Electronics > Projects, Designs, and Technical Stuff

ADAR7251 AFE for Photodiode Data Acquisition - Too good to be true?

(1/1)

mattico:
I ran across this ADAR7251 AFE and it seems too good to be true so I had to run it by you guys.

It's a 16-bit 1.8MSPS Quad channel Sigma-Delta ADC with LNA, PGA, Equalizer, and on-chip Reference. It costs less than my single channel ADS1602. The built-in amplifiers mean you don't have to worry about ADC drivers and can possibly remove a gain stage from your design. Sig-Delt with decimation filter means you don't need a high order active antialiasing filter. It's got 2 aux ADCs for monitoring temperature or whatever. It's got parallel out or single/dual serial out with a SPI control interface. Of course all that integration is the whole point of an AFE but this is the first one I found which seems generically useful rather than being limited to a single application.

The only issues I could think of are:
1. 0.5 db Channel-Channel gain mismatch is meh but it should be easy to cal out unless it changes a lot with time or temperature
2. Application notes focus on radar and aren't much help for general applications
3. 65dB PSRR is lower than some alternatives
4. Dual-SPI data + SPI control is annoying

A bit about the application. Trying to accurately measure the width and average value of ~40us light pulses at about 10nA photodiode current. So looking for about 500kHz bandwidth, high transimpedance gain and low noise. The attached schematic shows the basic idea. Hopefully the optics can work with the smaller area photodiode (~2.5pF) but if not might have to live with more like 60pF of diode capacitance, hence the tuning cap. The RC network is from the datasheet to create a bandpass, the ground signal is AC coupled as suggested by Analog Devices for single-ended signals.

Please nitpick away. I've not much experience in precision analog stuff so I'm hoping the AFE let's me get away with this extremely simple design.

moffy:
Looks like an interesting chip but not particularly low noise based on the FFT graphs. If I was trying to determine the average power of the pulse I would be tempted to integrate over the pulse period then divide the final output by that period. It is the definition of average. Increases accuracy and reduces your overall data rate. :)

awallin:
- negative bias usually helps to reduce the photodiode capacitance (but gives worse dark current)
- 1.8MSPS seems like a low sample-rate for a ~us pulse?
- check if your signal only swings one way, and if the ADC has bipolar input - are you using the full 16-bit range of the ADC?

d-smes:
That 470 / 220 pF capacitor network on the inputs to the ADC will attenuate your pulse somewhat (acts as voltage divider).  Be sure and factor that into your gain calculation.

Navigation

[0] Message Index

There was an error while thanking
Thanking...
Go to full version
Powered by SMFPacks Advanced Attachments Uploader Mod