Dual-Channel PCB Design with Signal Conditioning - Seeking Input

[laurence-diack-pk]

I'm looking into designing a combined PCB that implements both the alpha and electron detector variants on a single board, with added signal conditioning for digital processing. I'm still waiting on my parts to arrive for the basic circuit so I don't have any hands-on experience with it yet - just spitballing ideas at the moment

Design Goals:

• Both detector circuits on one PCB (different component populations)
• Inverting op-amp conditioning to handle negative pulses and normalize signal levels
• Outputs to headphone jack (stereo - one per channel) + onboard I2S ADC
• Maximize ADC resolution by matching channel amplitudes
• MCU (rp2040, esp32 or similar) to read from ADC, do some basic processing / data storage
• Perhaps a little oled or LCD display

Technical Questions:

• Gain Scaling: Looking at the expected outputs on the troubleshooting page, should I target 40-50x relative gain difference between the electron and alpha circuits? Planning adjustable trimmers for fine-tuning. I would be DC biasing them up to 1/2 of what the ADC is expecting for a single ended input.
• Input Impedance: From what I gather, 1MΩ+ is preferable over typical audio impedances to avoid overshoot?
• Anti-Aliasing: Would it be worth adding LPF, particularly on the electron channel? Target cutoff frequency recommendations? My ADC has programmable DSP, so I could oversample and filter digitally.
• Isolation Strategy: Beyond isolating digital grounds, should I use separate regulators/batteries for each detector channel to minimize crosstalk? I imagine just cuplicating the circuit with the same ground might not be a great idea
• Op-Amp Selection: I was thinking just another TLE2072 for the conditioning stages - appropriate choice?

I'd be interested to hear if anybody had any suggestions / input on this plan, as well as any other suggestions for improvement. I will share progress when I figure out exactly what direction this is all going to take.

Thanks for any insights!

[timbk]

Hey, I tried a few of the things you are listing (integrating with RP2040 and using its ADC, alpha+beta) and it is somewhat documented in [1,2]. I tried sanding down the section of a BPW34 that does not contain the bond wire, but have not been able to make the alpha detection with that to work reliably yet.

Regarding your questions:

• Gain scaling: The pulses will only go in one direction. Therefore I but the biasing to the ADC at roughly 10-20% full scale. This prevents clipping and allows getting the best resolution on the pulses.
• Anti-Aliasing: If you use similar circuits as in the CERN DIY particle detector, they should already provide a first order LPF, so you might be fine without any additional filtering. You might want to add a voltage limiting circuit, depending on your digitizer.
• Isolation strategy: This is a complicated topic and probably depends on your use-case. A common ground plane has worked well for me. From what I've read, splitting a ground plane might not help and even be detrimental. I would just try to place components in a way that minimizes crossing the return currents. OpAmps usually provide a decent level of ripple rejection, so that path is probably less critical if you think about the coupling between channels. More critical is probably to create stable reference voltages (zener diodes and RC filtering have worked well for me). Stabilizing with a linear regulator definitely helps, but comes at the cost of a voltage drop. If you need to get rid of very low frequencies (20 Hz and below), an RC low pass might also be a good option for the low power analog section.
• Op-Amp selection: Most critical is the first OpAmp (input current/noise level, voltage range). My impression is that the later amplifier is significantly less critical.

Having a stable power supply is critical here. Powering with a battery is the simple way; if you use an MCU and power it through usb then you will need to be careful. I believe the output amplitude also scales with the bias voltage on the diode, so reduced voltage is disadvantageous.

If you are documenting your project, I'd love to hear where to find it!

[1] https://github.com/timbk/BetaBoard
[2] https://hackaday.io/project/196766-betaboard-tiny-particle-physics-lab

[laurence-diack-pk]

Thanks so much for the detailed response! Your project documentation looks incredibly thorough and well thought out - the PCB shielding design and overall form factor are really impressive. The integration with RP2040 and the comprehensive approach to power management/isolation is exactly the kind of systematic thinking I was hoping to learn from. I also see that the MCU ADCs would be more than sufficient for the task and that the bit depth of an audio ADC doesn't really add much in this case.

Would you be okay if I used your design as the basis for the modifications I'm planning? I'm particularly interested in adding features like SD card logging, OLED display, and similar enhancements, and your architecture seems like an excellent foundation to build upon.
Your point about biasing at 10-20% full scale makes perfect sense - much better approach than trying to center everything. And good to know about the existing LPF characteristics, that simplifies things considerably.

Quick question: do you think it would be feasible to implement the original alpha detector concept (using a BPX61 diode) with your design framework? I realize the isolation from external particle sources would be tricky, but I've come across some references suggesting gold leaf windows might work for letting alpha particles through while blocking light and other interference.

Really appreciate you sharing your project. I'm still a way off getting my initial boards together but I will definitely document my progress if I get anywhere useful with the modifications.

Thanks again for the insights!

[timbk]

I'm just a physicist and not an expert on mixed signal design, so take all I say with a grain of salt. I just hope I can make you aware of some lessons I've learned ;)
I agree with what you are saying about bit depth, especially if we talk about beta particles where it's mostly about counting. For alpha particle spectra it might be more relevant, but I don't have much experience with that.
Use all you can from the stuff I've documented, that's what it's for. If you have questions feel free to keep asking ;)

As mentioned I don't have a lot of experience with the original BPX61 alpha detector concept. Your idea sounds quite interesting. I guess there will be some scattering and therefore energy loss from the gold foil, but if it's thin enough that might be acceptable. My guess is that it will be primarily the mechanical challenge of mounting the thin gold foil on top of diode housing, especially if there are sharp corners after removing the glas.