We use cookies to provide our visitors with an optimal site experience. View our privacy notice and cookie notice to learn more about how we use cookies and how to manage your settings. By proceeding on our website you consent to the use of cookies.
The by-pass capacitor value depends on the application. Low gate count digital circuits usually are fine with 100nF, which is kind of “industry standard”, but on the other hand an audio amp may require millifarad size caps. Using several caps parallel with different values widens the bandwidth. The values shown are just a guess, you may need to tweak them if you observe that the supply voltage is not solid enough (e.g. dips when your APD starts to conduct).
However, be cautious, as the by-pass capacitor and the wire inductance form an LC resonance circuit. Rapid load current changes may induce oscillation on the power supply line. A small series resistor is usually enough to damp the oscillation.
There are many websites that explain by-pass capacitors better than I do. Take a peek:
That is correct. However, I have no idea if it truly works, that’s just my guess of the circuit that you are after. If the ionization current is in nanoamps, as you stated (i.e. there is no avalanche multiplication in the ionization chamber), then it may not be enough to trigger the APD (unless you actually use the photons that are emitted when high energy particles collide in the chamber, to trigger the APD).
Standard practice in electronics labs for the past century is to use two laboratory grade reference power supplies when you need two different voltages for your experiment.
Introducing an extra component with all the extra uncertainty it brings eliminates the need for the second reference power supply, but it yields lower measurement accuracy and trackability.
Can you comment on the Zener diode calculation I did in my last post? It looks to me that the current for the Zener regulation is in the range of ~10 mA for a 1W-rated Zener diode (not 100 uA, it is hugely different), resulting in a resistor R2 in 3.3 k, but not 330 k. I might miss sth there, or not read the datasheet of Zener correctly (it turns out all Zener diodes give this value of I_test, which looks to me like the bias current.) Please confirm.
Can you elaborate a bit more on the condition of triggering the avalanche process - which is critical in my design? I think I understand your subtle point but let me elaborate on my thinking - in my real design, there is no photon (it is not really a scintillation device), the ion current, more specifically the generated electrons inside the ionization chamber are directly falling on top of the PN junction’s P surface via 24V bias, NOT through a “wire” as the form of a current. Both the PN junction and the ionization chamber are not off-shelf, but I am integrating them (the ionization chamber sitting right on top of the P-side of the PN junction) in a clean room through microfabrication. Therefore, analogous to the photon going inside the PN junction and photon-generated carriers triggering the avalanche process, in my case, the ionization-generated electrons fall (biased by the 24V) directedly on the PN junction and get accelerated by the large electric field generated by the reversed biased depletion region, hence avalanche.
Of course, I don’t know if it will work or not, but from the physical point of view, it seems worth a try. I think if we look at this from a simple electronic perspective, “an nA current feeding into the APD to trigger avalanche”, yes, it sounds not feasible.
Outside of situations involving lightning or utility power service, “ground” is a term that more often than not refers to an arbitrary node in a circuit, chosen for convenience of representation. Most power supply outputs are isolated from the dirtball spaceship, allowing them (within reason) to be stacked in whatever order and polarity one might desire, to create positive or negative potentials with respect to any point a person might choose to designate as the common reference point for other measurements. Stated differently, there is no general requirement for the negative terminals of your bias sources to be connected.
Heke, another comment regarding the by-pass capacitor, see below the technical file you shared with me from HAMAMATSU, the red circled is a by-pass capacitor for the same purpose right? (it also mentions to shorten the connection (parasitic inductance) to the APD)
Yeah, 3k3 should work very well. Your calculation is correct, I’d say. Zener as a regulator has poor accuracy, line and load regulation performance, but you’ll need to determine by measurements whether the accuracy is adequate. There are more precise solutions existing, offering in addition automatic squelching of the APD (to prevent it from getting damaged due to self sustained avalanche).
Thank you for the explanation. Now it all makes sense. So, I’ll take some words back: The node “C” in my previous schematic does not need to be “high impedance” as the ionization current bypasses that node. In fact you may consider shunting the zener with some capacitance in order to improve the ADP response. I am afraid I cannot suggest any value.
To answer to your another post: Yes the Hamamatsu’s example schematics has the same supply by-pass capacitor shown. It is a pretty important part of the circuit.
So, all in all, your circuit should work. You only need to decide how you’ll measure the avalanche current. Also, well, you certainly know this already, but just hinting that the particle that you’re about to detect may also hit to the PN junction causing possibly damage.
When the APD’s avalanche breaks loose, the current goes through the zener (well, some goes through the ionization chamber, as it has 100pF of capacitance). The zener has dynamic series resistance and some parasitic series inductance, which will slow down the avalanche. A capacitor in parallel to the zener will bypass those making the response faster.
Yes, that is doable. The cap is used to make the avalanche pulse more detectable (a voltage dip with slow recovery) and provides the commutation (squelching). In your circuit though, it won’t work, as you’ll always have the 35V over the APD. In Hamamatsu’s circuit the cap is not parallel to the APD as the avalanche current goes through the OpAmp’s feedback resistor (which converts the diode current to voltage) and OpAmp’s output stage and OpAmp’s supply bypass capacitor (not shown in the schematics) before returning to the capacitor, that is, their circuit also operates over constant voltage. So, in the end, you need to put a known small value resistor somewhere in order to measure the diode current.
Thanks for the helpful comment! I will adhere to these suggestions.
Last several (hopefully!) questions, can you instruct me how to properly implement a pulsed voltage source? As I found that indeed the PN junction went through irreversible damage after a long time at a high reversed bias.
I tried an experiment on placing the PN junction at a DC reverse bias larger than the breakdown (-35V) while recording its current over time, see the result below.
Clearly, the junction cannot withstand high reverse bias as the current would be kept increasing - which can you also comment on? For the given experiment, I also record the temperature on the silicon chip by placing a thermocouple. When biased at -60V, the temperature rose from 31C to almost 50C during the 300 s measurement. While the -50V went up to 29C, and -40V stayed at 23C (RT). Is this called the thermal runaway?
I am thinking of using a pulse generator after the voltage supply to give it a PWM-like pulsed voltage to hopefully go around this issue. Also, I would bias the junction at ~-40V (sacrificing the gain) to minimize this effect, as I am worried it can easily bury my target amplified signal.
If I would like to run it at a higher voltage, assuming this is a thermal runaway, is it possible that I place a TEC cooler underneath the chip to keep it from being damaged?
Yes, it seems to be the case. The avalanche keeps on going by itself and looks like it is a function of the junction temperature. You can certainly try a TEC to remove thermal energy from the device.
Probably the easiest way to do it is to use a switch device in series with the APD. You could consider e.g. an opto relay, which makes it easy to do the switching (no need to worry about biasing, level shifting etc.).
Questions are always fine. Just concerned if this thread with over 50 posts serve the TechForum’s purpose…
Please feel free to correspond alternatively by email to me info(at)asamalab(dot)com . I’ll try to respond when schedule allows.
I sent an inquiry through the given email with some new results and questions. When you got time, please take a look and I look forward to the discussion.
