1
\$\begingroup\$

I'm coming to this with a Physics background. I work in solar cells. I would like to flash the solar cell with a laser and measure the voltage response. The University has a spare 10 GSPS (gigasamples per second) data acquisition system that I can use. Ideally, I'd like to measure up to a second - but with particular interest in the ns range.

My question is - *how I should pick an amplifier to amplify the signal from the sample in these time ranges *.

I understand I need to pick an amplifier with input impedance as close as possible to the source pick up. I'm not sure yet of what the output impedance of the sample will be.

Say, if it were 50 ohms output impedance, would something like this, an amplifier block with quoted specifications of

  • wideband, 10 to 2500 MHz
  • high IP3, 35 dBm typ.
  • low noise, 5.5 dB typ.

be reasonable? What else should I consider when making a purchasing decision for these time ranges?

I don't think any single amplifier will cover the range I want (1 Hz -1 GHz), so that suggests what would be the easiest way to switch amplifiers for different time ranges? I imagine even a switch could generate electrical noise in these nanosecond time scales.

Any background reading would be incredibly appreciated.

\$\endgroup\$
10
  • 1
    \$\begingroup\$ How large of a signal are you expecting this to be? A few volts or a few microvolts or somewhere in between? I would think you could just use an oscilloscope with deep sample memory for it. \$\endgroup\$
    – Hearth
    Commented Dec 26, 2020 at 18:44
  • \$\begingroup\$ What's the size of the cell? Do you know what its capacitance is? \$\endgroup\$
    – The Photon
    Commented Dec 26, 2020 at 18:55
  • \$\begingroup\$ How much illumination are you going to hit the cells with? What kind of open-circuit voltage are you expecting? Maybe the 50 Ohm input impedance of the sampling system is acceptable and you don't need an amplifier? Just a thought. Also, 10 Gs/sec means 10 gigabytes per second (if you have an 8 bit ADC). So sampling for 1 second would be... interesting. \$\endgroup\$
    – user57037
    Commented Dec 26, 2020 at 18:58
  • 1
    \$\begingroup\$ Yes, but you will need to place it within a few mm of the electrodes of the cell. But if your cell has 1 uF capacitance or more I am not clear why you think it will produce any useful signal at 1 GHz. You should capture everything useful about the signal produced by this cell if you measure with 1 MHz bandwidth, or even 100 kHz. \$\endgroup\$
    – The Photon
    Commented Dec 26, 2020 at 19:18
  • 2
    \$\begingroup\$ @Tomi I think you'll need to describe as much as you know, a priori, about the laser pulse itself as well as intimate details about the purpose/intent of the observations you hope to make. There must already be some theory to apply that you hope predicts the outcome, as well. You should share that, too. \$\endgroup\$
    – jonk
    Commented Dec 26, 2020 at 22:29

1 Answer 1

2
\$\begingroup\$

The world of amplifiers is vast and complex. I have had experience with a similar problem about 2 years ago, where I wanted to measure the impulse response of a fast photodetector (also in a research setting at university).

  • If you are using a 10GSPS acquisition system, then you must band-limit the analog signal to at most 5GHz to avoid aliasing. It's unlikely you were going to go higher, but it's something to keep in the back of your head.
  • At these frequencies, reflections of transmission lines usually become important in a measuring setup, especially if you plan on preserving the pulse shapes as much as possible. Impedance matching becomes important at all stages.
  • As you may have seen in the datasheets on Mini-Circuits, a fixed gain block will probably give you the best performance for wide bandwidths (I have used gain blocks from Mini-Circuits myself and they worked reasonably well for my setup). However, keep in mind that they assume both a \$50\Omega\$ input and output impedance to avoid reflections over \$50\Omega\$ SMA cables. This can pose some restrictions for you sample (eg. biasing).
  • If you need to strictly keep the bias voltage of the solar cell constant, then a transimpedance amplifier (TIA) as close to the sample as possible is in my eyes your only option.
  • If you're planning to use the fixed gain block as a transimpedance amplifier, the TIA gain will then at most be \$50\Omega\$ due to the termination resistance (and this TIA gain is usually considered low). The combination of high-gain, wide-band, low-noise and stability is notoriously difficult for transimpedance amplifiers. It is not just "use a fast opamp and put a resistor in the feedback loop".
  • It sounds to me that you're going to use a pulsed laser (I may be wrong), in which case you are in a position to average over many periods to drive down the noise. I'd say, use that to your advantage if possible. Note that averaging does not get rid of transmission line reflections, nonlinear behavior, aliasing, etc.!

At least, this is how I would summarize possible pitfalls of what I think you're trying to achieve. I've by now finished my PhD, but I remember the annoyance of trying to measure such a fast signal very clearly.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.