EEVblog Electronics Community Forum
Electronics => Projects, Designs, and Technical Stuff => Topic started by: Amper on July 12, 2019, 12:32:38 pm
-
Hi,
Im designing a small geiger counter meant to run of a single 135mAh lithium cell for extended times logging radiation. To extend the possible time between recharges i would like to have a very low power regulator for the roughly 500V supply. Maximum current would be only a few micro amperes. I guess some kind of burst controller would be nice but i dont have a good overview right now.
any recommendations welcome ^^
-
You'll have a hard time finding a single, integrated boost converter capable of going from ~3.7V input to 500V, and even harder time finding one which has very low quiescent current...
(You said "lithium cell" and said it was rechargeable, so I assumed a Li-Ion/LiPo cell of some sort here...)
But let's start with simple maths first: assuming the output current is 10µA and you can find/design a converter that is 80% efficient (which IMO would be already pretty optimistic here at this output current), that would yield an input current of 10e-6*500/3.7/0.8 ~ 1.7mA. With a 135mAh battery, you'd get ~79h (a little over 3 days) of continuous operation assuming that nothing else is drawing current (which I doubt if your device is supposed to be logging data)... would that be what you'd call "extended times"? Or would you actually have your device log radiation only a fraction of the time, thus having a low duty cycle? This preliminary remark just to make sure your expectations are realistic.
One approach could be to use an efficient, low-power boost converter to go from 3.7V to an higher voltage and then cascading a Cockcroft–Walton multiplier (see: https://en.wikipedia.org/wiki/Cockcroft%E2%80%93Walton_generator (https://en.wikipedia.org/wiki/Cockcroft%E2%80%93Walton_generator) , and I think Dave also made a video about it) to go up to 500V. IMO, a Cockcroft–Walton multiplier alone to go from 3.7V to 500V would be bulky and never be efficient enough to get you the level of current draw you're targetting.
One possible such boost converter could be the LT8494: https://www.analog.com/en/products/lt8494.html (https://www.analog.com/en/products/lt8494.html)
(typ. 7µA quiescent current!), set the output at the max. supported: 70V, then cascade a Cockcroft–Walton multiplier for a ~7x multiplication.
Another option would be to use a flyback converter, but this would require a transformer (don't know if it would fit your size requirements) and good luck designing one to draw that little current...
-
Hi,
thanks for the reply, i can see you already found the pieces of information i left out ^^
It doesnt have to be a boost converter, at the moment im tending towards a transformer driven by a boost IC. I have seen such a configuration in some analog datasheet but i just cant find it any more since its not the primary application of the part.
The device will have several modes ranging from continuous logging (for decay experiments and maybe gps logging while hiking in the mountains) to just general monitoring and taking a measurement once a day for 5 minutes. The energy is by far sufficient, i have experience from building a logger for industrial current loop sensor that took much more power. Standby should be around 5uA lasting more than a year without being in use.
Using a multiplier with may stages would be possible but since my space is limited i hope i dont have to go that rout.
Attached is a rough schematic of my plan to this point, dont judge please, its just a sketch and will be refined after im finished coming up with the features needed.
EDIT:
Yes, a flyback would be the obvious simple solution but i stopped liking them for reasons of reliability, also regulated voltage is a requirement and regulating a flyback sucks or at least has no advantage over using a proper switcher.
-
I just took a quick look. What you're working on kinda looks like a boost DC/DC converter used as a flyback converter. I didn't see such a topology in the suggested applications in its datasheet. Without further analysis, I'm not saying it won't work, but I'd suggest you simulate that first (LTSpice has a model for this converter) and see what happens. Also, have you selected an appropriate transformer yet? (As said above, I had ruled out the flyback topology assuming that the required transformer may be too bulky for your needs - I assumed that based on the choice of a pretty small battery. But maybe you've found a transformer that's small enough.)
Check that it can't destroy the converter under any circumstances. Check the input current. Even if it works, I'd suspect a much lower efficiency than what is typical when used as a basic boost converter.
Then, even if the average input current, once it's in regulation, may fit your requirements (to be checked), also carefuly take a look at the inrush current when it's starting. It may draw currents that your battery would not be happy with.
Regarding size, a 7x Cockcroft–Walton multiplier, at such low currents, would require only very small diodes and caps and would probably not take up too much space compared to a transformer.
Edit: I hadn't seen there was a reference for the transformer on your schematic: https://www.coilcraft.com/lpr6235.cfm (https://www.coilcraft.com/lpr6235.cfm)
Looks like it's rather small indeed and would be a nice fit. I'll try and set up a simulation in LTSpice.
-
You could look at using a piezoelectric transformer, no idea where to get them however.
-
Just an additional quick comment after looking at the LPR6235 transformer's datasheet: with a winding to winding isolation of 300Vrms, I'm not sure this is going to work very well...
To be continued.
-
This (well, a similar topology without doubler) was something i have seen in a datasheet and forgot which one it was. Nice enough, you actually found it for me without knowing, have a look at page 21 in the LT8494 datasheet you linked. I forgot about the primary side diode but except for that i guess it should work with the 8330 as well. The inductor is 1:100 which is most likely to much but thats the one i have laying around from a LTC3109 converter i played around with. They are 300Vrms rated so i guess it should be fine here and if it turns out the converter works like i invision it i will order the proper one at the right ratio.
Sadly they are not available with middle tap, they are called coupled inductors and i couldnt find one suitable for flyback.
Thanks&cheers!
-
why do i neeed those things?
@SiliconWizard the 300V is fine using my voltage doubler,going without yes, that could be to close especially since im not sure if moisture may be an issue in the future, i dont want to conformal coat it right from the beginning.
-
As I wasn't looking for that in the LT8494 datasheet, I actually missed it! But yes this is it. As I said, it's indeed a flyback converter made out of a boost converter.
Yes there's a missing diode in your design.
Yeah I haven't really taken attention that you added a voltage doubler as the last stage. So you should be OK.
I will still try and simulate that later.
-
The thing im worried about now is that i dont have enough negative swing on the transformers output due to this diode, that would render the doubler useless. If you want to do a sim that would be cool, im usually to lazy sadly... But i will try putting a test circuit together with a similar controller i already have laying around.
EDIT: And now im not o sure any more about the whole voltage rating thing, i only looked at the primary side but since the doubler is grounded im afraid i will get 500V relative to the primary which is wound below the secondary without additional insulation.... Maybe a cocroft walton is necessary after all, just not so many stages.
-
OK, here is what I came up with based on your design so far (I used/modeled the 1:10 version of the transformer with the respective DCRs).
It reaches 500V in about 5ms. Once stabilized, it appears to be drawing an average of ~100mA from the battery for a 10µA load which is not exactly efficient.
-
Also regulated voltage is a requirement and regulating a flyback sucks or at least has no advantage over using a proper switcher.
Discontinuous mode has the fastest response, that's an advantage.
Siliconwizard, how much worse does it without the doubler?
-
Thank you very much for your"independant evaluation"! I cana ctually confirm this result by my experiments yesterday. I used a Chinese chip running at 1.2MHz (8330 runs at 2) and had it operate without doubler measuring the output loaded by 2MOhm. It reached only 150V at its best and drew up to 400mA @ 2V. Obviously this is far from the final pcb version using the right parts but it shows that its not the right path. Just running the transformer from a function generator and amplifier gave me the same results, pushing it to 600Vpp the transformer actually heated up to a point near damaging. Though this way i found out that this particular 1:100 transformer works best at 500kHz and looses efficiency fast going into the 1-2MHz range, so no surprise linear is using 250kHz with theirs.
The next experiment was using an lt3757 boost converter as a basis (old prototype laying around) I modified it to fit this topology and it also took over 1W just to hold a voltage of 400V even though the frequency was just 200kHz.
Well, summing this up, transformers seem to suck more at this sort of frequency than i expected however a friend of mine pointed me the right direction yesterday. Its also very close to what you recommended first yesterday using a cocroft walton with a boost stage first:
https://www.maximintegrated.com/en/app-notes/index.mvp/id/3757 (https://www.maximintegrated.com/en/app-notes/index.mvp/id/3757)
This should be roughly at 1-2MHz as well and the no load input current is phenomenal, i will try to put one together today and then have a look how far i can reduce the size of it. Maybe reduce by one or two stages and use sot23 double diodes could get it to acceptable dimensions on the pcb.
-
Siliconwizard, how much worse does it without the doubler?
Average current draw drops a little (not much, about -1.5%), which makes sense since there is a bit less loss. But then it takes atrociously longer to reach 500V. And then there is the problem that the transformer selected here wouldn't handle it in real life.
-
Sooow i just tried the maxim approach without the regulation OP on a breadboard with film caps and 1N4007, it works beautifully. Just had to reduce the frequency to around 500kHz again for reasons of the FET i used having far to much gate charge to be driven by the 74.
Using only 7 capacitors and diodes i can easily reach 1,1kV without load. The input power using the not suitable components is .8W at 5V. This should drop significantly with the proper parts and regulation.
The most interesting thing is how load sensitive the device is. It takes several seconds to charge the bank of 100nF caps up to 1kV, voltage measurement was done by hp3457a and a 1G resistor giving a rough 1:100 divider, direct 10M loading makes the output instantly crash to 200V.
Running an Si8b Pancake tube from it is very interesting as well as the unregulated supply drops depending on radiation. Just having the thing sit will result in roughly 150 impulses every minute and thee voltage will rise up to 1kV. If 10kBq Strontium90 is used to ionize the gm tube the voltage quickly drops and even reaches 300V, at this point the tube will no longer operate properly and the count rate drops to an equilibrium.
I could imagine a mode of operation first charging the capacitor to a set voltage and then have the tube discharge it over a long period of time until the cap reaches a threshold triggering a comparator interrupt in the microcontroller. Coupled with a real time clock this could provide extremely low power measurement not even requiring any microcntroller to stay awake during measurement. The only disadvantage will be the need or a >1GOhm resistor to do the voltage measurement which will be a moisture problem.
Ill keep you updated as i continue...
-
That Maxim circuit is quite cool ... I completely forgot that type of boost converter existed (seen it called multilevel boost converter as well).
The way series voltage multipliers droop under load (ripple is not the same for every stage) makes it hard to regulate the output voltage using just the voltage on the lowest stage though. A parallel voltage multiplier might be better.
PS. if you really want it small, VMI makes small multiplier modules like PVM302P08 (example (http://physicsopenlab.org/wp-content/uploads/2017/02/SlowNeutronCoronaCounterTubes_1v10-1.pdf) of its use in a neutron detector). No idea what it costs though.
-
Yes, im curious too how well this regulation will work, on the other hand the average geiger tubes plateu is also around 100V wide so it may be alright to have 10-20% variation. Measuring the output directly would be possible but also requires very very high resistances not to create to much load. As i said before, 10M will completely overload the converter, 1G seems appropriate but even though 1206 parts in this value exist the board it self would be more conductive... Not even speaking of moisture.
I know about these amazing modules though besides the very high price tag they are also very difficult to find anywhere. My current layout is just about double the board area and requires only pretty common components. Changing the fet to a 300V type i could even raise the voltage to 1.5kV since all the other parts are already capable enough.
Something else i just remembered:
There is also the option to use a string of high voltage zeners in series to the sense reistor dropping the first 400V or so. If the 60M divider just sees the last 100V the current should be small enough and the multiplier is loaded evenly without insane resistors.
-
I've simulated the "flyback" topology around an LT8494 instead of the LT8330 and I got better results. It would draw ~50mA on average. Still a lot IMO, but it seems hard to do better than this, so if you manage to do better with Maxim's approach, I'll be interested in seeing that. Keep us posted.
The small 1:N coupled inductors from Coilcraft are great, but pretty limited still. They also have miniature flyback transformers that have better specs, except they are only 1:1, so not really usable here. Haven't found much better so far, except with MUCH larger transformers. Or you'd have to have them custom-designed...
I have also tried the approach I had in mind (in simulation only): a conventional boost converter, followed by some oscillator feeding a Cockcroft-Walton multiplier. I was happy to find the LT8415, which has an integrated dual half-bridge, making it ideal for that without the need of any external transistors or additional integrated high-voltage half/full-bridge. Unfortunately, the max switching current of the LT8415 is so low (25mA) that the output of the multiplier can never reach 500V, not even close... Could be doable for a lower output voltage though such as 100V to 200V...
-
Oh, there's (https://www.theremino.com/wp-content/uploads/2012/02/GeigerAdapter_Datasheet_ENG.pdf) already a complete open source design using the Maxim technique specifically for Geiger tubes.
They also use avalanche diodes for feedback BTW.
-
There's an easier method : Boost converter with an tapped inductor/autotransformer in place of the inductor. Flyback have an unecessary component count, transformers are not cheap, and you don't need the isolation between input and output.
Schematic for example here, on page 43 :
https://www.analog.com/media/en/technical-documentation/application-notes/an19fc.pdf (https://www.analog.com/media/en/technical-documentation/application-notes/an19fc.pdf)
-
An autotransformer has nearly the same complexity as a flyback, the couple of windings saved are hardly relevant, especially given they can be wound with thinner wire.
Isolation has complexity, but you can use the fly back transformer without it.
-
It could be a nice idea though considering thee isstandard boostconverter ICs up to around 60v so a 1:1 tapped inductor could get rid of the previously needed external fet, that is a legit point he has there.
-
Boost converters with lots of annoying features which will just make your life hard, compared to a nice and predictable hex inverter ... I'd go for the hex inverter with two transistor from the Theremino GA500 design.
-
The 7414 does not mean i cant use a transformer, a normal boost topology or what ever afterwards.
-
It could be a nice idea though considering thee isstandard boostconverter ICs up to around 60v so a 1:1 tapped inductor could get rid of the previously needed external fet, that is a legit point he has there.
And why not a tapped inductor with ratio 1:10, or even more? Buy a ferrite core and do it yourself. Sometimes a bit of handwork is better than browsing catalogs for hours searching the perfect off-the-shelf component.
-
We have gone through this already, at the size i need home brewing is not possible, the inductor we are looking at here is a maximum 5x5mm footprint and there is a height limit as well. Also as said before i have problems with losses at those high frequencies and high winding count, the capacitance at this voltage is not to underestimate at 1MHz and above.
-
Ans another update:
Sometimes its a nice idea to ave a look at what others have dome and luckily some people have reverse engineered commercial equipment already.
https://www.youtube.com/watch?v=WH8wNaTSN3M (https://www.youtube.com/watch?v=WH8wNaTSN3M)
Sadly german but you may still see how it works, a single stage boost converter but controlled by the microcontroller and voltage sensing decoupled by diodes to reduce the discharge while the voltage is still there. This is of cause only possible by a micro "looking after" the voltage every few seconds or minutes by pulsing the boost converter a few times but reduces the total current enough to power the entire device from a single AA sized lithium battery for years (over an entire decade) constantly holding 500V of tube voltage.
-
Sometimes its a nice idea to ave a look at what others have dome
You say?
[...] a single stage boost converter but controlled by the microcontroller and voltage sensing decoupled by diodes to reduce the discharge while the voltage is still there. This is of cause only possible by a micro "looking after" the voltage every few seconds or minutes by pulsing the boost converter a few times but reduces the total current enough to power the entire device from a single AA sized lithium battery for years (over an entire decade) constantly holding 500V of tube voltage.
LTC3525 does that without any µC :
https://www.analog.com/media/en/technical-documentation/data-sheets/3525fc.pdf (https://www.analog.com/media/en/technical-documentation/data-sheets/3525fc.pdf)
Iq = 7µA with a 135mAH cell => Autonomy ~ 2 years.
-
But doesnt this one have an internal synchonous buck and no seperate FB pin? I dont see how i could use it in that topology.
-
Yeah. Looks like it's running in circles a bit. I think we already have been through the idea of using a conventional boost converter to generate a high voltage...
Regarding the commercial equipment you mentioned, I'm not completely sure I got how it's generating the high voltage. Care to share a simplified schematic?
-
They basically have a box standard boost converter but the output capacitor is isolated by a string of high voltage diodes. In a usual boost converter this wont work without a load but if you do intelligent pulse skipping up so minutes at a time this way the actual boost converter can be at 0V for the majority of the time while the cap stays charged. Since the discharge rate is so slow its enough to do an active measurement every now ant then at predicted intervals depending on count rate and time.
-
Well, it's just a discrete high voltage boost converter... ;D
Of course you need a transistor that can take the punishment (the BSP125 has a VDSmax of 600V), the inductor should be able to take it as well.
Putting the feedback network BEFORE the diodes (whereas we would normally rather put it after) allows to have it NOT draw current from the output capacitor, but then it requires something a bit more intelligent than just a simple comparator since the feedback voltage is not exactly a fraction of the output voltage... using an MCU to do this solves a lot of problems. :-+
-
Thats exactly what i was saying ^^ Thats also why it wont work with any off the shelve part that i know of as ist not just pulse skipping. Since every ionization of the tube is a very well defined amount of charge its possible to count them and know the voltage of the cap very closely. Then just have a timer for low count rates and fault conditions, more or less a watchdog and make the converter do a short sounding pulse to the boost.
Though i suspect my tube will reduce battery life a bit more since the si8b is faar more sensitive than the original one used in the gamma scout. I will have a whopping 150 pulses a second with no radioactivity present and up to 5000 per second with sources close to the legal limit.
-
I'm not very comfortable not being able to monitor the output voltage though... I suppose you can indeed estimate it, but I wouldn't be so sure about the "very closely"... but at least here, being able to monitor the transistor's drain voltage allows the circuit to 1/ estimate when the target output voltage has first been reached (approximately) and 2/ avoid burning up the transistor and inductor... unless of course, there is a software bug.
-
nooooooo. You didnt get how it operates ^^ Its not guessing the voltage, its just predicting the voltage drop so it knows how long it can stay silent taking zero power. When timer or prediction tell that its time for a checkup the boostconverter is powered up, in this moment the diodes will conduct as the boostconverter obviously has a higher output voltage as the cap. This way you can exactly measure the capacitors voltage as long as the converter is running. After shutdown the decay of the voltage is estimated with a safety margin and when it drops a few volts in the calulation the boost converter is started up again for a few cycles and a new measurement can be taken. According to that measurement the boost converter is run until the nominal voltage is exactly reached again, then the next sleep cycle starts and everything is completely currentless besides the microcontroller waiting for interrupts.
-
Well, yes, I got that! ;D
I'm just saying that the system is "blind" until the converter starts again. I have no clue how accurate the output voltage must be here - I guess not very as long as it's high enough, I don't know much about Geiger tubes - so yeah it may not matter at all. But estimating the voltage drop would be pretty inaccurate IMO (of course would depend on the maximum "off" time). You also have to take the capacitor's leakage current into account. Again I suppose it may be negligible in this application, but even ceramic caps have leakage currents that significantly increase with the voltage, so at 500V, it's probably something to consider as well. It could be leaking more current than you think...
As to software bugs: obviously a purely software-controlled high-voltage boost converter is not completely without risks. There are many ways something could go wrong and the transistor (and many things around it) could literally explode. ;D So you better make it robust software-wise...
If the system is not active at all times but only for a small fraction of the time, depending on the average power draw of your tube, it may be simpler or even more efficient to just switch off the converter completely when the system is idle and start it only when needed. Certainly something to be considered by comparing average power in typical scenarios - scenarios which I have no clue about!
-
Allright :D
The voltage is not very critical, more stable is better but +-10V is acceptable. Considering the tube has only 13pF discharging every pulse and the usual pulse rate is 2Hz on average in standby, the storage cap is 200nF and i accept 10V discharge before a measurement has to happen, thats 300 pulses or two minutes. Not a relevant time for capacitor or pcb leakage since everything has to be potted anyways. Since i know the rate will never drop significantly below 100cpm i can just have a watchdog running that will do the checkup every 60s and it will be enough unless i have some sort of hardware failure in which case i will also not get any pulses in those 60s and the device will go into failure mode. In case it wakes up after one minute and notices a voltage drop much larger than expected it will decide something is wrong with the HV side. If a number of pulses more than 200 is detected while asleep it will wake up as well still having a margin to go in voltage and do a recharge cycle. I cant see a way for it to go out of regulation this way, actually if voltage control needs to be tighter for some reason i could still just wake up after every pulse and recharge, still much less energy required than doing proper regulation and holding the voltage divider under voltage throughout wich would draw very very significantly more power than any thing else in the system. For comparison the maxim circuit could do 1kV with no issue in seconds but would completely break down under 10MOhm load, i had to go to 1GOhm to get a usable voltage reading.
Regarding the software damaging the fet you are right considering the gamma scout circuit but i will just put a capacitor and a slight pull down resistor on the gate, using logic level fets thats enough of drive voltage and in case something gets stuck it will just turn off in time. It worked well on half bridge drivers with bootstrap capacitor, they will just run empty in a few milliseconds and thats it, fire prevented. I understand your concern though, if it wasn't for such a wired application i would never ever drive a boost converter from software.
The reason i now want it this way is that i have seen that its possible and i want to go one better on the commercial thing. The gamma scout is a machine hated amongst hobbyists for user unfriendliness and the high price of 400+€ even though it has some very nice features. I have the obligation to build something better after investing so much time i just cant settle before this thing can run continuous mode :D
-
Yes it looks like a decent approach. The added benefit (don't know if the original device does it that way, and if this point matters) is that you can do all the measurements while the converter is off, which would thus get rid of any noise - don't know if noise on the high voltage could cause measurement issues, but if so, that's a good point.
The risk damaging something is not just with the IO driving the gate getting stuck IMO. It could also be switching away without ever stopping, especially if you use a PWM peripheral in your MCU and some bug prevents your internal control loop from stopping it when the output voltage is reached: in this case the converter could reach dangerous voltages possibly damaging some components. But yes some additional hardware protections can limit the max output voltage whatever happens software-wise.
-
Yes, turning off the converter for low noise is nice, not really necessary with gm tubes though, they are just dump flash lamps completely ionizing and then extinguishing again. However it can be very useful with neutron counters as they require <2mVpp noise on the supply. Though also at three times the voltage, so a cascade will be needed again. For now ill be happy with just having "ok" regulation.
Regarding a hangup in pwm yes, that would suck and i didn't consider it...
But as you already told, most likely some zeners will be enough to keep it from destroying itself. The Tube can survive 1kV if its only short periode, so most likely the fet will just die and blow the fuse. I will worry about it later, if it will be a problem i will add a zener string for the development board and later i guess it should be sorted out.
After all its not going to be a safety critical device, its just a nice toy and gadget for finding radiation sources in every days life. If it breaks its broken, i wont use it to monitor my dose rate ^^
-
Maybe this recent event will motivate your quest some more:
https://www.cnews.fr/france/2019-07-18/du-tritium-radioactif-dans-leau-potable-de-64-millions-de-personnes-862178 (https://www.cnews.fr/france/2019-07-18/du-tritium-radioactif-dans-leau-potable-de-64-millions-de-personnes-862178)
(Google can translate that for you, didn't find the info yet in foreign media.)
-
Mhhh well...
Sadly my detector will not be sensitive enough for this particular incident but i can see exactly where your idea is coming from and its definitely a thought in the back of my head i cant get out. My family owns a small hydroelectric power station near lyon (see my older posts about the lithium welder if you are interested) and all the way from the eighties when my dad bought it he wanted to install a permanent radiation logger there for the reason of not trusting governments about this particular topic. Its not like we trust our own one any further but you see where everything is coming from... Maybe some time i will get around to building a better detector, material is almost all in my stock already but the time issue....
Oh and there is much more radiologic stuff going on then you may expect, i have a friend who worked in decontamination as a students job and he made me aware of an incident in Munich, some guy in a company took an angle grinder to a 1TBq source and roughly 1% of its contents left the building via the hot cells exhaust, the entire neighborhood was covered in radioactive dust in an amount that is actually a health risk. But except for a tiny article in a local newspaper there was no mention to the public, you could have walked there and absorbed the stuff...
-
Aaaaand im back.
Had some spare time and continued the prototype a little bit. It now is made almost entirely from surface mount components that are affordable and the footprint will be tiny once on a pcb. Im now controlling it by only an attiny13 and the control scheme mentioned before. Atm im down to 50uA but my method of measuring it is not the best and likely the actual draw will be even smaller. On a single 2Ah lithium battery 40000h or 4.5 years of continuous supply to the geiger tube is now possible. I still aim to get at least a decade just to beat the gammascout and m pretty confident it will work with jus a few tweaks here and there now.
And here is the mess that this ended up being for now but it seems to work beautifully :D
-
Hi Amper
Glad to see things are going well. This is obviously a difficult design.
I have a couple of thoughts that might help with things.
- The internal resistance of your battery will go up a lot as the battery discharges, especially if it’s cold. This can significantly reduce the useful mAh capacity of the cell. The remedy is to try to limit the peak current applied to the cell. You may be able to pursue strategies like trying not to turn on beepers or LEDs or radios at the same time as the boost converter.
- Large capacitors across the battery can help support peak currents, but their leakage current will drain the battery. There will be a trade off...
- Have you considered a tapped inductor boost converter? This is a method to get higher voltage gains out of a boost converter without really large duty cycles. Downside is you’ll need a higher voltage diode (due to transformer action in the tapped inductor when the main switch is on).
- Diodes with good switching characteristics (eg very low capacitance and Qrr) may not have the best leakage current. You could combine a very good switching diode with a very low leakage diode to get the best of both.
- If you want to sense the main capacitor voltage, you could switch the divider in and out of circuit (eg use a 600V logic level FET). This way you can have reasonable divider resistors without burning too much current
- Circuit behaviour could be quite different at higher and lower temperatures (diode Qrr and leakage are particularly sensitive to temperature)
And finally, please be careful of the high voltages, especially if you have any large-ish caps about.0
-
on a breadboard with film caps and 1N4007, it works beautifully.
Are you absolutely positive that you are using plain vanilla 1N4007 diodes?
The recovery time for those is orders of magnitude longer than your switching period of 500 Khz.
-
@jbb:
Now lets break this down step by step :D
-Yes, this is a problem, though there is sufficiently low leak super caps and larger lithium batteries will be able to supply quite a bit of current without this problem. The average charge pulse is slightly more than 15mJ (charging 50nF from 0 to 500V and trickling the storage cap). One other reason is that i can actually control the shape of the pulse since i use a micro controller. At the moment i do a few hundred pulses to the boost converter, then wait for 500us, take a voltage reading and then continue charging. This way i can stretch the input current pulse to a certain extent and the 500uS can be swallowed by any ceramic input cap.
-Yes, i considered and even tried this a while ago but i wasnt happy with it. It could work much better than igot it to run but in the end there is a very small selection of smd coupled inductors, its a pain to select one and they are almost never rated 500+V and i dont like to use parts out of spec any more. Due to the winding they usually ave the inductances and maximum currents also dont match this application very nicely. One other downside is, there is geiger tubes and other interesting kit that need up to 1500V, so trying a multiplier topology right fro the beginning gives me an opportunity to just extend and have more voltage. Maybe in the future ill try again.
-The diodes im using right now for all the steps are GSD2004S. They arerated 50ns but at full current which they will never see in my case and a nominal leak of 100nA@ 300V. Since i only do 50kHz and for the "checkvalve" on the storage cap i use two pair of them (150V per diode = turns out to be less than 10nA) there is no real problem.
-The topology i using actually does almost this as i drain the multiplier after each charge. This way im using 66MOhm right now but most likely later it will be only a single 10M precision resistor, mostly for reasons of having less possibility for moisture to change the divider.
Your Idea using a FET is interesting, it may come in handy later... Though in this case it will be a Problem since even using logic level with 2Vth when adding the 1.5-2V divider voltage i wont be able to turn it on using a 3-3.3V battery. Also while im still using the multiplier approach, i will have to discharge the multiplier anyways since the ceramics have bad leakage characteristics and a dozen film caps are not very surface mount :D
For example the 1.2uF 630V ceramic i first intended turned out to have 80nA leak which corresponds to 6GOhm, the Vishay film cap i turned to now has less than 5nA leak @ 1uF confirmed by the datasheet stating >=100GOhm). Though this also means, any 100GOhm parallel resistance added by parasitics will half my run time between recharges and doing proper continuous measurements of the voltage decay can only be done by such a 100G resistor and an electrometer like my keithley 616.
-Yes, the temperature still worries me. But it will mostly be a Problem of the double pair which can be extended
without to much additional dropout. The gammascout uses only two BYV26E which are actually really bad, so i hope i will be fine too.
Regarding high voltage, dont worry, i have my fair share of experience with very deadly levels of high voltage up to 20kV 1.5A 3phase, pulsed stuff, tesla, and a whole lot of mains installation. One small Project i did lately you can see here: https://www.eevblog.com/forum/projects/zvs-driver-lockup/25/ (https://www.eevblog.com/forum/projects/zvs-driver-lockup/25/) 20uF@500V with a boost converter delivering 100W+, much better chances of killing myself :D
@schmitt trigger
I used 1n4007 only int the first tests because thats what i had stock of. I dont remember the Frequencies i tried but 50-100kHz are no Problem at all. The leakage current is a bigger problem though, these things show their age, 10uA leak compared to just 50-100nA with modern devices. Also the size is not suitable for my application, even the sot23 double diodes take away quite a lot of board space...
-
If you were operating in DCM, it is OK to have slow diodes.
-
Im actually not even sure im in ccm, never calculated or bothered to measure the timing. Though using the multiplier im not sure if its actually ok since no matter if pushing or pulling on it, there will always be one set of diodes conducting and the other ones isolating. by now everything seems to work nicely, so maybe ill start playing with such detail soon, its just software after all.
One thing im still a bit worried about is that when im between charging cycles and at very high radiation levels i can potentially discharge the capacitor by a continuously conducting tube. This may happen with x-ray as the dose rate is actually large enough so saturate.
-
It could work much better than igot it to run but in the end there is a very small selection of smd coupled inductors, its a pain to select one and they are almost never rated 500+V and i dont like to use parts out of spec any more. ... Maybe in the future ill try again.
Fair point. In fact, I'm sorry I brought it up again, as you'd already discussed it up-thread.
-The diodes im using right now for all the steps are GSD2004S. They arerated 50ns but at full current which they will never see in my case and a nominal leak of 100nA@ 300V. Since i only do 50kHz and for the "checkvalve" on the storage cap i use two pair of them (150V per diode = turns out to be less than 10nA) there is no real problem.
The catch with diodes in series is that they may not share off voltage evenly...
Your Idea using a FET is interesting, it may come in handy later... Though in this case it will be a Problem since even using logic level with 2Vth when adding the 1.5-2V divider voltage i wont be able to turn it on using a 3-3.3V battery.
Yes, the downside is that you might only be able to count on a 1V full scale output, which might sacrifice some ADC resolution.
... voltage up to 20kV 1.5A 3phase ...
Yowza. I was nervous enough about 5kV...
Im actually not even sure im in ccm, never calculated or bothered to measure the timing. Though using the multiplier im not sure if its actually ok since no matter if pushing or pulling on it, there will always be one set of diodes conducting and the other ones isolating. by now everything seems to work nicely, so maybe ill start playing with such detail soon, its just software after all.
DCM / CCM will still be relevant for the main switch and some of the multiplier diodes. Could you maybe post a schematic?
The nice thing about DCM in a low power converter is that you can load a measured amount of energy into your inductor (proportional to (Ton * Vin)^2), then wait for this energy to flow out into the output stage.
There are two possible control aims on the input side: limiting peak inductor current to avoid saturation, and limiting average input current to avoid crashing the battery supply rail (considering a cold, discharged battery).
To limit peak inductor current in DCM:
Vin = L dI/dt
=> Vin = L Ipk / Ton
=> Ton = L Ipk / Vin (1)
Where Vin = input voltage, L = inductance, Ipk = peak inductor current
We have now loaded some energy into the inductor, and need to let it out into the assorted capacitor banks:
Vc = L dI/dt
=> Vc = L Ipk / Toff
=> Toff = L Ipk / (Vc-Vin)
=> Toff' = Toff + Tmargin (2)
Also:
T = Ton + Toff'
Where Vc = bus capacitor voltage scaled by number of charge pump stages, Toff = time for inductor to discharge, Tmargin = extra time to leave some margin, T = total switching period.
We see that Toff' will vary a lot with Vcap. When Vcap is empty, Toff' will be long. As Vcap fills up, Toff' gets shorter. (This is why old school camera flashes make the wheeee sound with increasing pitch.)
We can relate Toff' to Ton:
Ton * Vin / L = Ipk = Toff * (Vc - Vin) / L
=>Ton * Vin = Toff * (Vc-Vin)
=> Toff = Ton * Vin / (Vc-Vin)
Vc and Vin should change fairly slowly, so we can calculate Toff. (Note, initially Vc - Vin = 0 and Toff is allegedly infinite, but resistances and diode drops will take care of this.). Helpfully, L disappears from this equation (assuming you don't saturate it!).
Now let's look at average input[/t] current Iavg:
Iavg = (1/2 * Ton * Ipk + 1/2 * Toff * Ipk) / T
=> Iavg = 1/2 (Ton + Toff) * Ipk / T
Hmm. With some rearranging:
Ton + Toff = Ton + Ton * Vin/(Vc-Vin)
=> Ton + Toff = Ton (1 + Vin/(Vc-Vin))
=> Ton + Toff = Ton ((Vc-Vin)/(Vc-Vin) + Vin/(Vc-Vin))
=> Ton + Toff = Ton ((Vc-Vin+Vin)/(Vc-Vin))
=> Ton + Toff = Ton (Vc/(Vc-Vin))
=> Iavg = Ton^2 * Vc/(Vc-Vin) * Vin / (2*L*T)
So, if we just set T = constant (i.e. fixed frequency), we see that Iavg varies a lot (and we can't make it arbitrarily high or the inductor will saturate). (In this case Tmargin varies a lot.)
But what if we set Tmargin to be small, i.e. Tmargin = 0 (note, in practical designs Tmargin must be > 0 so we don't enter CCM!)?
=>T ~= Ton + Toff
=> Iavg ~= (1/2 * Ton * Ipk + 1/2 * Toff * Ipk) / (Ton + Toff)
=> Iavg ~= 1/2 * (Ton + Toff) * Ipk / (Ton + Toff)
=> Iavg ~= 1/2 * Ipk
=> Iavg ~= Vin * Ton / (2*L)
Solving for Ton, we get:
Ton = 2 * L * Iavg / Vin
And:
Ipk ~= 2 * Iavg
Toff = Ton * Vin / (Vc-Vin)
Tmargin = constant
As a practical matter, you could round Vin into a few bins and use a small lookup table (maybe 8 - 16 entries) to get Ton (scaled directly to clock cycles). A similar approach would probably work for Toff.
This method gets you reasonable control of the input current (good for battery) without saturating the inductor or requiring a current sensor. I think it could reasonably be implemented using an 8 bit MCU. Also note that because no current sensor is required, you probably don't need to update every switching cycle.
If your MCU PWM unit includes a one-shot mode, this would be perfect. Your software can request the one-shot (duration Ton) from the PWM unit, then configure a timer interrupt to go off at (Ton+Toff) at which point you calculate the next step. If the software hangs on something, the PWM unit will push out one pulse and then stop, so nothing blows up :-)
I did a lot of unnecessary algebra below...
Substituting for T:
T = Ton + Toff + Tmargin
=> T = Ton (Vc/(Vc-Vin)) + Tmargin
=> Iavg = [Ton^2 * Vc/(Vc-Vin) * Vin] / [2 * L * (Ton * Vc/(Vc-Vin)) + Tmargin)]
If we want to solve for Ton, this is an unmanageable mess. Let's make some assumptions:
Tmargin is small. This means variable switching frequency!!
Vc >> Vin
=> Vc / (Vc-Vin) ~= 1
Now we can simplify a bit:
Iavg ~= [Ton^2 * 1 * Vin] / [2 * L * Ton]
=> Iavg ~= Ton * Vin / (2*L)
Solving for Ton:
Ton ~= 2 * L * Iavg / Vin (3)
For a control algorithm, we can use the minimum of equations 1 and 3. This provides limited peak inductor current when Vc is small (i.e. startup) and approximately limited input power when Vc is large.
Ton ~= min(L * Ipk / Vin, 2 * L * Iavg / Vin)
=> Ton ~= min(Ipk, 2*Iavg) * L / Vin
Dammit, I could have save a lot of work!
-
I know you're looking for a schematic that works from 3.3V but I have some inside pictures of a Radaler 50. It works from a 9V battery and the supply current is only around 2mA. And it has a transformer and multiplier.
-
Perhaps a charging unit from a photostrobe would work. I've taken them out of disposable cameras and they can get 300 or 400 volts and they run off of a 3 volt disposable battery. Maybe take a look at these:
https://www.goldmine-elec-products.com/products.asp?dept=1480 (https://www.goldmine-elec-products.com/products.asp?dept=1480)
https://www.goldmine-elec-products.com/prodinfo.asp?number=G22744 (https://www.goldmine-elec-products.com/prodinfo.asp?number=G22744)
https://www.goldmine-elec-products.com/prodinfo.asp?number=G22986 (https://www.goldmine-elec-products.com/prodinfo.asp?number=G22986)
https://www.goldmine-elec-products.com/prodinfo.asp?number=G23390 (https://www.goldmine-elec-products.com/prodinfo.asp?number=G23390)
-
@jbb Thank you for the effort! I pretty much built it in dcm without realizing it was an actual thing. I guess Really going in and calculating times and frequencies would only be reasonable if power output was relevant. Since i almost dont care how long it takes to charge i can just go for maximum turn off time.
Voltage division is always a thing of concern to me, though in this case the currents are so low and there is no large changes the internal resistance is enough to even the voltages across all parts. Though I will try to use diodes rated at the full capacitor voltage next time i order some, even if they will survive any ways, higher voltage rating also means lower leak.
@Miti Allways nice to see other ones for reference! The first geigers i built were actually very similar to this one minus the readout part. is it actually a string of zeners for voltage measurement next to the multiplier?
@profdc9 Sadly thats just high power and very low efficiency, im looking for pretty much the opposite :D But yes, you can actually use those for a simple geiger, if you replace the cap with a low leak one you may even get long battery life just charging it from time to time. The nes i know even run of a single 1.5V cell, also something i really like.