# Peak voltage detector schematic

Discussion in 'Electronic Design' started by Ignoramus20878, Nov 29, 2005.

1. ### Ignoramus20878Guest

I would very much like to make some sort of a "peak voltage
detector". A device that, when plugged into a circuit, after a
while, builds up an adequate representation of peak voltage. That
would be again for my IGBT inverter.

That's for peak voltages under, say, 1 kV.

I tried making one from just a cap and diode, but multimeters would
discharge it too quickly and getting reliable readings is difficult.

After reading the art of electronics, something like this comes to
mind:

Make a small capacitor charged by diode (and a series resistor to
limit current). After a while, voltage would build up there.

Connect a voltage divider (say, 1 megohm and 1 kOhm in series). To the
cap. That would be a 1:1001 divider.

Connect that as input to some sort of FET based amplifier (that takes
voltage and does not require input current) that produces same
voltage, but at higher current. That would be the input ot a regular
multimeter.

The multimeter reading would need to be multiplied by 1000 to get the
actual voltage.

Alternatively to all this, does anyone know if higher end multimeters
like Fluke 8050A (which I have) can measure voltage without
discharging the cap so much? I have one lying around somewhere.

i

2. ### The PhantomGuest

Such a divider wouldn't be much of an advantage since the typical input
resistance of a modern DVM is about 10 megohms. The DVM would be 10 times
slower at discharging the cap than your divider.

You could put 9 resistors of 10 megohms each in series to make a 90
megohm resistor and then put the meter at the bottom of the string for a
total of 100 megohms resistance across the cap. Then multiply the reading
by 10. If you have a couple of DVMs, use a second one on the ohms range to
measure the input resistance of the other (set to volts range) to make sure
it's really 10 megohms. If it's not exactly (to 1% or so) 10 megohms, use
the measured value to figure your divider ratio. You might also measure
the 10 megohm resistors in the string to see what their total is, and use
that value in figuring your divider ratio.

The 100 meghohm load on a 1 uF cap should be a slow enough discharge to

3. ### John - KD5YIGuest

Maybe I'm wrong, but I thought most multimeters today came with 10 megs
input impedance. That's 10 times more than you propose.

You have an oscilloscope, as I recall. Can't you set up the triggering to
give a trace when the voltage exceeds some value you set with the trigger
level control?

Good luck.

John

4. ### ehsjrGuest

AIRC, Art of Electronics shows a peak detector with op amps.
Why not use what they show?

Ed

<snip>

5. ### Ignoramus20878Guest

I have two handheld MMs, one is a nice but old Extech and another is a
\$3.99 harbor freight unit.

Both discharge the caps rather quickly, I can barely get the reading.

I also have a Fluke 8050A, which seems to have 100K resistance. (I
I suppose I can, yes. I am open to suggestions. Tek 2245A.

i

6. ### Ignoramus20878Guest

Sure, I guess I missed it or did not get to it. Do you know what
chapter or page it was on?

i

7. ### The PhantomGuest

I explained what to do in another post:

" Such a divider wouldn't be much of an advantage since the typical input
resistance of a modern DVM is about 10 megohms. The DVM would be 10 times
slower at discharging the cap than your divider.

You could put 9 resistors of 10 megohms each in series to make a 90
megohm resistor and then put the meter at the bottom of the string for a
total of 100 megohms resistance across the cap. Then multiply the reading
by 10. If you have a couple of DVMs, use a second one on the ohms range to
measure the input resistance of the other (set to volts range) to make sure
it's really 10 megohms. If it's not exactly (to 1% or so) 10 megohms, use
the measured value to figure your divider ratio. You might also measure
the 10 megohm resistors in the string to see what their total is, and use
that value in figuring your divider ratio.

The 100 meghohm load on a 1 uF cap should be a slow enough discharge to

If the DVM you would be using to measure the voltage at the bottom of the
divider isn't 10 megohms, then find out what it is by measuring it with a
second DVM, and use the actual input resistance of the DVM to figure out
what your divider ratio is. If the actual input resistance of the DVM is
Rdvm and if you have used 9 resistors of 10 megohms in series, then you

8. ### GregSGuest

Is an oscilloscope out of the question? It would be easier to record peaks.
Most ohmeters only measure up to 10Meg, but there are some that go to 100 Meg. Perhaps their
impedance is also higher. In any case, adding series resistance will increase input impedance.

greg

9. ### The PhantomGuest

It absolutely would. That's how voltmeters work; you're just increasing
the multiplier resistance (as it's called).

Think about it. If your DVM has 10 megohm input resistance and you are
measuring 1 volt, the same current is seen by the meter as if you add 90
megohms additional and then measure 10 volts. If the meter can do the
first case, it can do the second case.

10. ### The PhantomGuest

An easier way to calibrate would be this: Take one of your power
supplies and set it to its highest voltage, maybe 20 or 30 or even 50
volts. Measure it with a DVM, and write down the voltage. Put the 90
megohms in series with the DVM and measure the voltage. Divide the reading
without the 90 megohms by the reading with the 90 megohms. That is your
multiplier factor and it takes into account the input resistance of the DVM
and the actual values of the 90 megohm string.

11. ### Ignoramus20878Guest

Thanks Phantom. I am not sure if a divider with such high resistance
(10 MOhm) would produce currents that are high enough to allow the DMM
to reliably measure voltage. I will try your suggestion however.

i

--

Page 217

14. ### Ignoramus5455Guest

thanks, will check tonight.

i
--

15. ### Ignoramus5455Guest

Jim, sorry if I am missing the obvious, but this peak detector on page
4 would quickly droop (discharge cap) if connected to a multimeter,
right? I am talking about the part starting with U3.

i

16. ### Jim ThompsonGuest

For just peak detector purposes, delete the rectifier sections and
input your (attenuated) signal into the plus input of U3.

Increase R4 to suit your required droop rate.

Use another OpAmp section, connected for unity gain, to buffer

...Jim Thompson

17. ### Ignoramus5455Guest

That's exactly what I am wondering about, what opamp would be most
suitable for this purpose (infinite input impedance, voltage driven or
whatever it is called). To avoid any droop altogether. Something in
DIP package or just a piece with contacts would be best.

i

18. ### Jim ThompsonGuest

My favorite OpAmp (when I last did discrete designs) was the TL084
(TI) or MC34084 (Motorola).

...Jim Thompson

19. ### Ignoramus5455Guest

Yes, looks just what I need, thanks a lot!

i

20. ### Ignoramus5455Guest

One more question, is the output of his opamp going to be enough to
reliably read it with a multimeter? It looks like yes, but I want to
double check. I really want to build a nice PVD.

i