Simple low freq Am modulator

[MRIguy]

Looking to xfer some low varying DC data.
Tired V to F converters and they proved problematic.

Thinking of simple AM scheme.

Looking at Gilbert cells/multipliers but is there a simple op-amp type AM modulator circuit?

Ralph

 

[jsolar]

Yes. I assume you want to convey your "varying DC" as the amplitude on an AM carrier. As long as your carrier vfrequency is much higher than you max DC rate change your idea should work very well. Ham people use a dual gate FET, but you can do this with one BJT if needed. The RCA IC's are not easy to get a hold of, but there are replacement parts available. You can get the osc. and the modulator to be the same part if you want to be tricky.
John Solar

 

[MRIguy]

Thanks John;

I was planning to use the carrier of 1KHz, the modulation is signal from a slowly changing (< 50 Hz) pressure transducer. Pretty much DC.

I'm stuck with this low freq app. due to some systems issues. I usually do RF circuits and I have used DGFets, but I was hoping for an op-amp or IC solution.

I'm thinking of a AD633 set-up in AM mod (where they leak the carrier back into the DSB-SC mod) cause I need to use a simple peak detector.

I just thought there was a straight forward op-amp technique using a diff amp.

Thanks again.

Ralph

 

[Warpspeed]

A lot of the communications type modulators are really designed for speech transmission, where a bit of distortion or random shift in the dc baseline is of no real importance.

For a pressure transducer I would assume reasonable linearity and temperature drift of the modulator might be issues to think about. Something a bit better than a crude voice "communications" type of modulator may be called for.

Your idea of using a gilbert cell, or even a proper precision pre trimmed analog multiplier would be my choice.

 

[sreid]

How about an Analog Devices AD633 analog multiplier. Sine (or square wave ) in one input, amplitude modulation input to the other input.

http://www.analog.com/en/prod/0,,AD633,00.html

 

[jsolar]

You can buy the translinear parts, but you can make them yourself for the BW and dynamic range that is good enough by strapping a diode in the feedback path of an op-amp. The idea is that op-amp ckt gain is proportional to feedback resistance. If you add a non-linear resistor, using the exponential characteristics of a transistor, you can achieve log amps or AM modulators. I do not have my copy in front of me, but I remember learning a lot from:

Nonlinear Circuits Handbook, Engineering Staff of Analog Devices, Inc., Sheingold, D., ed. Norwood, MA: Analog Devices, Inc. (1974)

You may want to try a simple 555 like timer for your application. A fixed frequency of 1 kHz and vary the PW proportional to your DC signal. A simple low pass or active bandpass will give you something that is easily AM detectible. If you have access to some old stock switching power supply parts out can do the same thing. Even with just op-amps you can build and osc. and PW control.

Best Regards,
John Solar

 

[MRIguy]

The AD633 is first on the list, thanks sreid.

I do have a copy of the Analog Devices handbook ( now you know I'm an oldtimer!) that is where I got the multiplier idea, thanks John

I found another IC used for feedback generation, a Unitrode UC1901. It has an on board osc., multiplier, and drivers. Kinda cool if I can get it to work.

But I hope it has the accuracy for my app that Warpspeed mentions.

Thanks guys, as I said I'm an old RF guy snob, but doing this low freq. stuff can be just as challenging.

Anyway, AD633 and the UC1901 will be ordered today.

 

[sreid]

If you need circuits for a peak detector to demodulate the AM, if you post a FAX number I will send them to you.

 

[jsolar]

What are your accuracy requirements? I would guess that your detector offset is your largest source of error. If this is not within a closed loop correction, and you have to know the correct DC voltage and not just the changes, you might want to consider offset amelioration.

The Unitrode takes care of your osc., you still have to deal with that with the AD part, although I heard the Wein bridge is nice for audio op amp work.
file:///C:/Documents%20and%20Settings/solarjo2/Local%20Settings/Temporary%20Internet%20Files/Content.IE5/J03TPQYZ/274,16,Results of Diode Network

You might try DC restoration, similar to setting the black level on old fashioned TV. See "Kiver, Milton S. Third edition Television Simplified" D. Van Nostrand Company, Inc. new York, 1950". Now that book I do have in my hands right now. Don't tell me you have that also?!!

You mentioned that you have to use a simple peak detector. That sort of eliminates solving the DC offset problem. So I hope you do not care about the arbitrary offset constant you will have.

If you do have to know the actual DC level, let us know!

Best Regards,
John Solar

 

[MRIguy]

When I get to that part , I'll post it. Thanks sreid

I do not need absolute DC readings. Just the change over time.

I need to get the info out and locally stored so AM was the choice for simple implementation and remote transfer.

I can pretty much dial out the DC at the load cell, so the change is all that needs to be analyzed.

Back porch clamp! Ah yes, I got started in CCD video. So I know what you're suggesting, but as I mentioned DC errors won't be a big issue and I can cal them out. Thanks John

I'm trying to avoid a full blown digital solution (where my digital buddies are leaning), A/D with clocks, latches, S/H all that yuck. I'll be storing the AM on video tape for later analysis. So simple is the key word for this one time test fixture.

So I'll give the ole' multiplier a shot and see how much trouble I get into.

I'll be back if I mess up.

Death before digital!

 

[jsolar]

MRIguy, It sounds like you have the right choice, not overdesigned, not under. Very good!

 

[cbarn24050]

Hi MRIguy, your right there is a simple diff amp method. A long tail pair. Feed your "rf" to the ltp base and the modulating voltage to the tail transistor.

 

[Warpspeed]

A basic two quadrant multiplier chip will already have exactly that, and the transistors will be better matched than you can possibly do yourself with discretes.