LC tuned circuit using parasitic C's (photodiode preamp)

[Gunchars]

Hello, guys and girls!

There is something I could use your help with. See, there is this project I am working on that requires me to detect a very weak optical signal (20-30 MHz, sine). So I was thinking of using a photodiode with a tuned LC circuit like the old day radio guys would and this is what I have come up with so far. Schematic is attached. The gray capacitors aren't really there. They are parasitic components for photodiode and opamp respectively.

From what I've read, for this purpose I need as big as Q as I can get. To do that for a parallel tank circuit the inductance has to be as big as possible and therefore capacitance as small as possible. So why not use parasitics from the photodiode and input of the op amp ?

This works like a charm in Multisim, but that's all theoretical and I'd like some more confidence in this idea before I order any parts.

Now, this is all theoretical as I have never made a single working LC circuit and odds are I could be missing something obvious. Like those capacitances doesn't really work like that. Or they aren't too stable. Or maybe the losses are too big for this to be considered.. I don't know. Any feedback would be appreciated.

Many thanks!

G.

 

[mneary]

I think you're right, the parasitic capacitance will be too small and not stable enough. Parasitics don't always have the Q you would want, either.

Several separate topics:

Inductor Q---In practical inductors, a higher inductance does not mean better Q. With increasing number of turns, a smaller gauge wire is usually used, leading to a higher resistance. There's a balance to be struck

LC disadvantage----But I don't usually see an LC directly on a photodiode. That's probably because the resulting physical dimensions of such a circuit would be too susceptible to E/M fields.

Diode load---The output variable of a photodiode is current, not voltage. Converting to voltage too soon (such as right on the diode junction) invites interference and parasitic loads. Normally the first thing I see on a photodiode is a transconductance amplifier. That way the diode junction sees no voltage and capacitance essentially disappears from the frequency response equation.

Do you really want high Q---When you say "20 to 30 MHz" do you mean ONE frequency within that range, or to be sensitive to ANY frequency within the range? If the latter, high Q wouldn't be what you want anyways. A Q of 100 at 25MHz would leave you with a bandwidth of only 250 kHz.

 

[Gunchars]

Thank you for your reply. I checked couple of inductor manufacturers datasheets and you're right - bigger inductance doesn't mean bigger Q. I looked at nanohenries range and any increase in inductance is countered by almost proportional increase in resistance. Also I couldn't find any inductor with bigger Q than 30 for my frequencies of interest.

I still think that using high Q tuned circuit right after photodiode is the way to go because it gets rid of DC bias, 50hz fluorescent lightning and keeps the bandwidth narrow from the start. Less noise that way, I think. And less chance of saturation.

Okay, so using parasitics is out of the question. What if I add a cap that's 50-100 larger than parasitics in parallel and reduce inductor accordingly? That let's me assume parasitics are insignificant, no ?

Also I didn't quite understood what you said about photodiode's current to voltage too soon and how that invites trouble. Could you explain a bit ?

And when I say frequency from 20-30 MHz, I mean one, entirely within my choosing. Too many things change from that so I'd rather not choose one just now.

 

[mneary]

Capacitor size can't be too large; you have to keep the inductor from becoming too tiny. (There are parasitic inductances, too.)

Well, hopefully I've been helpful. Keep us posted.