Op-Amp AF millivolt meter problem

[Bazz4CQJ]

I wanted to make a small unit to enable easy adjustment of a testpoint to read 15mV at ~14kHz (what could that possibly be?).
Searching Google for ideas, I found the attached design for an AF millivoltmeter good to ~20kHz based on a LF353 IC and published on the internet by David Bull. What’s quite unusual about his design his that he includes quite detailed notes on the design and testing of the circuit http://www.uk-dave.com/wp-content/up...o-mv-meter.pdf so offering confidence that it is a good choice (schematic only is attached).

My unit seems to be working well (only tested at 430Hz so far) the only problem is that at switch on, the meter hits the FSD stop very hard (meter is 100uA, 1.3k internal resistance, FSD needs 0.13V). A 2uF non-polarised cap across the meter provides a little alleviation. I wonder if anyone can suggest a way to get around this other than putting components across the meter?

Thanks

B

[Bazz4CQJ]

Thinking about this more, I think that my cap across the meter should go. Maybe, a fairly hefty capacitor (at least 100uF) across both 9V lines and a carefully selected resistor in both lines to give a "soft start"?

B

[Terry_VK5TM]

I'm thinking it may be either C3 or C5 (or both together) that is causing it to pin the meter.

Not sure how to cure that at this stage without upsetting the normal operation of the circuit.

[Radio Wrangler]

100uA meter.... What's its coil resistance?

This wil allow you to calculate the voltage across it at FSD. Then maybe a back to back pair of diodes could make a clamp. The diodes might range from back to back schottkies for low voltages to back-to-back silicon, to strings of diodes for higher voltages.

If this isn't good enough, some of the resistance in R8 can be moved inside the bridge. EG drop R8 to 12k and add 3k in series with the meter, do back to back diodes across the combo of the new resistor and meter.

By playing with choice of diode arrangement and partitioning R8, you should be able to make an effective clamp that should save the day for the movement.

Plan B:

Use a depletion mode MOSFET to short the meter in a circuit with the diode bridge and R8 swapped, so that the meter lives closer to ground voltage, put the MOSFET gate to the -9v rail via a high value resistor, stick a capacitor across gate to source to give a time delay. so on turn on your meter is shorted for a while to allow the turn-on transient to finish.

David

[Bazz4CQJ]

Originally Posted by Radio Wrangler

100uA meter.... What's its coil resistance?
David

R is about 1.3k, so Vfsd=0.13V

I guess that you know what it's for .


B

[Radio Wrangler]

Actually, I haven't figured out what it's for. I can be damned thick at times.

OK, 0.13v means a diode clamp will not do anything until several times FSD current is flowing. So if we do ...

Um, interrupt, I just missed a trick! It doesn't need back-to back diodes, just a single one! Doh!

So if we put a single diode in as a clamp, say 1N4148 general purp Si switching diode with anode to meter + terminal, it will start to turn on at 0.6v. Let's say 150uA as the clamped current, so 0.6v/150uA gives 4k Ohms.... less 1.3k gives 2.7k Isn't that lucky! A diode across DC port of the bridge rectifier and 2.7k on the way to the meter.

The David Bull circuit is neat in that it current drives the meter. The current is set by R7 and VR1, NOT by R8. R8 just sets the swing the opamp second stage has to to to give the required current. So to avoid increasing the swing demands on the opamp, let's reduce R8 by roughly 2.7k. So 12k should be nice.

The meter will still kick on turn on, but the overcurrent will only be 50% If you up the 2.7k resistor, you'll bring the clamp level lower. The clamp however won't be abrupt at these ciurrents, if you have it too low, your readings will be distorted to the top end of your scale.

David

[Electronpusher0]

Your problem is C3 and C5, these provide a very low frequency rolloff and prevent the dc offset at the inputs being amplified and appearing on the output of the op-amps.
At switch on they are effectively short circuit so the offset is amplified and drives the meter over.
Once the capacitors charge the thing settles down and the amps become unity gain buffers at dc.
As suggested in post 4 some method of shorting the meter out for a period at switch while the circuit settles would be best. This could be as simple as a relay driven by a simple time delay circuit, or a fet as suggested.

Peter

[Bazz4CQJ]

Thanks for the suggestions. If I were to use a 2p 6w switch for the ranges, I could use the second pole for the meter and have the first position as "Warm Up" where the meter is out of circuit, but then goes in to circuit in the subsequent positions. It's crude, but means no changes to the little strip board or anything else?

@David; think VCM163, which calls for the use of the rare Avo electronic meter to set the level on the oscillator board. I've usually used a scope, but this will be handier.

Thanks

B

[retailer]

I built a similar ac millivoltmeter a few months back - uses a moving coil meter but can also use a digital DPM with no circuit changes, only has two ranges 30mV and 150mV fsd but could be wired for more if needed, I'm sure there is no kick from the meter but if there is it is only minimal. If you haven't invested a lot of time and funds into your build you might want to give it a try. PM me if interested and I'll post the circuit.

[Radio Wrangler]

Originally Posted by Bazz4CQJ

@David; think VCM163, which calls for the use of the rare Avo electronic meter to set the level on the oscillator board. I've usually used a scope, but this will be handier.

Ah. VCM163s are rather thin on the ground, but EA113s are even rarer. They're rather good with 10mV ranges on both AC and DC. I put mine on a Datron standard labs calibrator and found the accuracy in-spec on all ranges and the AC voltage was 3dB down at 200kHz. I was quite impressed.

Oh and my EA113 bangs the pointer on the endstop when the AC button is pressed!

David

[Radio Wrangler]

Oh, with opamps on the go, there's no need to use a 100uA meter, something substantially more robust would be viable.

David

[Trevor]

try another 100 uf capacitor from the feedback junction to the positive supply thus creating an artificial center point from the ac point of view
Trev

[stuarth]

Doesn’t this point to a DC imbalance in one or other of the op-amps? The DC gain with C3 or C5 shorted (the switch on condition) is not huge.

When the circuit has settled, what is the DC voltage at C3 and C5? Assuming the capacitor “ground” terminal is connected to the centre point of the two batteries, the voltage could be positive or negative, so those capacitors should be non-polarised.

Would an op-amp with provision to set the DC balance (with C3 & C5 shorted) be better? Maybe the original builder happened to have devices with better balance.

Stuart

[Bazz4CQJ]

Mmmm... my profound lack of understanding of opamps is something of a hindrance here. I did wonder about the polarity of caps when I was assembling.

However, apart from clouting the meter, it seems to work as required. I've just found a largish 1mA meter which offers some 'mechanical' alleviation and gives better resolution than the current meter. I'll do some tests and report back.

Thanks

B

[Radio Wrangler]

Scale down R7, R8, VR1
Scale up C5

Actually, If you've got only a fixed frequency to measure, you might not be interested in going any lower in frequency. so maybe C3 and C5 started out bigger than you needed? THe original circuit will probably be scaled to be flat down to 20Hz. What frequency is the VCM163 oscillator?

Lower C in these parts should lessen the transient.

David

[Radio Wrangler]

Let me see 680 + half of 560 is 960, which with 100.47 uf gives 1.6 Hz Yep, you don't need to go quite so low, I suspect

5.6k and 100uf gives 0.28 Hz Methinks C3 could be reduced somewhat

So keeping C5 at 100uf and rescaling those resistors for the 1mA movement gives a 16Hz -3dB point.

C3 could be reduced to 1.77 uF to put its roll off at 16Hz. 17uF would put it at 1.6Hz

I think he just used a 100u here because he had them.

Isn't calculation wonderful! Opamps are fun, but there are a few tricks to learn about them.

David

[Tony Jaques]

Hi there
I built the PW February 1974 "ac millivoltmeter" design, yes, that long ago!
The on/off switch is a two pole three way, one pole being used to short out
the 100 microvolt meter until transients are passed. This also provides the sometimes useful facility of seeing the output on a 'scope.
If needed I can attach the article (when I work out how to do it) but the complete magazine can be found on americanradiohistory.com.

Hope this helps
Tony
(G3PTD)

[Tony Jaques]

Oops!
My apologies that should have been PE (Practical Electronics) Not PW.
Tony...

[Bazz4CQJ]

Thanks for all your inputs. I've spent the day in the back garden rebuilding a Suffolk Punch mower, but I'll try get back the meter later tonight.

The 163 runs at ~14kHz, and although it will be useful to have something that covers the full audio range for general use, a little bit of roll off at the low end is quite tolerable for me.

As it is already assembled on a compact area of strip board, some changes are easier than others, but C3 and C5 at located right at the edge, so are very easy to get at. In the original design, he used a meter which had three times the internal resistance of the one I've started with, so perhaps his was hitting the stop, but a little less brutally?

B

[Bazz4CQJ]

OK, C3 changed to 1.84uF (didn't have a 1.77uF ) resin cap and the problem has pretty much gone away.

The lowest frequency my present AF generator will provide is 400Hz and I couldn't see a 'before and after' change in the response of the meter at that frequency. So, I just need to confirm it's good to 14kHz and above and then I can get back to sorting out the 163.

Thanks for all your inputs .

B