P-P reading voltmeter with hold function

[Jaguarjoe]

I searched the forum and searched the web but I still haven't found what I'm looking for (stolen from U2).

I have a very low frequency (~.5-1Hz) bumbling waveform that has positive as well as negative excursions. I want to measure the highest positive peak, then the lowest negative peak, and add their absolute values to obtain a peak to peak indication. Then I'll need to store that P-P value for at least a few minutes.

There are numerous "bi-polar" peak meters but most of them are simply a precision FW rectifier with a peak holding capacitor hanging off the tail end. These do not measure actual P-P amplitudes, they are voting systems- whichever negative or positive peak is highest "wins".

My thought was to make a negative peak meter and a positive peak meter and subtract their two outputs to get a P-P indication but this seems a tad clumsy.

Any ideas?

Thanks,

Joe

 

[dougy83]

You can make a simple adapter to fit between your signal source and a regular voltmeter, e.g.


You didn't mention the voltage range you are measuring; if it is outside the range of the opamp voltage rails (in the above circuit), then you can place a voltage divider on the input. If you're measuring a tiny voltage, you can put gain on the opamps. If it's a large voltage and you don't care about an error of ~1.2V, then you can just use two diodes and two capacitors (no opamps or extra power supply required)

 

[MrAl]

Hi,

I like dougy's idea, or even just the two diodes and a little calibration if the input voltage is high enough. Schottky diodes might be good there.

 

[dr pepper]

If you want a voltage referenced to ground rather than a differential, then use a differential op amp on the end of dougs circuit.

 

[The Electrician]

There are numerous "bi-polar" peak meters but most of them are simply a precision FW rectifier with a peak holding capacitor hanging off the tail end. These do not measure actual P-P amplitudes, they are voting systems- whichever negative or positive peak is highest "wins".

It's not clear just what you want. Do you want a meter which gives you a number (volts P-P) in a display which you can then write down? Do you want a voltage proportional to the P-P volts which you can then feed into an A-D converter?

Is your objection to what you call "bi-polar" meters that you have to look at two values and add their absolute values, rather than just getting a single number to look at?

 

[MrAl]

Hello there Electrician,

My take on this was that he has a signal of some sort that goes positive by a value that is less or more than how much it goes negative and that the waveshape is probably not always sinusoidal, and he wants to know the value between the positive peak and the negative peak even though strictly speaking for this app this will not really be volts peak to peak the way it is for a sine wave.

If it was a sine wave i would say just use a capacitor between the meter and the source to be measured (some meters need this) and have the meter on the AC scale and then do the math, but it sounds like it may not always be a sine wave.

 

[Jaguarjoe]

Looks like Dougy's way will work with a little tweaking. Looking in the Art of Electronics book, there's lotsa good info about his circuit and more.

The data from this will feed a DPM. The longer it holds, the easier it will be for the doctor to write the value down on a piece of paper.

Thank you, guys!

 

[The Electrician]

Well, I think you indirectly answered my questions, but now I have a couple more.

How often does your bumbling wave change its max and min values? You said you wanted to hold the P-P number for a few minutes. Does this mean the wave doesn't change often?

Dougy's circuit as it stands has no way to discharge the output capacitors. If the peak values of the wave decrease, the circuit won't reflect that fact until the voltmeter impedance discharges the caps, which could take a long time if the caps are big and the voltmeter impedance is high.

You also haven't said if this is a situation where you will be making these measurements on an ongoing basis, maybe for weeks, months or years. If it were a long time, building a special circuit could make the measurements more convenient.

If you want a quick solution and you're only going to do this a short while, just use a suitable DVM. My Fluke 187 (and numerous other hand-held DVMs) has a min-max mode. You just put the meter in that mode, wait a while and you have both the maximum positive voltage and the maximum negative voltage. The sum of the absolute values is the P-P voltage, and the DVM will hold the values as long as you want. Assuming that the wave may change values and you want another reading, reset the meter and do it again.

If your wave only changes its peak voltages very slowly, and you really want a single number and no arithmetic, you could do this with one capacitor: simply pass the wave through a suitably sized capacitor and clamp the output side with a diode so it can't go negative. Then the voltage there will be a positive only wave whose positive peak will be equal to the P-P value (minus the forward drop of the diode, a Shottky maybe), which you could measure with the peak holding capability of a DVM; this way, you only have one reading to note.. You could use an opamp based ideal diode to clamp and avoid the forward drop error. With any of these capacitor holding methods, you need to reset the capacitor voltage(s) to zero before you take another measurement.

 

[Jaguarjoe]

Well, I think you indirectly answered my questions, but now I have a couple more.

How often does your bumbling wave change its max and min values? You said you wanted to hold the P-P number for a few minutes. Does this mean the wave doesn't change often?

The wave could ideally have 0 fluctuation, or predominately + traces, or predominately - traces. It bumbles randomly.

Dougy's circuit as it stands has no way to discharge the output capacitors. If the peak values of the wave decrease, the circuit won't reflect that fact until the voltmeter impedance discharges the caps, which could take a long time if the caps are big and the voltmeter impedance is high.

The plan is to use MOSFETs to disch. the caps. Using the 1N914 leakage spec and a 1uf cap and a non-inverting follower after it, it should hold the peak for a relatively long time (~3 min)

You also haven't said if this is a situation where you will be making these measurements on an ongoing basis, maybe for weeks, months or years. If it were a long time, building a special circuit could make the measurements more convenient.

Nothing ongoing, just another tool in his toolbox. Might get used 3x/wk. He will push "start", a 5 minute timer will allow the peak meters to track the output from an integrator. Highest P-P value wins]/QUOTE]

If you want a quick solution and you're only going to do this a short while, just use a suitable DVM. My Fluke 187 (and numerous other hand-held DVMs) has a min-max mode. You just put the meter in that mode, wait a while and you have both the maximum positive voltage and the maximum negative voltage. The sum of the absolute values is the P-P voltage, and the DVM will hold the values as long as you want. Assuming that the wave may change values and you want another reading, reset the meter and do it again.

Huh, coincidentally, I have a 187 also! Until Harbor Fright came out with there free little red DVM's my 187 was the workhorse. This is a self contained unit, the Fluke wouldn't fit in very well.[/QUOTE]

If your wave only changes its peak voltages very slowly, and you really want a single number and no arithmetic, you could do this with one capacitor: simply pass the wave through a suitably sized capacitor and clamp the output side with a diode so it can't go negative. Then the voltage there will be a positive only wave whose positive peak will be equal to the P-P value (minus the forward drop of the diode, a Shottky maybe), which you could measure with the peak holding capability of a DVM; this way, you only have one reading to note.. You could use an opamp based ideal diode to clamp and avoid the forward drop error. With any of these capacitor holding methods, you need to reset the capacitor voltage(s) to zero before you take another measurement.

I'm gonna try that!

 

[dr pepper]

You can get dpm's that have a differential input.

 

[Jaguarjoe]

I already purchased the DPM, so I sunk another $1.00 into the project and added a op=amp subtractor.

 

[ronsimpson]

I have done a similar project using 'software'.
The little 8 pin PIC10Fxx, PIC12Fxx and PIC16Fxx can measure an analog voltage 5,000 times a second and through math find the +/- peaks. Last time I did this with a PIC I only got 10 bit resolution. How accurate do you need to be. One part in a 1000? Analog answers are probably not that good. There are many types of computers that are well under $1.00 us.

I am now working on a XMEGA that is doing much more math than this on 4 analog inputs at the same time. My signal is 4000 times faster than yours.

 

[2nero]

[MODNOTE]off topic[/MODNOTE]

 

[dr pepper]

I've used the 10f200 many times, I've never used the comparator in a pic as an atod, did you just keep changing the reference volatge to find the voltage?

 

[ronsimpson]

Analog "bumbling waveform" goes to analog in pin on the micro.
The software looks for the most positive and most negative reading in a period of time.

ADC=value from ADC.
+PK= positive peak voltage.
-PK= negative peak voltage.
Loop:
Get analog voltage.
If ADC <= +PK then do nothing.
If ADC > +PK Then make +PK=ADC. (up date +PK)
IF ADC >= -PK then do nothing.
If ADC < -pk then update -PK
go to loop.

 

[dr pepper]

Yep seems simple enough, a bit more processor time req but still a simple way of getting an a to d done, I'll consider that when I need an a/d, esp something like a battery indicator.

 

[ronsimpson]

A little PIC (6 or 8 pin) is not much of a part but:
In the space of one op-amp I can get functions like:
voltage compare, amplifier, low pass, high pass, PK detect, average, etc all at the same time.
The response is a little slow but fine through the audio range.

Peak Detectors can respond too fast. They can capture a small bit of noise or glitch and think it is data. Your program can read in 10 samples, average, then do the PK detect. This way one sample being off will not effect the pk value.

Most PK detectors need to get back to "0". Normally I, every x-time periods, (PK=PK-1) so the peak number heads back to zero. The down ward formula can be much smarter than that if you want. Example: if the signal is close to the PK value then slope down slow but if the signal is very low then slope down fast.

The good part is you don't like the results you can change the formula with out making a new PCB.

 

[Jaguarjoe]

I think that's what I'll do, Ron. This is a one hertz freq, project so even a Basic programed Picaxe could handle it.