I need to make a non linear *DC only* amplifier. It only has to handle low voltage and current - an input range of about 25 - 300 mV at negligible current, and the output to be approx 2:1 on an exponential curve - ie with an input of 25 mV the output would be 25, 50 100, 100 400 etc. It's for instrumentation purposes, but doesn't need to be *that* accurate, just reasonably stable. I've got both +/-15 volts and a 5 volt rail available.
Now if it were a compressor, I've got circuits for those...
Have you tried asking in sci.electronics.design? But then you'll probably get answers on how to do just about everything but... :)
Lee
From the question, I think you mean that you want an antilogarithic amplifier. I'd suggest something designed around a tl081 etc fet opamp with a diode as the log element. Try looking up log amps in Google for more detail.
Hope this helps.
Regards Capitol
I don't know where to get one, but are you sure you want an exponential function? From the figures quoted, it looks like the output needs to be the square of the input (times a constant).
i.e. for an input of 25x the output needs to be 25x^2
I'm no electronics expert, but maybe there's a standard chip which does this - or maybe one whose output is derived by multiplying 2 inputs, in which case you could feed the same signal (with buffering if necessary) into both inputs.
Sounds like a trip to Maplins or RS Components could be on the cards.
[I may be talking rubbish. Doubtless IMM will put me right if necessary!]
Actually its not that hard.
If you use two op amps and two matched diodes, it should work.
Buffer the input to go low impedance using one op amp.
Now connect the +ve input of the second to ground via one diode, and bias the diode forward with a few mA from e.g. -ve supply.
Now use diode in series between input buffer output, and normal shunt feedback opamp, as second part. Diode current to voltage ratio is essentially exponential so you will get what you want. The second diode will absorb most of the temperature variation. Out put volts will be of the form exp input volts.
Yes. Don't need FET input, but its nice.
Another way I have done it is by means of a series of diodes and resistors in a ladder network, using stepwise approximation...this is more stable, but not such a smooth curve and more complex to calculate.
Dave, Try something like this as a starting point.
HTH
Bob
I've every confidence in this group. ;-)
At least that never happens here. Perhaps twin combis are the answer to my problem?
Brilliant. I had an idea diodes were the way to go, but wasn't sure how to implement this. I'll knock something up tomorrow and report back.
I was hoping there might be something available.
I've looked at the Maplin cat, but nothing suitable. RS have a vast range, but it's knowing what to look for.
How about an A/D converter followed by an EPROM lookup table (for the conversion function) and a D/A converter for the output.
All bog standard bits and no "funny" analog stuff, if you d> I need to make a non linear *DC only* amplifier. It only has to
handle low
negligible
available.
Sadly, twin opamps. And twin diodes.
I know it may sound a silly solution (and a bit over the top for this design) but it's the principle behind an awful lot of digital signal processing!!
If you need to perform an unusual function on an analogue signal this is more often than not the way to go!! However in this case you probably just need a couple of op-amps and diodes.
Opamps were made specifically for this. Operational amplifiers were conceived in order to produce analogue mathematical functions, or operations - exactly what youre doing.
If you dont need full rail to rail swing above 6kHz, LM324 often is the way to go. Its a quad opamp, which gives you lost to play with, for a price thats very close to free.
Regards, NT
If you have a logarithmic compressor it can be used in the feedback path of an opamp summing integrator to generate the antilog function. Sketched below. View with a fixed-width font such as courier.
R1 C Vin-----/\/\---+-----||-----+------->Vout | __ | __________ +----|- \ | | | | |OP >---+--|Log device|--+ | +-|+_/ |__________| | | |0v | | | +----/\/\---------------
Just as long as they are more than 100% efficient :-)
Thanks, Tony. I meant it as a bit of a joke, since compressors are used in audio. But these are pretty complicated affairs, as they aim to keep distortion to a minimum. I'm only dealing with DC, so to use an audio compressor circuit to provide the gain control on an opamp would be rather OTT, cost wise.
Compressors are also not what you want anyway. They are timne variable things, and basically pertty orrible.
Had a go at this, but without success. Using 1N4148. One problem is that biasing the diode to a couple of mA from the neg rail puts a large offset on the output, being DC coupled - even with the feedback set to give unity gain.
Huh>? should have been relatively balanced.
I'll mail you direct - if you can post me the cqt diag i'll take a look
HomeOwnersHub website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.