Voltage-controlled amp design is hard

Can'y you adjust the APD gain with the APD voltage the same way you adjust the PMT gain with the PMY voltage?

--
Uwe Bonnes                bon@elektron.ikp.physik.tu-darmstadt.de 

Institut fuer Kernphysik  Schlossgartenstrasse 9  64289 Darmstadt 
--------- Tel. 06151 1623569 ------- Fax. 06151 1623305 ---------

Thanks, Win.

I've looked at those several times, but never found a use for one, mainly because the transconductance linearity is the pits except at high current (OPA860 datasheet, Fig. 16).

To make a VCA out of an OTA, you put the signal into the bias current input and use the normal inputs as the gain control. Using a highly degenerated diff pair and two OTAs gets you twice the signal and nulls out the DC bias.

Cheers

Phil

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Yes, but not over the required 100X range, and not without degrading the noise very badly. The APD is running at a nominal gain of 20, which (for noise purposes) effectively makes the TIA feedback resistor look

400 times larger.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

That works great unless the diode is a reachthrough device, which doesn't conduct at all below some fairly large bias.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

Looks like you have to write in and send box tops to get one, though. The Skype contact is a hotmail address. ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC 
Optics, Electro-optics, Photonics, Analog Electronics 

160 North State Road #203 
Briarcliff Manor NY 10510 

hobbs at electrooptical dot net 
http://electrooptical.net

You could cascade the two sections with 20 dB range for each and even afford some fixed gain in between, since for the 2nd section the life is already 20 dB easier than for the first.

But the emitters are fed from the modulated tail current source and even when you change the current distribution of the pair, their emitters are still low-impedance in comparison. That is not very different from a cascode, and I have never seen a cascode transistor add significant noise.

Even if the shot currents of the pair do not correlate, the signal has already seen some amplification from the input to the emitter split.

I have a 100 MHz crystal osc on the back burner where I want to do that with 2 JFETs in a Driscoll style circuit. I want to avoid the usual Schottky diode pair as a limiter because it will upconvert the

1/f noise to 100 MHz. I need only a few dB of gain variation and the JFET's depletion makes the gate bias cheap. Still untested.

regards, Gerhard

ver to

h

a gain

ise

the

ise

ut a

off the

he

t

re 16dB

's

wer

y o

log

Hi piglet, the issue here is that it's an APD. (Avalanche photodiode) You can run this open, but then loose the gain of the avalanche part.

GH

Hey Phil, can I asked a related question? So this is replacing a PMT run in current mode (not pulse counting) (I assume). I've used PMT's in current mode, but have never looked at the noise. One assumes it looks like the shot noise times the tube gain squared. Is that right? And if so, you are hoping to make a substantial improvement in the noise performance with the APD.

George H. (curious minds want to know :^)

Thanks, Gerhard.

"Ah, the old laser noise canceller trick, eh, 99?" ;)

When you do that, if the tail current has full shot noise, then in the high-beta limit both collector currents have full shot noise also, regardless of the splitting ratio.

It's the splitting that adds the noise, and you can't degenerate the diff pair or the signal will split nearly 50:50 regardless of the DC splitting ratio. Cascodes have no current splitting and hence no additional noise to speak of. On the other hand, that circuit gets only one copy of the shot noise instead of several, so it ought to be better, especially with an extra diode-connected transistor in each arm.

At low gain it still suffers from the attenuation problem, so it would need to be done in a few stages, which starts to raise the parts count. I'll think it about it some more.

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Yup, with two caveats. First, you have to take into account the lower quantum efficiency, at least if you don't use one of those TIR coupling techniques we've talked about in the past, and second, in a PMT _all_ the secondary electrons are stochastic, whereas in an APD the original photoelectron is deterministic (once it's been detected, that is), and only the avalanche electrons are stochastic.

That means that if the first dynode has an average secondary yield of N electrons, the PMT adds sqrt(N) noise electrons per detected photon. In an APD with a gain N, it only adds sqrt(N-1) noise electrons (neglecting excess noise).

Thus a PMT with a gain of 1 would have a SNR far worse than the shot noise, whereas an APD just turns into a normal (potentially shot noise limited) photodiode.

In practice the difference isn't important until the secondary yield at the first dynode drops below about 3. It's often worth running that stage at higher bias and doing gain control further down the string. That also improves the tube's linearity by reducing space charge effects in the PC-D1 space.

And if so, you are hoping to make

Mainly they want it to be cheaper and more compact without reducing the overall performance. A super-quiet front end plus an APD running at only moderate gain was the ticket.

The motto of the National Curiouser (and curiouser) ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

They got a $2G minimum order with a $100 handling fee. :|

Here's a US distributor that sells them for $2.99 each in quantities of

4 or more, probably less for 50+:

Thanks George. Forget about the Avalanche part.

piglet

Okay, box tops and a Brinks truck. Picky picky. ;_

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

I can just picture how BigBiomedCorp would react. "Hey guys! Remember that audio VCA I told you about? There's a second source, some guy in a shed out in the woods of North Carolina, near where Judge Crater was buried!

(crickets)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Phil,

The two classic VCA design books are

Nonlinear Circuits Handbook, Daniel E. Sheingold Editor, Analog Devices, Inc. Publisher. I like the 1976 Printing. ISBN 0-916550-01-X

Function Circuits, Design and Applications, Yu Gen Watt and William E. Ott, The Burr Brown Electronics Series, Second Edition 1976, ISBN 0-07-071570-X

Beware of "Dumbed Down" versions in later issues.

You might find a workable old topology in the second book.

I have two copies of the Burr Brown book, but the first book will only leave here pried from my cold, dead, hands... :-)

Abebooks is your friend.

Steve

Thanks, Steve.

I have the Sheingold book, but not the Watt&Ott one. I bought Sheingold probably eight or ten years ago, but there wasn't a lot of new news there--I picked up most of that stuff either on my own or from National app notes BITD. (Of course some of them may well have been indebted to Sheingold.)

I found a copy of Watt&Ott for a few bucks online, so we'll see. As the old saying goes, "a couple of weeks in the lab can save you an afternoon in the library." ;)

Cheers

Phil Hobbs

--
Dr Philip C D Hobbs 
Principal Consultant 
ElectroOptical Innovations LLC / Hobbs ElectroOptics 
Optics, Electro-optics, Photonics, Analog Electronics 
Briarcliff Manor NY 10510 

http://electrooptical.net 
http://hobbs-eo.com

Watt and Ott go deep into other multiplier topologies as well as a deep coverage of error sources. I doubt there is much new under the sun, but it is a good read from before microcontrollers.

Steve

Available at

couldn't find any online