My planned detector topology & Voltage regulation for HV.

Linear and non linear

My planned detector topology & Voltage regulation for HV.

Postby chrismb » Mon Aug 29, 2011 2:23 pm

I've [just about] decided on a topology for my circuits for the neutron tube. I will make a shielded socket with short coax length to the preamp box. In the preamp box will be the feedback amp we've been discussing recently, signal conditioning that feeds the signal back onto an HV input, and a rechargeable 9V battery and recharging socket.

A longer length of coax will then run out of this, down to a box with the HV supply inverter in, along with some further signal conditioning to pluck the signal back off the HV line to feed to an internal counter display and out onto an audio line output. I've ordered some shv plugs for the preamp-to-HV supply connection, so that I can drive the preamp with either my box, or my crate, or, indeed, any future 'Geiger'-like supply that is looking for a units-volt pulse on the feed line.

I've decided to run the preamp and HV supply continuously at regulated volts supply, so as to be assured of consistent circuit behaviour while I am still trying to figure everything out. (I can save adventures into battery efficiency for later). This means I won't be introducing any energy-saving measures in this iteration (more to come I am sure, like Doug's scheme of starting the inverter up intermittently).

My options for powering the inverter; I've tests how I need to power it, and I need to feed it with 7V to maintain the correct voltage. I could use PP3 rechargeables. It's a 50mA draw, so two in parallel should last me 6 to 8 hours or so. That's fine, I've no issue with plugging in a charger at the end of a day's work (maybe it'd even be a good thing to go flat, limiting the time I spend on it and sending me to bed at a reasonable hour!!). I would like to use two pp3's as I can package them in one box. BUT, how do I get a reliably steady 7V from a PP3? As the voltage drops, so the voltage of the inverter drops markedly, so how to do with so little overhead for regulator voltage drop? I want dead steady volts so I know where I am at - I want assured repeatability for now. The option at the moment is that I house a 12V lead acid 800mA within the HV inverter supply box, so that I can regulate to 9V and then drop the volts with a plain resistance (prob adjustable, too).

But is there not a better way? In fact, is there a way to regulate the HV line voltage directly? The only scheme for that I could come up with though was to pulse a FET, enabled if the volts on a divider string are above a set point. The fet, via a large resistor, can then drain the capacitors in pulses if the value is reading a bit high. Or maybe I can do the same somehow with the power feed and end up implementing 'energy saving' as a natural consequence of trying to regulate the volts?

Or I could just regulate the input volts to the inverter and hope the output stays constant, and if I do that then how can I achieve a steady 7V for the whole discharge time of the pp3, or should I use 12V?

Summary of the two main questions;
1) How can I best regulate to exactly 7V (pref small adjustments too) from a 9V battery with so little overhead for voltage drop, or best just to go for 12v in that scenario?
2) Can I regulate the HV directly by monitoring a voltage divider potential, and if so how do I best do that (e.g., disable supply current, enable a discharge circuit, put a bipolar transistor somewhere directly into the HV line?)
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 7:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Starfire » Mon Aug 29, 2011 5:12 pm

Chris - if you use a 5v regulator with a resistor in the ground leg you can adjust the regulator o/p voltage up from the (7v) 5v or down to 5v and still retain good regulation - better to go for a 9v (pp3) than a 12v source - you will have less dissipation - tail-off maintenance needs special handling and Li-ion batteries are better
Starfire
 
Posts: 143
Joined: Thu Aug 05, 2010 5:26 pm
Location: North Ireland

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Mon Aug 29, 2011 5:46 pm

You seem to have found one of the least efficient HV supplies for no load current drain. 50 ma is what the JKL ones draw when driving a phototube/resistor load! Mine are running about 20 ma here. What I've found with the ones I use is that they draw less quiescent current (and therefore way less power) if you add another doubler stage to them and run them at lower input. Bingo - dropout problems solved - now you only need 4v or so.

As JohnH suggests, you can use most three terminal regulators in a manner similar to an LM317 with two resistors and get a variable, but I'd skip the common ones and get a low-dropout one instead to do that with, since batteries tend to have such a wide voltage over life range (and the rechargeables tend to be worse). Some of the newer ones have a pnp pass transistor and don't need the extra 2v or so that the common 3 terminal ones require. Youcan't use a 9v battery except at the beginning of its life, end of live volts for one is usually regarded as 5.4v or so (worse with some rechargeable types).

Alternatively, if you can get all your power needs into the 5v range, I'd suggest something like an LTC 3400 switcher chip with a good inductor off a pair of D cells -- more battery for the money, and those will run at very low input voltages, getting all there is out of the batteries. I used these on a body-worn prosthetic that had to drive a piezo transducer to the point of feeling it on the skin along with a PIC uP and 9 RGB leds - plenty of power if the inductor and output cap are "best you can buy" (still cheap). For low current loads like the preamp, a little RC decoupler in the power makes it quiet enough, but do any comparator separately so it's noise doesn't feed back into the preamp and make an oscillator. We did that whole prosthetic thing, including a mic preamp, and a 120v switching driver for the piezo on a 2" sq pcb that also held the batteries....two NiMH AA cells.

As for remote-ing the preamp from the tube -- just say no. Don't do it. Look at Richards, Carls, Doug's, Jon (either one) setup -- and just go with the herd here, we are on to something with that. Richard and I have both had problems using a preamp box the screws directly onto a connector on a 3He tube top - it's that picky - all it takes is a slightly dirty silver plated connector to get into trouble there, or a few inches of extra wire.

The tube has perhaps 25 pf worth of C intrinsic. Double that with a short piece of coax and you cut your signal in half (same charge into twice as stiff a load), while adding a nice antenna for ground loop noises to get into things. This is not the direction you need to be moving in. Here, we find it's wise to put the HV supply in a separate shielded box. We decouple both output leads with series R to the preamp, since those wires need to carry very little current. This avoids pushing the noise the little switcher puts onto the ground line from getting injected into the ground of the preamp. In general you can use 10k to 100 k resistors in series with both HV leads, and let ground be defined by whatever is connected to the preamp output, or in your case, whatever your bench supply does.

This was a really hard one. I *hate* using batteries, but I succumbed due to the advantages in noise rejection of having something that floats perfectly off ground and can go right into the preamp box, avoiding most of the "I didn't think that coax would be such a great noise antenna" as well as ground loop problems. Now this advice is somewhat conditioned to having a real HV supply in the room, that occasionally puts out amp peaks when a tiny arc happens (easy to get well over that from the stored energy in the capacity of the wiring, post-ballast) -- that's 30kw right next to your millivolt thing!

As you noticed, just turning off a lamp made your thing false-pulse. This means you need about 40-60db MORE noise immunity than you have now. If you add a fusor to the mix, that will be another additional 40db or so needed for the output to then mean anything other than it's a great EMI detector.

Adding a divider string to the HV CCFL will add a ton more wattage draw. I don't think you need that for anything but a phototube which is real picky about voltage for its gain -- here you have something that sort of self-regulates anyway. You can't just pulse a fet - amps = dv/dt times C (and with a fet, dv/dt is more or less infinite)...you need to learn switching supply basics and realize you then will need an LC filter for that, along with a catch diode. You can, however (and I have) use a series pass transistor and an opamp with a reference voltage to make an adjustable HV supply off these, I have one on my bench to test detectors with. It's all analog and continuous, low noise. Not very efficient, but then I'm running it off a lead acid gel cel - 12v 7ah in a big box. It took a little fiddling to get the feedback loop closed without making an oscillator, due to the delay through the CCFL and output filter.

You can however, if you control risetimes, switch power to the CCFL on and off, but not so well for these tubes. It's just too much noise, and they actually draw some current between pulses, unlike a geiger tube. You *might* get away with that with maybe a 1 uf HV filter cap...but that's getting dangerous to work around, not to mention physically big. I use .47 uf in some designs here, as I had to buy 500 to get any of that specialized polypropylene pulse qualified part, so they're laying around. Gives one a nasty shock...

As is I usually use about a 10k or 100k series R from the CCFL rectifier/filter to another at least .01 uf cap as a filter to get the ripple on the HV down lower. After that, the decoupling resistors, and another filter cap proximate to the detector tube, right in the box with it. There are enough noise problems without adding ripple on the HV.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 2943
Joined: Wed Jul 14, 2010 8:05 pm
Location: Floyd county, VA, USA

Re: My planned detector topology & Voltage regulation for HV

Postby chrismb » Mon Aug 29, 2011 6:54 pm

Thanks for the comments. Actually the one that hits [maybe stings] is using two stages. Doh! I did originally plan to do that - I didn't think the inverter would give the full 2kV on one stage. Then when I tested it and got good volts at 6-7V I bloomin' forgot to retest for two stages or more!! I'm so forgetful sometimes, I forget why I walked across the room for something these days! I guess I was just hoping to use as few component as poss to help layout.

I did try lifting the voltage regulator, but for some strange reason it didn't like it. I think part of the issue is that if you do that then it becomes a little dependent on the current draw, and as the inverter pulls non-linear current for higher voltages, the 'lifted' earth didn't work. One way I did this in another build was to use diodes to add to the earth, lifting it up by the voltage drop. That didn't work here, either. Not sure why, but think maybe it's something to do with this non-linearity of current draw.

You seem to have found one of the least efficient HV supplies for no load current drain.
Yeah, and the worst of it is how much I paid for these things! 1.50 euros each! Gee, you just don't get quality these days!! ;) Actually, I do regard them as outstandlingly good quality for the money. They are very neat things for what they do. When loaded the run at around 65% efficiency and I tested a few at 30V/25W (nominal 12V/4W!), at which point they are putting out 3000V peak unloaded AC, or 15W into a load, I left a few like that as a test for 30 mins, they got real hot, but no failures! (What killed the ones that have died was having too high a coupling capacitance when an arc flew, or allowing the internal transformer to float away from ground too far inside the can it sits in, both of which have fixes now that I understand those issues.) I have tested a stack of 4 of them to 20kV x 4mA (tested on a 5Mohm resistive load), and still use those for all my current work. They are now my principle HV supply.

In regards noise and cables - I already ran the tubes with the cheap inverter, located 2 foot away along 50cm of wire in both directions! I have tended to keep the tube close to the amp, but I don't think the distance is that critical for these tubes. The balancing act here is that to get a good and practical fit around the tube, it would need to be 'a connector' and it's just a body-in-white problem - I don't see how to engineer a closed box that also fits well around the tube. I'm talking 4 inches of coax here, not feet. It's just to allow the 'socket' to fit to the tube the best chance of doing exactly that. I think I have not had the same issues you've seen because the pulses from this 18-1 tube are so much slower than the ones you've pictures. I mean, we're talking AF here - 10 to 20 kHz, whereas yours are more than an oom quicker. I did see more bounce and more noise with the SNM-10 tubes that had a quarter of the pulse length of the 18-1, so I think I am seeing some of what you are talking about, but it looks to be tube-specific.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 7:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby chrismb » Thu Sep 08, 2011 2:53 pm

Doug, here we were talking about whether it is OK to take a feed from the tube and displace the preamp. You've commented [before] how the additional capacitance of the cable will swap the capacitance of the tube.

I have done a number of tests, and I don't see this happening. I'd like to clarify what you think is the issue. The way I see it, as the tube voltage collapses some little trickle of charge is made available that makes a small voltage drop that is measured across the resistor to ground. If you have an additional coax capacitance between the 'input voltage' and the tube, providing that capacitance is not lossy then surely all that happens is that this additional 'capacitor' gets pulled down by that voltage? Therefore, the voltage pulse should be equally detectable at the other end of that coax?

I have tried with 30cm of my big low loss 4-screened 75 ohm coax (low capacitance per length), and just now I have completed my 'cap' to fit on the end of the tube (I'll post a separate thread on constructing that) using about 60cm of URM76 which my voltmeter (not particularly accurate capacitance meter) seems to be telling me is in agreement with the stated capacitance for such cable (viz, about 60pf added). The tube doesn't register any capacitance at all on the meter, so is substantially lower. But I am getting much the same signal out of it all, no problem, whether I have the length of coax or not.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 7:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Thu Sep 08, 2011 3:20 pm

Well, that's just good -- never hurts when things are a little better than expected. But working from across the pond and with no control over technique, I have to suggest for worst case.

Your premise would be wrong with a constant charge -- adding capacity would reduce the voltage on the pulse. Same amount of amp-seconds into a larger capacitor = less volts.

The fact that you have 75 ohm stuff (less C per foot) is good, and the fact that it's multi screened is better yet. Till you've had to shield against 120db or so you can have no idea how ridiculously hard it becomes past 60 or 80 db. A bud box with all four screws in tight won't do it, not even close. Our coffee can is marginal with 5 screws into the copper top and copper sides we soldered onto it. Connectors *should* be good, but even Richard and Carl report common problems with them due to the tiniest amount of dirt that creates an effective change in series resistance and are fiddling them all the time even with the preamp screwed right onto the tube (no coax at all). Of course, you weren't going to put the HV on that coax or connectors, right? Or you're going to find out just how easy it is to make millivolts out of an electrostatic microphone....and in fact it was an issue here with the "air wired" 50 meg R and 270pf C till I made them stiffer mechanically.

Here the tube doesn't dump a pure charge - it's a somewhat different effect than a proportional tube or phototube due to the already somewhat ionized state of things I'd suppose. So it's not going to be quite as bad. However, just for example, my gamma spec software requires knowing the cable length to 0.1" for calibration purposes...out of 36" max for a far far louder signal out of a large PMT.

Really, noise is going to be your issue here. When I have EMI such that the scope probe on 10x picks up about 3v of it in open air, I start having troubles with false counts from the tube. That's not much noise in a fusor context -- it's nothing to see a couple hundred volts from a tiny arc you can't even hear. As usual, your mileage may vary. I just think you're making extra work for yourself and slowing yourself down doing that. You can't put stuff tight against the tube end anyway due to corona noise, so why not just make the tube cap a little bigger and stuff the preamp in there with it? It's not a big size increment over the huge moderator, that's where I'm coming from.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 2943
Joined: Wed Jul 14, 2010 8:05 pm
Location: Floyd county, VA, USA

Re: My planned detector topology & Voltage regulation for HV

Postby chrismb » Thu Sep 08, 2011 5:05 pm

Doug Coulter wrote:Your premise would be wrong with a constant charge -- adding capacity would reduce the voltage on the pulse. Same amount of amp-seconds into a larger capacitor = less volts.


the thing is, this is assuming that that extra coax capacitance isn't playing a part in generating the signal itself. I mean, that coax would be holding 0.1uC (100uW) which it could dump into the tube when the tube fires. The pulse is >10us long, but the tube doesn't draw more than around 200uA (based on previous testing when I was investigating the tube), so it will only partly discharge some of that charge.

My thoughts are not so much that 'it doesn't work' with a length of coax there, but that I don't understand why it does and whether that additional capacitance is causing some adverse load on the tube.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 7:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Thu Sep 08, 2011 6:32 pm

It depends on how this tube really works internally, which frankly has a bit of mystery around it. If it's a source of charge -- some uC per pulse - then the coax capacity will cause a lower signal voltage during the charge pulse.

If it's more like a neon bulb firing - then the extra capacity means more peak current and total wattage is available for the discharge, which is potentially hard on the tube itself. In that case, as long as it's pretty low additional C, probably no problem - I am already using a smaller coupling cap to what amounts to ground (the preamp input) than specified, so there's some headroom there. The actual tube characteristics here seem to say it's a little of both, so you're both harder on the tube, and get a little bit less signal with the cable.

The real concern is going to be the EMI if you run this near any HV like a fusor. In your other apparatus, probably not much of an issue if it's not jamming the local radios (or doesn't show much on a free floating scope probe). Remember, this signal is smaller than that from a magnetic phono cartridge, yet the impedance is much higher and how much issues one had with hum in phonos. Here, we can pick up noise well above audible frequencies (tube response time again has precisely nothing to do with the preamp's ability to pick up and amplify high frequencies).

uC and uW aren't the same type of units and can't even be related without looking at the voltage, I assume you know this already. If you want reliability, reducing the total connectors is of course your wisest course, and you didn't have to take the design course for military satellites as I did to know that one. If you have volts on the coax, you're going to make a microphone and on this level, a pretty darn sensitive one -- look out for that if you don't have the coupling cap such that it makes the cable center conductor at near-ground. They do make special coax that isn't as microphonic, but they don't make it for HV that I know of (ran into that as a musician, it's a problem there too, even with only a couple volts DC on your cables). They use a special graphite coating inside the shield so the shield moving around the insulation doesn't change the capacity as much. Even that isn't perfect, and laying a microphone cable near the speakers can cause feedback!
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 2943
Joined: Wed Jul 14, 2010 8:05 pm
Location: Floyd county, VA, USA

Re: My planned detector topology & Voltage regulation for HV

Postby chrismb » Thu Sep 08, 2011 6:56 pm

Doug Coulter wrote:It depends on how this tube really works internally, which frankly has a bit of mystery around it.
You got that right!

If it's a source of charge -- some uC per pulse - then the coax capacity will cause a lower signal voltage during the charge pulse. If it's more like a neon bulb firing - then the extra capacity means more peak current and total wattage is available for the discharge, which is potentially hard on the tube itself.

Well, that's what I'm trying to work out by talking about it here. Thing is, if it were an 'extra capacitor' dumping its load into the tube, then wouldn't I expect to see signficantly more voltage-delta than the same 10 to 20mV we expect off of these tubes? Like you say, maybe there's a bit more capacitive current when it fires and a bit less voltage. Seemingly, they cancel out perfectly as the pulses seem unchanged.

In that case, as long as it's pretty low additional C, probably no problem - I am already using a smaller coupling cap to what amounts to ground (the preamp input) than specified, so there's some headroom there. The actual tube characteristics here seem to say it's a little of both, so you're both harder on the tube, and get a little bit less signal with the cable.
On the one hand, I obviously don't want to be hard on the tube, there again if it gets the job done I'm happy. These seem pretty tough anyway. But anyhow I think it is a little moot as the purpose was not to have a huge long lead, just a few inches to make the socket geometry easy and for the line to pass through the moderator. The whole issue of 'geometry' and ease of manufacture is the thing - I couldn't see how to create a fully-enclosing moderator whilst including a preamp and batteries.


uC and uW aren't the same type of units and can't even be related without looking at the voltage, I assume you know this already.
Of course. I was giving the levels for the 2000V operating voltage I'll be running it at, just thought I'd give the ball-park of 'energy' to get a feel for whether any damage could really come from it. Looks pretty benign to me.

If you have volts on the coax, you're going to make a microphone and on this level, a pretty darn sensitive one
That's why I picked this coax - small, and solid PE dielectric. I'll test for microphonics. It's a good call for a test. My immediate impression is that there is nothing noticeably above the noise, let alone the aggressive thresholding I'm using, but I'll be double-checking that. Difficult to check whilst on a bread-board, though, because I might just be knocking the circuit around, until I solder everything up properly.
chrismb
 
Posts: 620
Joined: Thu Aug 05, 2010 7:32 pm

Re: My planned detector topology & Voltage regulation for HV

Postby Doug Coulter » Thu Sep 08, 2011 8:55 pm

No you're not getting what I mean about charge. The tube "pulls out" some charge from whatever the total capacity is (in this case, tube + external stuff). The bigger that is, the less it moves the voltage across the total C for a given number of uC pulled out by the tube. But, as we think, the tube isn't a pure charge-puller -- and as always reality is the key -- if it works, it works. I am working on some improvement on the preamp, BTW, and will report that on the preamp thread somewhere if I get what I want. I'm going to use a different bias arrangement so it works over wider volt ranges with less reduction in output level, and double the gain (so it will drive TTL levels directly). The upshot is that changing the gain won't require re-figuring the bias.
Posting as just me, not as the forum owner. Everything I say is "in my opinion" and YMMV -- which should go for everyone without saying.
User avatar
Doug Coulter
 
Posts: 2943
Joined: Wed Jul 14, 2010 8:05 pm
Location: Floyd county, VA, USA

Next

Return to Analog

Who is online

Users browsing this forum: No registered users and 1 guest