Ultra low power, clean 30V power supply + ultra low power high speed comparator?

[Someone]

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.

There are many linear regulators hitting these specifications, but things like capacitor leakage and board contamination start to become an issue down at these low currents. Its not difficult to reach the OPs requirements for power. The glaring problem is thermal sensitivity as noted already:

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

And that is before considering the thermal characteristics of the SiPM its self, though the discrimination technique planned by the OP makes it less sensitive than it could have been. For reference Hamamatsu include temperature compensation on their MPPC (SiPM) supplies.

[Marco]

20 mV shift of the cathode voltage is going to do very little to the gain of the SiPM, hardly a problem.

20 mV ripple is more annoying.

[IconicPCB]

lm723?

[Wolfgang]

lm723?

Yes, absolutely. But with a PNP pass transistor, and a Wenzel cleanup shunt. If you need to be absolutely sure, make a preregulator, too.

[Marco]

Not exactly ultra low power.

[Spirit532]

It also seems useful for very high counts, if high energy photons strike the scintillator every say 10 ns you can still detect that.

If high energy photons strike the scintillator every 10 nanoseconds, the issue you should be worried about is your body evaporating. What are you doing inside a live fissioning reactor anyway?

And that is before considering the thermal characteristics of the SiPM its self, though the discrimination technique planned by the OP makes it less sensitive than it could have been. For reference Hamamatsu include temperature compensation on their MPPC (SiPM) supplies.

With a clean power supply, temperature compensation isn't really required for counting. It'll be clean and linear enough to provide a rough(+-10% is standard) dose rate estimate, and certainly more than enough to alert you if things go sour.

[David Hess]

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.

I looked at something like this for a proportion counter a couple years ago.  The charge amplifier catches everything through continuous integration but the ability to distinguish adjacent pulses depends on bandwidth just like a transimpedance amplifier.  The difference is every pulse is captured even at low bandwidth; pulses which are too close together are combined.

The problems are how to handle drift since a charge amplifier will happily integrate any leakage current at its input and how to reset it.  The leakage current can be cancelled with a temperature compensated bias current cancellation circuit or through an automatic calibration routine.  Limiting DC gain with a resistive feedback network like I described (AC shunted t-network) also helps and may be sufficient.

Resetting the output between counts can be done with the various integrator reset methods but I suspect a better way would be to subtract a fixed amount of charge similar to how a charge balancing ADC works.  This allows the charge amplifier to continuously integrate the input without missing anything.

[Spirit532]

The issue here is the power supply. Signal processing doesn't incur that much cost.

[Marco]

With a clean power supply

On second thought I don't think any ready made boost converters will do you much good, they are tuned for far too high a frequency and far too low a turn on time AFAICS. How about simply having a MSP430 in low power mode 3, pulsing a BSS123 for 1 tick of a 32768 Hz crystal clock at ~1 kHz to boost the voltage using a 100 mH inductor? Then just a question of how to tune the pulse frequency, I'd guess error amplifier and using the ADC in the microcontroller (or maybe use a microcontroller which can do ultra low power PWM but has a better ADC like the MSP432). Due to the low currents and the relatively large smoothing capacitor it doesn't need to be very fast. No linear regulator or ripple killer is really necessary, with a 100k/100nF smoothing filter ripple would be dominated by the SiPM current pulses (might have to increase the output capacitor a little more to give the microcontroller time to respond to varying pulse counts).

The TLV7011 seems the best ultra low power comparator around, with ~5nF SiPM and ~200 Ohm anode resistor it should be fast enough ... connect the output to a MSP430 external clock input for a timer to count.

[David Hess]

The issue here is the power supply. Signal processing doesn't incur that much cost.

The digital side does not but the analog side will if it requires high bandwidth.

I do not consider the power supply to be the most difficult part.

[Spirit532]

I do not consider the power supply to be the most difficult part.

Read what I said about the analog part with MIC860 on page 1 and what Marco said about the comparators.
The power supply is the most painful part right now.

[GEOelectronics]

I
2: How do you detect a 20-80mV, 50-200ns pulse and turn it into a longer digital output for counting(interrupts)?

This preamp might be a start: pic attached

George Dowell

[Spirit532]

I love the scrubbed out chip names.

[GEOelectronics]

all are CA3096

George Dowell

[Neomys Sapiens]

If I recall it correctly, one of the famous LT circuit collections did contain something on detector supplies.
Maybe AN45, but could habe been another DC/DC-related one too.

[Marco]

These detectors are really into a class of their own as far as power consumption goes though.

How about using an ultra low power oscillator with output enable such as OV-7604-C7 and using a low power comparator to switch their output on/off as a pulse skipping boost converter? Similar to the Theremino power supply, but even lower power because you're not running any logic ICs in linear mode.

[GEOelectronics]

"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.".

Can you post results of the detector portion? I haven't seen many successful attempts at building these regardless of the total current. You mention 1mA current as abysmal. To me that's good, if it is for the whole system, that would put it in line with a PM1703M system which can run for a month on a single AA battery, logging etc right on along.

George Dowell

[Marco]

AFAICS simply using a large enough anode resistor (which increases the pulse duration) with a TLV7011 will allow detection with very low power consumption (5 uA for the comparator).

Use an OV-7604-C7 oscillator driving a BSS123 and some large inductor (1-100 mH) for a boost converter to get 30V. With an ultra low power comparator like the TLV7031 to turn the output of the oscillator on or off and to pull the gate down through a diode when it's turned off. This would have an irrelevant quiescent current of around 1 uA.

[Spirit532]

Can you post results of the detector portion?

There's a short video on page 1 showing how it works, but other than that - I can get around 0.9-3cps/uSv/h(compared a few other meters) with a 10x10x30mm CsI:Tl crystal and the MicroFC-60035 SiPM from ON Semi(SensL), depending on how touchy I get with tweaking SiPM bias and amp gain. Not particularly reliably, but a prototype is a prototype.

If it is for the whole system

PSU and amp only at the moment. Given the recent suggestions, I should be able to get it much lower, which will get it closer to the Radiation Pager lifetimes, but much smarter.

AFAICS simply using a large enough anode resistor (which increases the pulse duration) with a TLV7011 will allow detection with very low power consumption (5 uA for the comparator).

Can you elaborate? The SiPM will go through a MIC860 to get amplified first. I don't think the TLV7011 can operate at an extremely low voltage(3-25mV pulses).

Use an OV-7604-C7 oscillator driving a BSS123 and some large inductor (1-100 mH) for a boost converter to get 30V. With an ultra low power comparator like the TLV7031 to turn the output of the oscillator on or off and to pull the gate down through a diode when it's turned off. This would have an irrelevant quiescent current of around 1 uA.

That is a great idea! I'll throw something together when I get the time, working on a different project right now(DIY SEM, no microamps required here...).
This could also be very useful for a high voltage SMPS for generic geiger-muller tubes, since the plateaus are much bigger there(10V ripple @ 410V is acceptable). A GM tube power supply that pulls <10uA on average would be a hit as well, though there's a bigger issue of high voltage measurement and hooking a boost up to a multiplier there.

[GEOelectronics]

Can you elaborate? The SiPM will go through a MIC860 to get amplified first. I don't think the TLV7011 can operate at an extremely low voltage(3-25mV pulses)."

Have you considered going with a CZT direct?

By the way the PM1703 uses something very similar you your prototype.

Geo

[Spirit532]

CZT is an order of magnitude less sensitive and an order of magnitude more noisy, not to mention the rather painful prices, with not much benefit.
SiPMs with scints are the way to go at this point. Maybe in the future something will change.

[Marco]

The openlabs project I linked earlier which used an inverter for amplification said they got a 200 mv out of their scintillator sipm combo. That would be enough for the comparator.

[Marco]

The diode is not necessary with the ov-7604-c7, I thought it went high impedance when OE is low, but it actually pulls the clock output low according to the application manual.

[Marco]

MIC860

BTW, I have no idea how accurate it is, but Spice says a BFP720ESD common emitter amplifier has a gain bandwidth product of 340 MHz at 30 uA bias current ...

I suspect it will be quite a bit higher than MIC860 even IRL. You don't really need the DC coupling or class AB push pull output of the MIC860 so a high ft low current RF transistor amplifier will almost certainly do better. A BFU710F might do a little better still, but the BFP720ESD is less fragile.