High voltage regulator circuit.
1 Attachment(s)
I've built a high voltage regulator to supply screen grids with 300v, see attached. It's a simple circuit that uses a mosfet as the series pass device and a bjt as a current regulator.
My circuit implementation is different in that there's no bridge rectifier, the circuit is fed by HT which is 420v. I'm using a zener string instead of the pot to give a voltage reference on the gate,I'm using a 50k preset pot to adjust current to the zeners for fine adjustment. Am using a 2SK2545 instead of the IRF740, no problem there AFAIK, the mosfet is rated at 600v/40w, but I'm wondering whether Vds & Vgs are too great. Vds will be 120v Vgs a tad under that. I'm also wondering if there's a better protection circuit, I've used this circuit quite a few times and the mosfet's dies pretty easily if there's too much current draw. TFL, Andy. |
Re: High voltage regulator circuit.
Surely Vgs can only be the few volts required to turn on the MOSFET? As in Vdg being a few volts less than Vds.....
|
Re: High voltage regulator circuit.
Quote:
I have a rule of thumb concerning audio amps/pre-amps which turn up here for repair. If there is a MOSFET HV regulator in them and if the kit actually has any fault at all then the fault will include a blown MOSFET HV regulator. So far the rule has proved very reliable indeed. Much, much more reliable than solid-state devices trying to deal with badly-behaved thermionic ones. EDIT: You could look up 'Maida Regulator' as an introduction to the world of solid-state HV regulation. Cheers, GJ |
Re: High voltage regulator circuit.
"Surely Vgs can only be the few volts required to turn on the MOSFET? As in Vdg being a few volts less than Vds..... " right Chris, brain fart there.
Thanks G. That doesn't bode well though at present the fet runs cool. I've looked at the Maida reg in the past I think and numerous other HV reg circuits, would probably be better with a valve regulator but with a big valve for the series pass it has a large footprint. Is the gate of fets their Achilles heel, would a BJT be less prone to faults? Also should i be looking for a fet with a low gate capacitance or would using a beefier device help, something like a TO247 package? Andy. |
Re: High voltage regulator circuit.
Have you looked at the supplies used in the Sussex valve tester?
This uses an LR8 with Mosfet current boost and incorporates over current protection. Peter |
Re: High voltage regulator circuit.
Quote:
Buried deep in nearly 30 pages (yup !) of posts on a DIYAudio thread on the subject was the advice (I'm remembering what I can) that: Protecting against limited overcurrent can be quite straightforward. But that's very different from protecting against an instantaneous hard short with a large reservoir capacitor behind the FET. If you want to build a genuinely bomb-proof lab-grade supply then study the circuits developed by HP and incorporate the tricks that they built in. At that point I found room for a 6080. Cheers, GJ |
Re: High voltage regulator circuit.
Aaaah, ....6080s!
The mainstay of 60s and 70s 250 and 300 V regulated supplies. We had hundreds of these very forgiving brutes quietly working away. Virtually bullet-proof. |
Re: High voltage regulator circuit.
1 Attachment(s)
Quote:
Looking at MOSFET operation within the circuit attached to OP but instead with 420V supply, current limiting starts to occur at approximately 120mA and any further current load will settle at, say 180mA. At this point the poor old FET is destroying itself trying to dissipate about 76Watts (absolute max = 40W) in constant current mode! Due to three thermal time constants between the die and heatsink the dangerous rise in temperature will permanently damage the silicon. Any protection circuit would have to detect and make safe in <1ms. There is also another caveat with using a MOSFET in constant current mode, that is the rise of RDS bulk resistance with temperature. 2SK2545 RDS increases approximately 2.5 times and this results in runaway power dissipation and is a product of square of the current multiplied by resistance. Rich |
Re: High voltage regulator circuit.
"Have you looked at the supplies used in the Sussex..." No, I will do though, thanks for the heads up Peter.
Thanks G, yep, I've tried to trawl through it too, after a while I zoned out, that's a good summation. There's another protection circuit that cuts power quickly if a valve starts pulling too much current, so hopefully that'll cover that type of fault, however valve regulation looks tempting, no room unfortunately. I did and do look at SOA graph's when using trannies, but is Vds 420v? Vd = 420v but Vs = 300v so surely Vds is 120v? I was going to tweak this circuit by changing R2 after measuring Ig2. thanks for the info on CC operation, I find the thermal info on datasheets hard to get to grips with. I suspect a bigger fet is going to be needed for this job, there's eight 807 screen grids to supply. Thanks for all your IP, Andy. |
Re: High voltage regulator circuit.
To give an idea of how to make something bullet proof I designed a 450V 10A constant current source, this had to cope with a dead short. Final design had 96 TO220 MOSFETS in parallel. Six where made 15 years ago and not one has blown up.
|
Re: High voltage regulator circuit.
By the time your over-current protection operates, into a short, ~200mA (?) the mosfet is sinking ~80W. You could try a device capable of dissipating that, maybe a 250W device with a monster fan cooled heat-sink, but it is then rather over-engineered :)
Better to go for a 'fold-back current limit' - in this case the current limit goes DOWN as the output voltage goes DOWN so limiting the max power in the series mosfet. Plenty of ideas on the web ! In reality you don't need a PSU like this to supply 100mA at 1volt so it would still be usable. dc |
Re: High voltage regulator circuit.
Thanks Dave. I found an interesting protection circuit whilst reading an old book, it used a "voltage dependant relay" which was in a BJT PSU, from memory it was a resistor on the base of the series pass device, the went to one end of the relay coil, the other end being fed by Vcc+. The SW shorted the rails, blowing a fuse, I guess a VDR is a normal relay + resistor.
Andy. |
Re: High voltage regulator circuit.
Quote:
|
Re: High voltage regulator circuit.
Just one observation, the original circuit, back at the start of the thread, isn't actually a regulator. It's a pot with a current booster and a current limiter. Any variation of the input voltage comes through proportionately onto the output.
Mosfets can be reliable, but you can't take any shortcuts with them. Their destruction is fast, very fast. You can't abuse them for even a short time and think 'well, it's not for long' A common misconception is that their gate-to-source resistance is immense, and so you can use very large drive impedances. You can, but only at pure DC. Power Mosfets have surprisingly large input capacitances into the nanofarads region and this gets forgotten when you have your DC goggles on. This can make reaction of regulator devices slow to transients and surges, and by the time the device reacts, something has gone outside its capabilities and has been damaged. If you want them to be responsive, you have to have fairly low impedance driver circuits. Mosfet gate oxide is fragile, ratings are in the 8-20v range for power devices. Bipolar transistors are also very fragile. Reverse bias of the base-emitter junction is the easiest way to kill one. Reverse b-e ratings are typically not much more than 5 volts. In this direction it's a reverse biased diode, so the resistance is very high, and the capacitance is very small compared to the mosfet. Th what extent the abused device lives depends on the current available when the junction avalanches. Low currents do damage, reduction in gain and Ft happens the transistor will never be the same again. Multiple events do cululative damage. Higher currents will do immediate and total destruction. You can use the b-e junction reverse biased at low current avalanche as a noise generator. With an RF transistor the capacitance can be low and the output can go to UHF. It's a common trick in amateur radio circles for assessing the noise figure of receivers, but the transistors used this way are ruined for normal use and their noise level changes with use as well. Back to Mosfets, the original circuit can be modified to be more robust. Firstly it needs a big enough device or bank of devices (with current sharing resistors) to handle the fault condition. The drive resistors and potentiometer resistance will need to come down in value. You need to also recognise that ordinary zeners are not fast. Quite sluggish and quite high capacitance. One trick is to use a properly fast diode to a point which is regulated by a zener, where the zener has a bit of current applied to it to set it up ready for action. I have tens of thousands of mosfet transmitters out in the field doing pulsed operation approaching 500W dissipation during the pulse. I cannot for a picosecond let any voltage rating be exceeded, or any current rating. The devices have power rating specially for this application. Average dissipation is just a couple of watts. Very little heatsinking is needed, but the devices include enough thermal mass to handle the pulse. Should anything go wrong and the pulse last too long, the device fries. VERY expensively. But, with care, it can be done and millions of device-hours done so far without any design issues. Oh, and they have to survive induced lightning surge tests :-) The pulse modulator mosfet is chopping 50v at up to 12A in nanoseconds. The drive circuit has an impedance of only a few tens of ohms. So rule number 1: Don't even think about short cuts or 'keeping it simple' Rule number 2: Mosfet gates are high resistance items, they are NOT high impedance. Rule number 3: Ordinary Zeners are too thick and slow as bouncers. Rule number 4: Mosfets have infinite gain at true DC, but this comes crashing down when there are AC components and transients on the loose, so circuits behave quite differently to how we look at them with our DC goggles on. David |
Re: High voltage regulator circuit.
And simulate with ltSpice or the like, not perfect but a good “understanding generator”. Can save you from schoolboy errors, saved me many times.
|
Re: High voltage regulator circuit.
1 Attachment(s)
Quote:
Quote:
But it will not surivive *sustained* shorts. If you want to survive sustained shorts then add a 100mA fuse and don't be clumsy. |
Re: High voltage regulator circuit.
Use a voltage doubler instead of a bridge rectifier.
You then have some intrinsic current limiting so the MOSFET does not have to work so hard in short circuit situations. |
Re: High voltage regulator circuit.
Quote:
|
Re: High voltage regulator circuit.
Quote:
I speak from experience. I once worked on a large power amp (150 W/ch) which had an HT voltage of ~580V if I remember rightly. The manufacturer had very properly fitted an HT fuse rated for a higher voltage than this. It was about the size of the last two joints of my little finger and each one cost more than a fiver. The fault that the amp came in with, of course, caused it to blow these fuses. To try to keep costs under control I temporarily replaced the fuseholder with a 0.25" x 1.25" open one and fitted a glass-bodied fuse. The first one failed explosively (very exciting !). So I boxed the assembly to catch the flying glass. After a few had gone bang one tracked, allowing the fault current to continue flowing via an arc on the inside of the glass which this time hadn't shattered. "Every day's a school day" as they say. Cheers, GJ |
Re: High voltage regulator circuit.
AFAIK, the above circuit can be turned into a 'foldback' limiter with just 1 more resistor!
Increase the sense resistor to 68R and connect a 62K resistor from Q2 base to GND. With the output shorted, 10mA through the sense resistor gives 0.68V which will will turn on Q2. At a 400V output, 100mA through the sense resistor will give 6.8V but this is divided by 1K/62K turning on Q2. I didn't build it ! Nor did I sim, YMMV dc |
Re: High voltage regulator circuit.
Quote:
The reason I go for a doubler is that it limits the current even if the MOSFET melts down to a short. But I now see that this circuit is not being run off the mains - so not relevant for that situation. |
Re: High voltage regulator circuit.
Thanks all. Posted replies but the post seems to have disappeared, that or I didn't click submit. Anyhoo, I'm plugging away at this, have several versions including the Sussex PSU jotted down and will post again when I've had a play.
Andy. |
Re: High voltage regulator circuit.
2 Attachment(s)
Right, I've done some more work on this and tried a few circuits. First off I built the anode/screen supply out of the Sussex. Essentially its the same as the one I built but it has a CCS supplying a zener string, this gives us a hopefully stable reference voltage, this goes into a voltage divider, which sets gate/OP voltage.
It worked pretty much first time, but like mine isn't stable, OP V drifts a little as the circuit warms and V out drops with increased current demand. Next I tried a few floating regulator types, a few based on the LM317 application notes on HV voltage regulator's, sometimes called a Maida regulator on DIY audio. I also re-visited a similar circuit out of The Art of Electronics that used a LDO LT3080. All use a similar approach, a mosfet to drop excess V followed by a floating regulator. I built three circuits, The AOE one, the 21st Maida and the LM317 HV reg. I'd built the AOE one previously and found it unstable, this was my fault, I left off a 2.2u cap off the OP. This time I found a 10u 450v film cap in my stash, this improved things, but the circuit was still unstable, everytime I tried to adjust it, the OP oscillated. All of these suffered from the same fault too, adjustment resulted in the OP jumping to the lowest V setting, then no I was able to make no more adjustment. All of these circuits were a pain to get working. So, I write these floating types off, due to their instability and the need for massive brick size expensive low ESR. HV film caps. Conclusions, for a simple HV reg the 1st circuit and the Sussex version works ok, but they're not 100%. To make a proper HV reg I think a more complex approach is needed, IE opamp foldback which means a separate PSU. For what I need at present this is a bit OTT and I haven't the room. In future I'd definately go down the valve route; valves laugh at HV & HV V reference valves give stable results, a string of zeners drift too much with temp. Just my inexpert view, Andy. |
Re: High voltage regulator circuit.
Quote:
IME it is better to build the Maida reg with no capacitors on the regulator IC 'in' or 'adj' pins. Those capacitors just create problems. |
Re: High voltage regulator circuit.
Actual ZENER diodes are only up to a couple of volts. Above that you get avalanche diodes although everyone still calls them Zeners. What goes on inside the diode may be different, but how they look to the world is very similar.
Avalanche diodes have significant temperature coefficients that depend on their voltage. What's interesting is that it crosses over from a positive to a negative signed coefficient a little above 5v. So 'Zeners' around 5v ara much more stable, as a percentage, over temperature This is why you see designs all over the place using 5.1v zeners as references theat get multiplied up by resistor ratos (Nice stable resistors, Mmmmm :-) ) Motorola Made a family of 'Voltage Reference Diodes' with very low tempco at several voltages. Inside each was the avalanche diode and a straightforward silicon diode in series. The avalanche part was engineered to have a controlled tempco that matched and cancelled that of the plain diode. The resulting thing was like a 'zener' but with a most unusually low tempco. 1N823A is the 6.2v part, probably the best one of the series. Like any other zener, you have to pay attention to specified current for accuracy and tempco best behaviour. They were the best reference there was until the three terminal IC references using bandgap references came along. IC fabricators also sometimes use 'buried zeners' as references. David |
Re: High voltage regulator circuit.
Thanks both. Re zeners, or avalance type diodes, I wonder if a PTC thermistor or maybe an LED and LDR would serve to adjust for temp? I know you can get better voltage references but there are non I'm aware of that'll do a few hundred volts. Having said all that it's probably best to use gas valve VR in combo with SS regulation. Probably talking out of my hat, just pondering.
I'll keep dipping into this subject and do some more testing etc when the chance occurs., Andy. |
Re: High voltage regulator circuit.
Quote:
|
Re: High voltage regulator circuit.
Yea, the original design (and I agree, pretty much the same as the one I saw from the Sussex) will drift thermally and with load current. My mod for 'fold-back' would make that much worse since it increases the output resistance (in order to get enough voltage drop to reduce the current into a short circuit).
To get better performance you will need feedback but that doesn't have to be an op-amp :idea: dc |
Re: High voltage regulator circuit.
+/- 5% David according to yon datasheet for the 807. However this is for AB2 conditions, so I don't need that level of accuracy right now but while I'm at it.... "My guess is you are seeing the effects of the current-limiting resistor, not Zener variation" could be.
"To get better performance you will need feedback but that doesn't have to be an op-amp" I'm all ears Dave 2. Andy. |
Re: High voltage regulator circuit.
Quote:
Look up GAP - George A Philbrick. Rob Pease cut his teeth working there. Then came discrete transistor designs, then monolithic ones. David |
Re: High voltage regulator circuit.
1 Attachment(s)
Andy, I knocked this up quickly last night, just an idea to see what it would involve without using too many parts. Looking at it now I could probably use 1 less transistor.
There are multiple problems as it is. it will only work down to ~35V and the regulation is not brilliant (~ a volt at 350V). Having 350V across a pot anyone ? I didn't have a MOSFET model so used a bipolar for the series element. The top right current source will be a pain, the others can be replaced with resistors (one sinking several watts). There must be a better way ... dc PS I tried to attach the model but I couldn't upload it |
Re: High voltage regulator circuit.
1 Attachment(s)
The fog is clearing ...
LTSpice model attached. |
Re: High voltage regulator circuit.
A few more issues ...
Should be a 6V2 zener not 6V8 for best thermal drift, but in any case there is not enough current through it to work properly. Ditto the 30V zener. More current is a problem as its being pulled rather wastefully from 400V. 1V2 shunt regulators are good and cheap (band-gap) BUT that wouldn't leave any headroom for the long tailed pair. Is there a cheap / available ~ 10V shunt regulator ? Or even a 30V ? Maybe Q2 would get killed by a shorted load :) dc |
Re: High voltage regulator circuit.
Could the reference diode be replaced by a TL431?
Take care if you build a circuit with high voltage on a variable resistor and make sure it can handle the power dissipation in the resistive element. Reverse biased power diodes across the output terminals and the pass transistor are usually recommended. |
Re: High voltage regulator circuit.
TAA550 is a varicap tuning supply shunt regulator IC, temperature compensated 33V. A bit old hat but still around. High voltage "Zeners" have rotten temperature stability, you need to try them in a test rig to appreciate just how bad they are even well within their dissipation rating.
With PSUs, it's wise to imagine what happens if a variable resistor wiper goes open circuit (or maximum resistance if rheostat-connected to one end of the track)- generally, output going low is preferable to output going high! |
Re: High voltage regulator circuit.
Yea,
TL431 could replace the 6V2 zener (6V8 on the schematic) Should work at 2mA. Drop the ref to 2.5V so no additional resistors needed. Let R3=2K5. The pot with the voltage is NOT nice :) dc |
Re: High voltage regulator circuit.
Quote:
|
Re: High voltage regulator circuit.
3 Attachment(s)
Thanks Dave. Is Q3/4 a current mirror? i'll try and find time to do some work on this but have a tfmr to rewind by hand today. I've noted all the IP, have a TL341 etc. Re pots across HV, the schematic I posted back in post #1 (re- posted here)uses a pot across 350v, a 470/500k. I've used two of these PSU's in a homebrew tester with no issues. i put two 10M from wiper to each end if I remember right.
Got you David, the Sussex PSU is essentially the same as the attached schematic, but uses a CCS, zener (avalanche) string to get a HV V ref then uses a potential divider instead of the pot. I only briefly tried it, saw how the zeners drifted, moved on. David said of the circuit " IT isn't actually a regulator. It's a pot with a current booster and a current limiter. Any variation of the input voltage comes through proportionately onto the output." I thought it possible to do a bit better by adding foldback. i'll admit to not being 100% on this stuff but have some idea, thanks for your concern though. Elektor published such a circuit that has a opamp and Farnell used a similar approach in their E350 ( though that use valves as the series pass element) and uses discreet FB circuit. If I can get Dave's schematic working it would be nice. Andy. |
Re: High voltage regulator circuit.
Quote:
"Version Nine, a daughter board (LR8 board) is mounted on main PCB which removes the series Zener Diodes, ie ZD1...ZD7 arrangement and uses a high-voltage, low-output current adjustable linear regulator (U1). to control two Power MOSFETs (TR2 & TR4) for the Anode and Screen grid voltages." The Schematic is in this post https://www.vintage-radio.net/forum/...postcount=1477 Peter |
Re: High voltage regulator circuit.
Thanks Peter for the 'heads up' on the amendment to the Sussex,
I think both Andy and I had only seen the original. dc |
Re: High voltage regulator circuit.
Andy,
Q5/6 is a long tailed pair (LTP). Q3/4 a current mirror with Q2 a 'current through-er' (cascode) :) Effectively, the current source I1 is a collector load for half the LTP. The mirror / cascode are really only there to steer the control current to the right place and avoid issues with the poor performance of high voltage transistors! dc |
Re: High voltage regulator circuit.
Ahhh, 1 transistor less and better regulation - replace Q3 with 430R :idea:
|
Re: High voltage regulator circuit.
Thanks Dave.
A. |
Re: High voltage regulator circuit.
1 Attachment(s)
Thanks Martin for the later Sussex HT/Screen grid supply, however am unsure what those "pads" labelled 1 , 2 & 3 under "JAN8W" (if that is what it is, hard to see), in other words where does the mosfet gate go, through that 1m resistor to...? On the old version it goes to a potential divider, which sets V out, so I presume it's a pot in this case.
I'm pootling away at this in the background, will post again when i have something to report, Andy. |
Re: High voltage regulator circuit.
Same confusion here Andy !
I would guess the regulator feeds the divider chain of switched resistors with the gate being connected to the tap (or a pot, as you suggest). In that case the major error contributions will be : (1) The LR8 regulator (but with a nearly fixed load) (2) Gate/source voltage drop, at delivered current / temp / device (3) Current sense series R (I can see 2 but I don't know which is which) Better, but still not a regulator ! dc |
Re: High voltage regulator circuit.
3 Attachment(s)
Quote:
The pads 123 go the original resistor chain of the Sussex, the LR8 just replaces the zener chain. When I built my version I dumped the resistor chain and connected the 1M from the Mosfet gate to the LR8 output. I then used the pot in the LR8 circuit to set the output volts Its Peter by the way, not Martin....... |
Re: High voltage regulator circuit.
2 Attachment(s)
Updated design : including TL431 ref, current mirror replaced by common emitter & MOSFET model / output. LTSpice file included.
dc |
Re: High voltage regulator circuit.
Thanks again Dave, now the amp is finished I'll have some room ont bench to fettle this.
Andy. |
Re: High voltage regulator circuit.
Hi Dave, I'm breadboarding this circuit today, a few questions. Re V2, why does the base of Q6 have to be at 2.5v, and do I really have to use a TL431, I hate these bl**dy devices with a passion. The resistors are always ones I'll never have so far I calculated 1r and 3.119r with a 2k2 supply resistor, can't I just use a diode instead? A blue diode gives me 2.65v, then i can just tweek R6, yes?
Re D2, does that read 24v? Re I3, that can just be a resistor yes? I've worked this out to 93.k, so could just use a preset and fixed R. I've sussed I4, I'm using a MJE350 CCS. Lastly as i understand the circuit if Q5's base sees a change in current, Q6 will go the other way switching on Q4 harder (or softer)that changes V on the gate of the fet. Q2/Q7 regulate Vgs. At present I'll run this at 100v, then when it's stopped smoking, crank Vin up. Andy. |
Re: High voltage regulator circuit.
Andy, IIRC the TL431 is 2.5V without any resistors!
But yes, an LED would work and adjust the R3 a bit on the other side, it will have the same voltage across it. I choose R6 to balance the current in the 2 legs of the LTP Q5/6 but its not critical. I specified D2 at 30V but 24 will be fine. Monitor it in any case, it probably wont be that stable but it should not matter too much ! Yea, if the output voltage goes up too much, Q5 turns off, Q6 on then Q4 on (throught Q2) pulling the mosfets gate down by pulling current through R7. Note, current can also be pulled through R7 by Q7 if the load is getting too much current. Good luck Andy ! dc |
All times are GMT +1. The time now is 9:51 pm. |
Powered by vBulletin®
Copyright ©2000 - 2024, vBulletin Solutions, Inc.
Copyright ©2002 - 2023, Paul Stenning.