking about how that works in solid-state equipment, you are asking about current flow and holes and electrons, and solid-state physics.  On one level a simple light switch is trivial to explain, but on another level, how current flows through a conductor has yet to be explained completely.  It is not clear at what level of abstraction you are asking your question?  Are you willing to take "for granted" what is inside the integrated circuits?  Or are you asking about how the integrated circuits operate?

[ejeffrey]

Actually my question was not about how to measure electricity, but rather how to firstly sense it so that it can subsequently be measured.

Then I think the answer to your question is "a transistor, usually a field effect transistor'.  A voltage applied to the 'gate' of a FET creates an electric field inside the channel, which either allows or prevents carriers to flow from the source to the drain, and it does it while drawing negligible current at the gate electrode.  If the gate voltage is above some threshold a large current flows and you have a one, otherwise a zero or small current flows which is zero.  You can either have a fixed threshold (as in ordinary logic gates used to implement digital logic) or a comparator that gives you a 1 if voltage A is greater than voltage B.

[wiss]

May I try?

The input voltage is feed via a buffer to a resistor, the current through that resistor charges a capacitor for a fixed amount of time (1 power-line cycle for example), the input voltage is replaced with a reference-voltage and the same capacitor is discharged through the same resistor and the number of clock-cycles it takes to get back to zero is counted.

For higher voltages an input-divider is used, for lower voltages the input buffer usually can apply 10 times gain. Sometimes the reference voltage can be switched 2 and 20 volts, sometimes not.

[jucole]

In the early days of electronics, a galvanometer was used to sense electric current.  The galvanometer used magnets and a coil attached to a needle to do this.  Can anyone explain to me what part of a modern digital multimeter replaces the coil and magnets of the galvanometer? Asked another way, what electrical phenomena are exploited by modern DMM's to sense current?

My explanation is simply the ADC replaces the "sense" coil of the galvanometer;  and there are many types of ADCs to choose from.  http://en.wikipedia.org/wiki/Analog-to-digital_converter

[Dave]

May I try?

The input voltage is feed via a buffer to a resistor, the current through that resistor charges a capacitor for a fixed amount of time (1 power-line cycle for example), the input voltage is replaced with a reference-voltage and the same capacitor is discharged through the same resistor and the number of clock-cycles it takes to get back to zero is counted.

For higher voltages an input-divider is used, for lower voltages the input buffer usually can apply 10 times gain. Sometimes the reference voltage can be switched 2 and 20 volts, sometimes not.

You have described a dual-slope integrating ADC.
There are, of course, many other types of ADCs, the most accurate being the multi-slope ADC, a successor of the dual-slope ADC. They are used in high-performance multimeters like the spectacular 3458A.

[wiss]

May I try?

The input voltage is feed via a buffer to a resistor, the current through that resistor charges a capacitor for a fixed amount of time (1 power-line cycle for example), the input voltage is replaced with a reference-voltage and the same capacitor is discharged through the same resistor and the number of clock-cycles it takes to get back to zero is counted.

For higher voltages an input-divider is used, for lower voltages the input buffer usually can apply 10 times gain. Sometimes the reference voltage can be switched 2 and 20 volts, sometimes not.

You have described a dual-slope integrating ADC.
There are, of course, many other types of ADCs, the most accurate being the multi-slope ADC, a successor of the dual-slope ADC. They are used in high-performance multimeters like the spectacular 3458A.

Of course! One point being that the same resistor/capacitor is used for both slopes. And you'd like to measure zero some way, in a fluke 8050 the zero-offset is stored in a capacitor during the integration-phase, in a keithley 192 the "zero" is counted and subtracted from the result for example. And it is possible to mix the integration/deintegration phases in various ways. I strongly recommend to read the fluke 8050, fluke 8800 and keithley 192 manuals (available via google).

[stitch]

May I try?

The input voltage is feed via a buffer to a resistor, the current through that resistor charges a capacitor for a fixed amount of time (1 power-line cycle for example), the input voltage is replaced with a reference-voltage and the same capacitor is discharged through the same resistor and the number of clock-cycles it takes to get back to zero is counted.

For higher voltages an input-divider is used, for lower voltages the input buffer usually can apply 10 times gain. Sometimes the reference voltage can be switched 2 and 20 volts, sometimes not.

That's a good answer.  Very much on target to what I was inquiring about.