Antilog amplifier
..., Tue Jul 18 2006, 12:49AMOk guys, in my internship helping design a terahertz spectrometer my boss wants me to make a circuit that takes the antilog of a signal...
Now back in the good ole days of analog multipliers, everyone ran the signal through a log converter, then added/subtracted, then through an antilog converter to do their math, but now we do things with dsp's and whatnot.
So, I need to way to emulate a icl8048 using parts available today...
From what I have been able to gather, the voltage across the c/e junction of a npn transistor is equivalent to the log of the current across said junction, in theory. So you can put the transistor in the feedback loop of an inverting amplifier to get an antilog amplifier.
But, no one seems to have published any complete antilog designs... I have a nagging feeling that if I use a 2n3904 and lf347 and build this circuit I am not going to get what I want (well the temperature stability seems like it would be horrible at least). I hope to prove myself wrong tomorrow (need to bring in some npn transistors to test out)
I am seriously thinking of building the circuit that is shown in the icl8048 datahseet and hoping for the best...
Am I missing something here? Ideas?
Re: Antilog amplifier
Tom540, Tue Jul 18 2006, 02:44AM
Good Luck. I don't even know what antilog is
[Mod Edit: then use a search engine. Unhelpful posts are rarely appreciated.]
Tom540, Tue Jul 18 2006, 02:44AM
Good Luck. I don't even know what antilog is
[Mod Edit: then use a search engine. Unhelpful posts are rarely appreciated.]
Re: Antilog amplifier
Sulaiman, Tue Jul 18 2006, 10:52AM
With analogue log- and anti-log circuits
the main problem is the dV/dT of the semiconductor junction.
This is why any serious log- or anti-log circuit uses a PAIR of transistors.
Usually a matched-pair in one package.
Look for manufacturers of dual-transistors and you'll find 'how-to' there.
e.g. Analog Devices MAT04
Sulaiman, Tue Jul 18 2006, 10:52AM
With analogue log- and anti-log circuits
the main problem is the dV/dT of the semiconductor junction.
This is why any serious log- or anti-log circuit uses a PAIR of transistors.
Usually a matched-pair in one package.
Look for manufacturers of dual-transistors and you'll find 'how-to' there.
e.g. Analog Devices MAT04
Re: Antilog amplifier
Steve Conner, Tue Jul 18 2006, 10:53AM
It's not true that "No-one has published any designs". The thing you're talking about is what I know as an "Expo converter." See this forum thread:
Though it's from an electronic music viewpoint, it's still the same circuit, and it links to tutorials. The electronic music applications were some of the most demanding in terms of stability, so music circuits should be good. Although, since you're too young to remember analog multipliers, you should be A/D converting, doing the math on a DSP, and D/A converting again.
Oh, and putting the trannie in the feedback loop gives an ordinary log amp. We have circuits like that in our products using a precision op-amp, a matched transistor pair, and a special voodoo temperature compensating resistor. For expo, you want the semiconductor junction in the input.
Steve Conner, Tue Jul 18 2006, 10:53AM
It's not true that "No-one has published any designs". The thing you're talking about is what I know as an "Expo converter." See this forum thread:
Though it's from an electronic music viewpoint, it's still the same circuit, and it links to tutorials. The electronic music applications were some of the most demanding in terms of stability, so music circuits should be good. Although, since you're too young to remember analog multipliers, you should be A/D converting, doing the math on a DSP, and D/A converting again.
Oh, and putting the trannie in the feedback loop gives an ordinary log amp. We have circuits like that in our products using a precision op-amp, a matched transistor pair, and a special voodoo temperature compensating resistor. For expo, you want the semiconductor junction in the input.
Re: Antilog amplifier
HV Enthusiast, Tue Jul 18 2006, 01:36PM
Antilog designs are plentiful on the web. Here are some in this document.
http://www.electronics.dit.ie/staff/ypanarin/Lecture%20Notes/DT021-4/6LogAntiLogAmplifiers.pdf
Search and ye shall find . .
Also, try going to digikey and search for "log" and check out the amplifiers. There are plenty of antilog amplifiers in there. . . some of which are in stock.
HV Enthusiast, Tue Jul 18 2006, 01:36PM
Antilog designs are plentiful on the web. Here are some in this document.
http://www.electronics.dit.ie/staff/ypanarin/Lecture%20Notes/DT021-4/6LogAntiLogAmplifiers.pdf
Search and ye shall find . .
Also, try going to digikey and search for "log" and check out the amplifiers. There are plenty of antilog amplifiers in there. . . some of which are in stock.
Re: Antilog amplifier
..., Tue Jul 18 2006, 10:53PM
Hmm, a quick look in the digikey catalog found two log amps labeled as 'demod' which I am going to pray means 'antilog' (ad640 and 641) which look like they should work (although they are just a tiny bit pricey at $42/chip)
In case you didn't notice the schematic I referred to in my first post was shamelessly stolen from the lecture you posted, which would bring me back to my first question of will any old transistor (matched pair) work?
Thanks for the suggestion to look in the datasheets for the matched transistor for help, never would have thought to look there
I do some searches on the expo converter later today if all else falls through....
Thanks guys!
Edit:
scratch my first idea, upon closer inspection of the ad640 it isn't what I want (in fact it is a very similar chip that is causing the problem).
What would be stopping me from taking a log amp (like the log101 from ti) and running it into the inverting input of an opamp (replacing the transistor in the schematic I posted in the first post)?
EVR, would you mind pointing out one of the plentiful antilog amps sold by digikey?
Edit2:
Ok, I think I have it (or all of this thinking of logs and op amps has driven me insane).
I wire an op amp as a simple voltage follower. But I put a log amp in the feedback loop, such that the amp will find the correct input voltage of the amp to give the output voltage equal to the voltage that I am putting into the whole mess. So then the voltage at the output of the amp will be the antilog of the input voltage (praying smiley).
sorry about the double post for the schematic, I am leaving soon and don't have time to mess with the uploads forum...
Mod: Fixed
[Edit: Sadly the picture was wider than 400 pixels. I thought about making it one pixel wide just to mess with your head but then I decided to set it to exacly 400 just to be kind and helpful.]
..., Tue Jul 18 2006, 10:53PM
Hmm, a quick look in the digikey catalog found two log amps labeled as 'demod' which I am going to pray means 'antilog' (ad640 and 641) which look like they should work (although they are just a tiny bit pricey at $42/chip)
In case you didn't notice the schematic I referred to in my first post was shamelessly stolen from the lecture you posted, which would bring me back to my first question of will any old transistor (matched pair) work?
Thanks for the suggestion to look in the datasheets for the matched transistor for help, never would have thought to look there
I do some searches on the expo converter later today if all else falls through....
Thanks guys!
Edit:
scratch my first idea, upon closer inspection of the ad640 it isn't what I want (in fact it is a very similar chip that is causing the problem).
What would be stopping me from taking a log amp (like the log101 from ti) and running it into the inverting input of an opamp (replacing the transistor in the schematic I posted in the first post)?
EVR, would you mind pointing out one of the plentiful antilog amps sold by digikey?
Edit2:
Ok, I think I have it (or all of this thinking of logs and op amps has driven me insane).
I wire an op amp as a simple voltage follower. But I put a log amp in the feedback loop, such that the amp will find the correct input voltage of the amp to give the output voltage equal to the voltage that I am putting into the whole mess. So then the voltage at the output of the amp will be the antilog of the input voltage (praying smiley).
sorry about the double post for the schematic, I am leaving soon and don't have time to mess with the uploads forum...
Mod: Fixed
[Edit: Sadly the picture was wider than 400 pixels. I thought about making it one pixel wide just to mess with your head but then I decided to set it to exacly 400 just to be kind and helpful.]
Re: Antilog amplifier
..., Mon Jul 31 2006, 07:02PM
update:
Well I tried it out...
If I wire it up as shown I get a comparitor with about 3v of histeresis... Then I tried it the other way and got a nice log-of-a-sign-wave generator at about 24khz
Do any of you guys know what is going on (or if this circuit it doomed to start with?)
..., Mon Jul 31 2006, 07:02PM
update:
Well I tried it out...
If I wire it up as shown I get a comparitor with about 3v of histeresis... Then I tried it the other way and got a nice log-of-a-sign-wave generator at about 24khz
Do any of you guys know what is going on (or if this circuit it doomed to start with?)
Print this page