Input impedance into amplifier - mismatch

You'll pay a considerable Johnson-noise penalty if you stick with low 50-ohm current-to-voltage signal impedances. Your 200MHz laser application may be a good fit for Phil Hobbs' common-base amplifier approach. Explore his web pages,

Properly implemented, a common-base amplifier allows you to present a low or 50-ohm impedance to high-frequency PD signals, while still developing the signal output voltage across higher resistance values, reducing the effect of Johnson noise, which goes as sqrt(4kT/R).

--
 Thanks,
    - Win

I worked on a very similar project 2 years ago. We had an array of heterodyne detectors measuring the beat frequency of a LASER and it's time-delayed light. We were sampling the heterodyne signal at 2 GHz.

Each of the 16 detectors had their own individual preamplifier, the inputs of which were matched to the detectors (high impedance). The outputs of the preamplifiers were 50 Ohms, so then the rest of the system was 50 Ohms. The gain of the preamplifiers was very high (on the order of 1000) and the noise figure was impressive for room-temp operation.

Unfortunately I don't recall the manufacturer of the preamps, and I no longer have access to that info.

Try Transimpedance amplifiers. That is what is used to match photo diodes to first amps, current to voltage conversion mostly.

It all depends on the photocurrent and the noise figure of the amplifiers, but generally speaking, you're right that 50-ohm amps are a poor choice.

What is the optical power level at the photodiodes?

Cheers,

Phil Hobbs

I have 3mW optical power at the minute, leading to photocurrents of around 1-2mA.

I have researched into transimpedance amplifiers for the photodiode, but the problem is finding op-amps which are relatively cheap, non-surface mount (I would prefer to avoid the cost/inconvinience of producing PCBs) and of sufficient bandwidth.

At the minute, careful photodiode selection has meant noise is fairly minimal (I'm quite surprised for what is essentially a simple resistor, current/voltage conversion) and should not infringe too much on my intended application.

Cheers

Sam

Can you tell us what is actually the problem with your current solution?

If you want the best signal to noise ratio, you will need to use a transimpedance amp since you can design its feedback network to match optimally to the photodiode impedance.

Beware that if you want to do anything serious at 200MHz frequencies, you MUST design a pcb to keep the parasitic capacitances and inductances low.

By the way you are lucky to receive so much light. :-)

A very useful rule is that shot noise dominates thermal noise when

2*e*I_dc > 4kT/R, i.e. when V_dc == I_dc*R > 2kT/e, 50 mV at room temperature. Thus with a room-temperature resistive load, or in your case an RF amp with a 3-dB noise figure, a photocurrent larger than 1 mA will put you in the shot noise limit.

Noise temperature in RF amps is a very useful parameter when the source is not impedance-matched, as photodiodes are not. Unlike the situation in op amp TIAs, the noise temperature is not closely related to the temperature of the resistor itself. You can get RF amps with noise temperatures down to 35K at room temperature, or 10K at nitrogen temperature (77K). Cooled varactor parametric amplifiers can go lower still. This noise temperature is what you plug into the formula above, so with a cooled amplifier, you could be in the shot noise limit with only (2k/eR)*10K = 85 uA of photocurrent, even at 50 ohms.

Thus if your RF amps are ordinary common-or-garden ones with NF~3dB, you're golden.

Cheers,

Phil Hobbs

Unless his photodetector is unhappy with so much light, and his 1mA exceeds the recommended maximum current-density for a spot of light. Just something that should be checked at high light levels.

--
 Thanks,
    - Win