Photo diode receiver modifications

Put a shroud around it? Mike

If you have a lot of light, namely noise isn't a big problem, you could AC couple the signal such that ambient light doesn't saturate the opamp, or even such that the opamp doesn't see it at all.

The pin diode could drive a resistor to ground or V- (double the bias!) with that node capacitively coupled into the TIA. Or you could dump the pin diode current into a paralleled inductor+resistor to ground, and then use a voltage amplifier, not a TIA. That would be stable and probably faster.

Now Phil will tell us how to do it *right*

John

Why did you use that opamp? It says it stable for gains greater than

There's a fast FET opamp from Analog. ADA4817 that might work better.

rt

What is high ambient light? indoors or outdoors? What's the laser power and how much current are you seeing...

Can you make the pulses longer? What is the bandwidth of the current design. Seems like you'd like something around 20 MHz or more to catch the 50 ns pulse.

A shroud is always a good idea as someone suggested.

George H.

He doesn't need the bandwidth - he might get away with a TL071

AC couple thru a DC restorer, as I've shown here recently. ...Jim Thompson

Instruments

was

high

short

To answer a few questions.

- Why did I use an OPA847? Basically, because it was avaiable. I started out with a (good old) LF157, and that didn't work.

- Put a shroud around it It is installed in a small metal case, without the right case, it doesn't work...

- If you have a lot of light then noise isn't a big problem The pulses are coming from a 20-100W IR pulse laser with optics to focus the beam better. When I connect the SFA205F photo diode directly to my scope, using a 9V battery for reverse biassing the signal is even visable.

- High ambiant? This is used outdoors mostly in (bright) daylight

I cannot make the pulses longer, so the amplifier has to work good at

--------------------------------------- Posted through

Joost mentioned:

I inherited a similar sounding circuit once. I believe the designer used an OPA847 because on paper, it has very low voltage noise. However, it was always right on the edge of stability. By simply changing the amp to a Texas OPA656, the circuit became very stable indeed, with no apparent increase in noise. I think the relevant characteristics were: OPA656 - bandwidth around 230MHz; OPA847 - 1.6GHz OPA656 - much lower Cin, so easier to make stable OPA656 - JFET input so low current noise. OPA847 - low voltage noise but relatively high current noise, which was probably more relevant IIRC.

As someone mentioned, the ADA4817-1 is good too, it is like the OPA656 except in some characteristics where it is better. Its pinout is sort of compatible with the standard one used by the OPA656, so with some careful layout you can have a dual-sourced IC.

Don't use the OPA657. Looks good on paper (higher bandwidth) but has higher Cin than its sibling the OPA656, so doesn't usually work any better, and is trickier to stabilise.

Does 'low frequency light' mean the colour (infrared, etc) or doe sit means pulse trains below say 10kHz? The former problem is usually solved by using filters over the photodiode (you can get cheap bad ones for tens of cents). The latter would be solved by adding a high pass filter to the output of the preamp.

With the 'scope set for 50 ohm input? You can always try just a resistor to ground and a voltage amplifer after that. There is a trick with a "T-coil" that can gain you a factor of two or four (?) in bandwidth. Phil H. talks about it in his book. I've always wanted to try it. Have you read Phil's book? There might even be a link on his web site to the T-coil idea.

George H.

to

=A0

We sometimes do the simpler version: V+, pin PD, resistor+series inductor to ground. Then a fast voltage amp off the PD:resistor junction. Now you can use an outrageously fast, like 2 GHz, amplifier and not worry about TIA type issues. The inductor peaks things up a bit. The resistor noise is now in your face, so that's the price of simplicity.

I guess you could dump a PIN diode directly into a darlington-type MMIC amplifier input. That sort of makes a fast, stable TIA for free. The DC performance would be terrible, but you could sense the current into the other end of the pin diode and use that to correct.

I suppose you could use the high-speed current signal off *both* ends of a photodiode, into dual TIAs or MMICs, and subtract their outputs. That halfway fixes the DC error of the MMICs, too. Gotta try that one day.

The next step would be...

John

. I

ry

less

n't

us

to

is

Yeah I think ye ole capacitance is an issue too. 10 pF can ruin your day.

But the only way I've done any thing fast is just like that. (I never tried an inductor) and then used a mini-circuits amp. (cheating?)

I don't know MMIC amps. Do they sense current?

Yeah I had that (or similar) idea. You've still got to reverse bias the photodiode. Which you can do by biasing the (+) input of on of the TIA's.. then you've got your ouputs at differernt DC levels... but Phil H. posted some circuit to take care of the DC offset... I can't remember what it looked like. Too many beers.

Sell it and design something else!

George H.

I mean something like an ERA-1, which you are probably familiar with. They are usually an InP darlington transistor with a c-b feedback resistor, practically a TIA already. They are characterized for wideband RF usage, with a nominal (nominal!) 50 ohm match on input and output, and are usually unconditionally stable. They work nicely as pulse amplifiers, but their DC stability is bad.

Sorry, I can't understand those words "too many beers."

I was cooking and I asked my wife (who is half Italian) whether I was using too much garlic. She replied "Sorry, I can't understand those words..."

Actually, I had a really cool idea for the next step. But I forgot it.

John

ode. I

r in

very

to

he less

te

ould

LF157,

oesn't

focus

a 9V

ed to

n his

ERA-1, I don't know it at all, but I've got goggle. I've only been doing 'real'* electronics for 10 years...(Wow it doesn't seem that long.) And most of that is mucking about at or near DC.

Hah, Yup never to much garlic or onions! All my favorite dishes start by carmalizing as many vadalhas(sp) as I can fit in my frying pan.

Oh, I was thinking that instead of adding the two signals you could amplify and then multiply. The amplifier noise should go down by a factor of two. (And more if you averaged.) I think my new Rigol DSO has a multiply function.

George H.