Author Topic: Ultra low power, clean 30V power supply + ultra low power high speed comparator?  (Read 6639 times)

0 Members and 1 Guest are viewing this topic.

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
I'm working on a radiation detector based on a scintillation crystal and a silicon photomultiplier, and I'm stuck with a few decisions regarding low power operation.
SiPMs are basically a massively parallel array of single-photon avalanche photodiodes operating in reverse bias in the Geiger region, at roughly 30V. When a photon hits the SiPM, some SAPDs fire, so you get a current proportional to the number of photons.

While I do have a working prototype of this detector already, it's consuming an abysmal ~1000-1100uA at 3V, which is completely unacceptable for a device meant to operate continuously in standby mode for weeks or months on small batteries.

So here's two interesting problems:
1: What's the easiest way to build a very low power(<30uA average) power supply capable of delivering ~30V at no load(the average SiPM current with a scintillator is in the nA range) with less than 10mVp-p of noise? The noise is critical, as the signal from a SiPM is very very tiny.
2: How do you detect a 20-80mV, 50-200ns pulse and turn it into a longer digital output for counting(interrupts)? Most nanopower/micropower comparators have a minuscule bandwidth and a very low slew rate. I'm guessing I'd have to build something discrete.
It's also possible to operate the SiPM in current mode with a transimpedance amp, but high slew rate, high bandwidth, ultra low power opamps simply don't exist as far as I'm aware.
« Last Edit: June 28, 2019, 08:46:36 pm by Spirit532 »
 
The following users thanked this post: GEOelectronics

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
1: What's the easiest way to build a very low power(<30uA average) power supply capable of delivering ~30V at no load(the average SiPM current with a scintillator is in the nA range) with less than 10mVp-p of noise? The noise is critical, as the signal from a SiPM is very very tiny.

Does the 10mVpp only apply to high frequency noise or does that include regulation?

If regulation is not a problem, then I think any burst mode or hysteretic boost controller can meet your requirements with passive filtering.  These have the advantage of the lowest quiescent power and are commonly used for micropower regulators.

Quote
2: How do you detect a 20-80mV, 50-200ns pulse and turn it into a longer digital output for counting(interrupts)? Most nanopower/micropower comparators have a minuscule bandwidth and a very low slew rate. I'm guessing I'd have to build something discrete.

I might have it set the state of a CMOS flip-flop which is then reset by the counter but that still leaves how to amplify the signal enough without drawing too much power.

Quote
It's also possible to operate the SiPM in current mode with a transimpedance amp, but high slew rate, high bandwidth, ultra low power opamps simply don't exist as far as I'm aware.

What about integrating the current output and periodically reading the output voltage and resetting it as needed?  Each pulse should create a step in the integrated output.

Or integrate the output with a low integration capacitance making a charge amplifier.  What is the minimum amount of charge which must be detected?
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7463
  • Country: hr
« Last Edit: June 28, 2019, 09:08:17 pm by 2N3055 »
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
Does the 10mVpp only apply to high frequency noise or does that include regulation?
If regulation is not a problem, then I think any burst mode or hysteretic boost controller can meet your requirements with passive filtering.  These have the advantage of the lowest quiescent power and are commonly used for micropower regulators.

Regulation can be within +-20mV or so, going beyond that will make a rather nonlinear response. The gain rises very very rapidly, with some of my SiPMs shooting through an order of magnitude with just 100mV.
It's 10mV p-p for any noise above ~1kHz, I'm afraid, or it'd need amplification(again, not low power for this high of a speed).

I might have it set the state of a CMOS flip-flop which is then reset by the counter but that still leaves how to amplify the signal enough without drawing too much power.

A discrete flip flop? I don't think 100ish mV is enough to flip something digital, and that's a very high energy photon(cosmic ray-ish).

What about integrating the current output and periodically reading the output voltage and resetting it as needed?  Each pulse should create a step in the integrated output.
Or integrate the output with a low integration capacitance making a charge amplifier.  What is the minimum amount of charge which must be detected?

The idea is for the host device to sleep until there's an interrupt on one of the inputs, rather than continuously sampling and determining whether there was another particle or not.
The charges are pretty tiny, though I don't have any figures.


Tons of stuff on Google...
High power, large stuff is easy. Slap in a few 100V/us slew opamps with a massive GBWP, a filtered high voltage supply, and you'll get a clean signal.
Miniaturizing this stuff and making it ultra low power is the hard part.


Here's an earlier test with an opamp and a bit of pulse stretching:
« Last Edit: June 28, 2019, 09:42:17 pm by Spirit532 »
 

Offline cur8xgo

  • Regular Contributor
  • *
  • Posts: 148
  • Country: us
Well if I wanted to have a fun afternoon I would get the sim going and see if I could detect these pulses by:

arranging a circuit which nulls itself with a time constant much slower than the pulse width but much faster than the typical time between pulses

and design it so it consumes basically no power once its nulled

then this pulse upsets the balance and causes some kind of cascade or avalanche which either greatly increases the pulse voltage or the pulse length or both

there are physical analogies to this so I think it might be workable..and has probable been done a million times

also might want to consider using the fast rise time of the pulse to ring a resonant circuit of some kind of in that way extend the pulse "length"

 

Offline cur8xgo

  • Regular Contributor
  • *
  • Posts: 148
  • Country: us
also I hate to be that guy but what about just using bigger batteries
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
I understand the "that guy" argument here, but a 3-4Ah battery, when used conservatively, would probably only last around 3 months. It's not bad, but I'd prefer that this device could be integrated into something that can be forgotten for a year or more, but still raise an alarm if you're in a high field. It's also basically the size of the entire planned device, which is not particularly nice or compact. An optimal scintillator size is ~8x8x30mm, the SiPM is another 6x6x1mm including the housing and PCB. The circuit could probably fit in a 10x10mm board, especially if it's a double-sided load.
 

Offline cur8xgo

  • Regular Contributor
  • *
  • Posts: 148
  • Country: us
I understand the "that guy" argument here, but a 3-4Ah battery, when used conservatively, would probably only last around 3 months. It's not bad, but I'd prefer that this device could be integrated into something that can be forgotten for a year or more, but still raise an alarm if you're in a high field. It's also basically the size of the entire planned device, which is not particularly nice or compact. An optimal scintillator size is ~8x8x30mm, the SiPM is another 6x6x1mm including the housing and PCB. The circuit could probably fit in a 10x10mm board, especially if it's a double-sided load.

Well I think this is a great project and the technical challenge here is a real juicy one...probably has been done before so there should be something out there..as far as long-term short pulse monitors..

I think a physical analogy would help drive the circuit design.....the pulse is a short low energy kick..its distinguishing features are its quickness.

If you flick a tuning fork it vibrates for a longggg time
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
there should be something out there..as far as long-term short pulse monitors..

There is - just not SiPM-based.
Regular dosimeters based on Geiger-Muller tubes have rather strong signals and generally don't care about massive voltage ripple. Most GM tubes have plateaus of 50V or more, so a 300V power supply with 20V of ripple is absolutely acceptable. They also don't have anything charge-proportional. A particle hitting always means it's a loud and clear discharge, rather than having to discriminate ultra low energy particles(in terms of NORM anyway) that are barely above the self-counting noise floor of the bare SiPM.
That's why most, even cheap, GM-based dosimeters can operate for months or even years on a couple of AA batteries. The Gamma Scout lasts so long it doesn't even have a battery compartment - once it dies, you just send it in for re-calibration and battery swap.
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Does the 10mVpp only apply to high frequency noise or does that include regulation?
If regulation is not a problem, then I think any burst mode or hysteretic boost controller can meet your requirements with passive filtering.  These have the advantage of the lowest quiescent power and are commonly used for micropower regulators.

Regulation can be within +-20mV or so, going beyond that will make a rather nonlinear response. The gain rises very very rapidly, with some of my SiPMs shooting through an order of magnitude with just 100mV.
It's 10mV p-p for any noise above ~1kHz, I'm afraid, or it'd need amplification(again, not low power for this high of a speed).

What I suggested is still doable but with that regulation requirement, I would use a micropower linear post regulator.  Micropower circuits do not have good frequency response so filtering between the output of the switcher and linear regulator will need to be carefully designed.

20 millivolts out of 30 volts is 667ppm so accuracy over temperature is going to be a problem.

Quote
I might have it set the state of a CMOS flip-flop which is then reset by the counter but that still leaves how to amplify the signal enough without drawing too much power.

A discrete flip flop? I don't think 100ish mV is enough to flip something digital, and that's a very high energy photon(cosmic ray-ish).

I meant using an integrated flip-flop as a pulse stretcher but it will require a fast amplifier which is problematical for a low power circuit.  See below.

A discrete flip-flop with low threshold voltage which is fast enough to respond to a 100 millivolt trigger is possible but I am not sure it can be done within your power budget.

Quote
What about integrating the current output and periodically reading the output voltage and resetting it as needed?  Each pulse should create a step in the integrated output.
Or integrate the output with a low integration capacitance making a charge amplifier.  What is the minimum amount of charge which must be detected?

The idea is for the host device to sleep until there's an interrupt on one of the inputs, rather than continuously sampling and determining whether there was another particle or not.
The charges are pretty tiny, though I don't have any figures.

A charge amplifier removes the requirement for high bandwidth as long as fast response is not required so supply current can be very low.  The tricky part then becomes drift.  The output of the charge amplifier can directly feed an interrupt line through a low current schmitt trigger; I would use a micropower comparator here.

Limiting the DC gain of the charge amplifier may be enough to remove calibration requirements.  For an integrator, this means placing a large value feedback resistor across the integration capacitor to absorb any changes in input leakage current.  Sometimes a T-network is used with two resistors and a capacitor to ground to remove the requirement for a high value resistor; the shunt capacitor lowers gain at DC to 1 to prevent amplifying the input offset voltage.

The processor then needs a way to reset the charge amplifier after being triggered.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15151
  • Country: de
I don't think the ripple would be such a big problem - it just needs a relatively large capacitor for filtering. At 30 V it could be tricky to find a suitable low power and low noise linear regulator though - though there could be a few. I would be more worried about low frequency noise than higher frequency ripple.

The idea with a charge amplifier sounds very good, as it reduces the need for high speed parts. The charge amplifier could end up being more like a pulse stretcher. However the amplitude after the charge amplifier will be likely rather small, so that it would need amplification first.

It is essentially impossible to build a fast discrete low power comparator, as parasitic capacitance is critical and very small transistors would be needed. The comparators build in low power µCs are relatively good, as they use a relatively fine structure CMOS process, normally not found with analog parts.

The nice point about the charge amplifier solution would be that it could still read the pulse hight
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
1: What's the easiest way to build a very low power(<30uA average) power supply capable of delivering ~30V at no load(the average SiPM current with a scintillator is in the nA range) with less than 10mVp-p of noise? The noise is critical, as the signal from a SiPM is very very tiny.
Ultra-low-power boost converter like MAX17220 with a cascode MOSFET followed by ultra low power linear regulator like STLQ020PUR with some more external transistors so it can form a higher voltage linear regulator? (Can't immediately find any description of such a circuit, but should be possible.)
Quote
2: How do you detect a 20-80mV, 50-200ns pulse and turn it into a longer digital output for counting(interrupts)?
The unbuffered inverter trick trick seems interesting, one of the inverters is biased linearly with a 10k-100k feedback resistor to become a transimpedance amplifier. Then more unbuffered inverters as amplifiers (if they are on the same die they should be biased linearly by the DC coupled output of the transimpedance one). Maybe some of the very low voltage logic families will be fast enough at low enough supply voltage to work and low enough power.

The On Semiconductor SiPMs with fast outputs look interesting too, no need for signal processing to detect multiple events close together.
« Last Edit: June 29, 2019, 04:41:15 pm by Marco »
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
The unbuffered inverter trick trick seems interesting, one of the inverters is biased linearly with a 10k-100k feedback resistor to become a transimpedance amplifier. Then more unbuffered inverters as amplifiers (if they are on the same die they should be biased linearly by the DC coupled output of the transimpedance one). Maybe some of the very low voltage logic families will be fast enough at low enough supply voltage to work and low enough power.

Operating a CMOS inverter in its linear region gives good sensitivity and frequency response but at the cost of high power.

I don't think the ripple would be such a big problem - it just needs a relatively large capacitor for filtering. At 30 V it could be tricky to find a suitable low power and low noise linear regulator though - though there could be a few. I would be more worried about low frequency noise than higher frequency ripple.

I have built these kinds of circuit before and the problem was always high frequency noise from the switching converter coupling through the detector (or supply connections) into the discriminator.  It is not insurmountable but care must be taken filtering the output from the switching regulator.

There are lower noise options but the micropower requirements in this case dictate a hard switched micropower switching controller.
« Last Edit: June 29, 2019, 06:07:12 pm by David Hess »
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
The nice point about the charge amplifier solution would be that it could still read the pulse hight

Does pulse height actually mean anything useful though? The number of optical photons per high energy photon depend on the energy, which is an unknown. AFAICS pulse height is either irrelevant or constant (if the SiPM is saturated).
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
Operating a CMOS inverter in its linear region gives good sensitivity and frequency response but at the cost of high power.

Anything which has amplification at >100 MHz is relatively high power ... but does it give good bang for buck when you don't really care much about linearity?

What would a RF BJT need to be biased at to allow a decent trans-impedance amplifier to be made at those speeds? 100 uA?
 

Online David Hess

  • Super Contributor
  • ***
  • Posts: 17427
  • Country: us
  • DavidH
Operating a CMOS inverter in its linear region gives good sensitivity and frequency response but at the cost of high power.

Anything which has amplification at >100 MHz is relatively high power ... but does it give good bang for buck when you don't really care much about linearity?

It does not matter how linear it is if it draws too much power.

Quote
What would a RF BJT need to be biased at to allow a decent trans-impedance amplifier to be made at those speeds? 100 uA?

It would be lower than the bias required for an FET.  At the same current, bipolar transistors have greater transconductance than FETs meaning higher gain-bandwidth product.
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
20 millivolts out of 30 volts is 667ppm so accuracy over temperature is going to be a problem.

That'd be 1200ppm, since +-20 is fine. Could probably push it a bit further, maybe +-40-50mV, but with higher drift the gain might become an issue.

A charge amplifier removes the requirement for high bandwidth as long as fast response is not required so supply current can be very low. 

I'm estimating that the maximum count rate will be around 2-5kHz for really high radiation fields on such a small crystal(several mSv/hr), so it may be an option worth exploring.

Ultra-low-power boost converter like MAX17220 with a cascode MOSFET
The On Semiconductor SiPMs with fast outputs look interesting too, no need for signal processing to detect multiple events close together.

A cascode setup would require quite a serious supply voltage and current to swing an external mosfet, which is outlined in that article, so I'd probably need another boost just to supply this boost...
You can see the OnSemi(formerly SensL) SiPM in that video, though using the regular output. The fast output is only really useful for single-photon counting applications, here we're counting thousands to millions of photons(depending on particle energy) as an analog signal(see below).

Does pulse height actually mean anything useful though? The number of optical photons per high energy photon depend on the energy, which is an unknown. AFAICS pulse height is either irrelevant or constant (if the SiPM is saturated).

Yes, it does. The SiPM is never saturated, as the only purpose of a SiPM is to not get saturated, by multiplying the number of mini-SAPDs(Which do get saturated, but there's thousands of them and some do, some don't. It's sort of like an R2R DAC in that regard.).
The output is directly(and mostly linearly) proportional to the energy of the incoming particle. Thus, you can do spectroscopy in a tiny package.

However, for this low-power version I explicitly don't need spectroscopy(or could have a second circuit that can be put to sleep when not in alarm mode), and I'd like to use the SiPM in a saturation-like mode, with just a simple digital output after a certain energy threshold.
SiPMs are not clean - they can have several MHz of self-count rate, but the pulse heights are generally a very small number of SAPDs firing, so it's usually much lower than actual photon amplification signals, so a basic height discriminator cuts off garbage below the ~10keV(equivalent-ish with a CsI:Tl crystal) level, which is not very useful regardless.
You can see that self-counting noise in the video linked above. It's just below the useful scintillation signal.
 

Online Kleinstein

  • Super Contributor
  • ***
  • Posts: 15151
  • Country: de
The comparator inside the AVRs (e.g. Mega88PA and similar) uses some 50 µA at 3 V.  I don't know for sure how fast the comparator can react, but it is not that slow and to my surprise also reasonable low noise / jitter (e.g. < 5 mV with a 1 V/µs slope). So the comparator may be just good enough to catch the pulses.
 

Offline cur8xgo

  • Regular Contributor
  • *
  • Posts: 148
  • Country: us
just to be clear...operational duty cycle must be 100%? cant do 10ms on 990ms off, or some kind of randomized duty cycle which statistically results in the same end result? Could increase battery life an order of magnitude and make the circuit much easier to design

I mean if you are in a high field do you really need to know that (can it be known?) only by being on constantly?

 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
It takes a while to stabilize the power supply and get a good background average, so yes, 100%.
 

Online 2N3055

  • Super Contributor
  • ***
  • Posts: 7463
  • Country: hr
These guys made front end that would use something about 60uA..
Is that good enough.. ?
Also a portable battery powered unit..

https://aisberg.unibg.it/retrieve/handle/10446/77221/126298/TDUnibg1002866.pdf
"Just hard work is not enough - it must be applied sensibly."
Dr. Richard W. Hamming
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
Their choice of opamp is not particularly ideal - my prototype uses the Micrel(now Microchip) MIC860, which has a bandwidth of 4MHz and a slew rate of 4V/us while only consuming 40uA. Faster, lower noise, lower power.
Their frontend(transistor-based, as suggested earlier here) consumes over 1.5mA in total, which somewhat crushes the point of having an ultra low power opamp.
I was thinking of using another MIC860 as a comparator with an adjustable threshold voltage(just above noise), with a bit of pulse stretching through feedback. Rise and fall times don't really matter in this case, since it's only going to be triggering a rising edge interrupt.

The real issue here is the power supply - say we have a budget of ~80uA for the amplification and "digification" of the signal, and another 50-80uA for the power supply. How do we get a clean, 28.5-33V(depends on the chosen SiPM, but that covers basically all of them) supply that, at no(or very little, nA) load would give us less than 10mVp-p of higher frequency noise(above 1kHz), and less than less than ~20-30mV of ripple?
The current consumption issue mainly sits in the realm of the power supply.
 

Offline Vgkid

  • Super Contributor
  • ***
  • Posts: 2727
  • Country: us
While not geiger mueller, have you thought about modifying an older epd. The ones we have at work last some time on a single aa batery.
If you own any North Hills Electronics gear, message me. L&N Fan
 

Offline Spirit532Topic starter

  • Frequent Contributor
  • **
  • Posts: 490
  • Country: 00
    • My website
That's like suggesting I buy a bicycle and turn it into a chair instead of building a table...
And FWIW, their sensitivity is 2-5 orders of magnitude less than the crappiest GM tubes. Scintillators have 2-4 orders of magnitude higher sensitivity than the best GM tubes.
« Last Edit: June 30, 2019, 04:28:57 am by Spirit532 »
 

Offline Marco

  • Super Contributor
  • ***
  • Posts: 7043
  • Country: nl
A cascode setup would require quite a serious supply voltage and current to swing an external mosfet
Don't see why, a simple BSS123 should suffice at 3V. You won't be switching a whole lotta current. Use a low noise avalanche diode for the upper leg of the feedback divider.
Quote
The fast output is only really useful for single-photon counting applications, here we're counting thousands to millions of photons(depending on particle energy) as an analog signal(see below).

It also seems useful for very high counts, if high energy photons strike the scintillator every say 10 ns you can still detect that. On the slow output those pulses pile up, the fast output will gives distinct pulses. You essentially get a well designed CR pulse shaper for free. That said, to detect pulses that fast you need even more bandwidth, so for low power and low counts I guess it doesn't make much sense.

TLV7011 seems the most appropriate comparator to me, increase the anode resistor to say 220 Ohm and the pulse should last long enough for it.
« Last Edit: June 30, 2019, 06:51:30 am by Marco »
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf