# 660 Mbits per second photo-diode amplifier on 3.3V rail

I've got a 10mW laser diode that can transmit at 660 Mbits per second, I've got the laser driver circuit that can do this, I've got a photodiode from Hamamatsu that looks good for this speed (-3 dB ~1 GHz) but I'm struggling to design the photodiode preamplifier.

In my innocence I thought I'd find an op-amp that will do this. I've seen a document posted by Spehro in another answer (question wasn't totally relevant) that gave me some clues and, I came up with a circuit that kind of works (in simulation) based on those ideas but I'm pushing my own limitations now and I don't feel confident so I need advice on the best approach to take. What I have come up with so far is

Am I missing some fundamental approach that could pay better dividends?

Hamamatsu are profoundly silent on the way they coupled photodiodes with a transconductance amp despite the device I'm trying to emulate being discontinued and Hamamatsu say they are moving away from this wavelength photodiodes and lasers. I'm trying to emulate that Hamamatsu device from their now apparently defunct range of 1.3$\mu$m products. (G10476)

My question is am I taking a reasonable approach?

• Note the maxim part has a low-frequency cut-off much higher than you had from Hamamatsu. The Kyosemi part doesn't spec a low-frequency cut-off, but they call it an OC-12 product which implies a lower low-frequency cut-off than Fibre Channel (which Maxim is spec'ed for). – The Photon May 28 '14 at 22:57
• Converted my comments to an answer. – The Photon May 28 '14 at 23:20
• Have you looked at figure 1-31 in chapter 2 of the Hamamastsu opto-semiconductor handbook? hamamatsu.com/resources/pdf/ssd/e02_handbook_si_photodiode.pdf – FL_Engineer May 29 '14 at 0:37
• @FL_Engineer - yes I've seen that diagram but unfortunately it runs from +/-5V supplies but I'm not ruling out that I can't get +/-5V from the incoming 3V3. I'm not all that convinced about the high-frequency performance though. – Andy aka May 29 '14 at 7:11

## 2 Answers

I think you will be able to find a close replacement for your original Hamamatsu part, if your requirements aren't too strict.

A google search for "InGaAs TO-46" will turn up as many parts with integrated TIAs as without.

A couple examples I found with a quick search are at OSI Optoelectronics and Kyosemi.

If you do want to use a separate amplifier, there are chips designed strictly as photodiode TIAs for optical communications applications. For example, Maxim has a couple of options.

• Cheers for that - I had contacted Kyosemi a few months back (but got no reply) on this subject but, reinvigorated by your answer I shall try again. The OSI link is a plain photodiode but I shall go to their website to uncover anything useful. – Andy aka May 29 '14 at 7:37
• @Andyaka, if you can deal with a different optical interface (fiber pigtail or receptacle) there are a bunch of other options. Maybe better to discuss in chat? – The Photon May 29 '14 at 16:39

Depending on exactly what you want, and how you're willing to get there, it's dead easy to find 1310 nM (You wrote 1.3nm but linked to 1.3um parts - aka 1300 nm) FAST parts cheap in the form of surplus 4Gb fiberchannel singlemode SFPs - last I looked still available for less than \$5 shipped on sleaze-bay if your search-fu is excellent, more money if you miss the poorly described cheap ones (longwave was a good keyword for those last I looked.)

I'll admit that I have not actually cannibalized any for parts, but I am running a gigabit ethernet network with them.