Need input on high voltage regulator design

[Frank Bemelman]

One of the circuits I based this on had a similar setup, but it just kept
the primary driver disabled while it was on. I tested it and it didn't seem
to do much good. I may experiment with this some more, I'm just a bit
worried that having only the single reference might make it hard to maintain
regulation.

Seems like the shunt would still be burning off power even in that setup -
though I guess it would depend on what kind of current limiting resistor I
put inline. Is that right?

I have no idea about the varistor, how sharp their knee is. In your
current setup you must have given the PWM a safe (too large) value
to make sure you are getting your 500V and have the varistors do
their job. Well past the knee. Adding the transistor, you could sit
somewhere halfway the knees.

Perhaps software can help a bit. As soon as you reach the ~500V,
and the transistor is on, turn off the drive. Turn on the drive
again, not before a long period of idle time has passed, or when
the tube has given a couple of pulses.

There's still the matter of the limited lifespan on the varistors... not
sure they should really be there at all.


Right. The version I built for myself ran off two AAs with a charge pump to
drive the LCD and MCU, but going with 9 volts lets me replace that with a
much cheaper linear regulator and reduce the number of multiplier stages.

Did you try a simple single inductor, with just one diode and cap? It
seems that a small 330uH inductor and pulsing it with 25uS would also
give you 500V in no time. You need a better fet of course. Or use more
windings on the transformer and get rid of the multiplier

 

[Robert Baer]

Hey, that's pretty slick. I'll have to try that. I know the A/D will work
with 10K impedance. I was also thinking of trying a comparator, but the
buffered A/D scheme only uses the one input pin, and at < 50 cents for the
TL081 that sounds like a cheap, flexible solution. Though it'd be nice to
find some cheap carbon film resistors in > 10 megohms to reduce the parts
count. The high-value thick film resistors go for about $4 at Digi-Key.

I'll do a full writeup on this project when I get a chance. Most of the
hobby kits and published designs I've seen have been pretty simplistic, with
just a speaker and maybe a meter, and the kits sell for about $150 and up.
Mine's got an LCD display and serial interface, and can be scrounged
together for under $50 in its current form.

Thanks for the input...

Scott

...and what is wrong with using a number of cheap SMD 1% 10 meggers?

 

[Robert Baer]

I have no idea about the varistor, how sharp their knee is. In your
current setup you must have given the PWM a safe (too large) value
to make sure you are getting your 500V and have the varistors do
their job. Well past the knee. Adding the transistor, you could sit
somewhere halfway the knees.

Perhaps software can help a bit. As soon as you reach the ~500V,
and the transistor is on, turn off the drive. Turn on the drive
again, not before a long period of idle time has passed, or when
the tube has given a couple of pulses.


Did you try a simple single inductor, with just one diode and cap? It
seems that a small 330uH inductor and pulsing it with 25uS would also
give you 500V in no time. You need a better fet of course. Or use more
windings on the transformer and get rid of the multiplier

MOVs, varistors, TVSes, etc all seem to be roughly exponential, so
there is no "knee".
A temperature compensated 500V zener would be far better, especially
if one could get one that was not noizy from nA to a few mA or so.
Using such a beast as a "reference" or the high voltage drop component
in a divider in the feedback might do the trick.
Below 500V the drain would be zero, and one could work in the 100nA to
10uA region for rather low power loading, making it act as a shunt
ergulator.
If you do not need a 185C part, then I could make some for you - not
cheap but not expensive either (cheaper by the thousands).

 

[Winfield Hill]

Robert Baer wrote...

A temperature compensated 500V zener would be far better...

Given the device physics, that would have to be made from 89
5.6V zener dies in series. :>) (If a zener reference was
desired, it would make more sense to attenuate the HV down
to 5.6V, compare and regulate.)

 

[Joerg]

Hi Frank,

Would I? I was hoping that when below 500V the current passing the
two 500V vdr's is so small that it does not turn on the transistor.

That would be a pretty scary circuit. Leakage might keep it turned on
and at the other end even the slightest overshoot could fry the transistor.

In the current setup, the pwm needs to have a safe overdimensioning
large enough keep those VDR's regulating, and wasting power. With
the transistor, you could back off the PWM, keeping it on the edge,
with the transistor turning off/on etc. With all the soft knees of
the VDR and transistor, you'd still waste power, but perhaps not
as much.

Most voltage dependent devices these days are MOVs which are not at all
suitable in any situation where they have to permanently conduct. Even
if it's just a little. They are similar to brake pads and wear out.

That seems the case. Perhaps it is better to use a comparator
or opamp + adc input and very high value resistors.

Yes, but I probably wouldn't waste an ADC input on that unless it's
there anyways. A comparator input should be fine. uCs with ADC on board
tend to be expensive.

Regards, Joerg

 

[Joerg]

Just to add to Graham's comments about doing this as an inverter application without feedback: If you were to scrap the regulator and, say, operate the whole thing from three AA cells (assuming a uC that would be happy with that) you could run this transformer bipolar. One side switched to VCC, the other to GND, then reverse.

This is off the cuff and you have to check that nothing gets fried: Since it is such a light load, maybe two uC pins could provide enough drive for that? You may need a couple of Schottky pairs to prevent spikes above VCC and below GND if the uC doesn't have enough protection. And maybe a cap or a routine that makes sure that 'rest' is either both port pins high or both low.

If the uC can't do that, maybe a stiff 74HC driver chip could.

Regards, Joerg

 

[Frank Bemelman]

Hi Frank,


That would be a pretty scary circuit. Leakage might keep it turned on
and at the other end even the slightest overshoot could fry the

transistor.

The resistor in the collector has to be low enough so that hfe * leakage
is of no importance - and there was a 470K resistor in series with
the voltage doubler ouput, so I can't imagine frying anything, really.

Most voltage dependent devices these days are MOVs which are not at all
suitable in any situation where they have to permanently conduct. Even
if it's just a little. They are similar to brake pads and wear out.

That rules them out, then, if it happens sooner than the life
expectance of the entire device.

Yes, but I probably wouldn't waste an ADC input on that unless it's
there anyways. A comparator input should be fine. uCs with ADC on board
tend to be expensive.

A comparator is okay, but sometimes they are giving away analog inputs
for free, almost. A few weeks ago I selected a PIC16F88 (7 multiplexed adc
inputs, 10 bit). Another version, the 16F87 is identical but only has
digital inputs. The 16F88 is $2.41, the 16F87 is $2.26. That's 15 cents
for the adc, not bad I didn't need the adc, but since I only need
to build 50 pieces, I choosed the 16F88, in case I get a crazy idea.

 

[normanstrong]

I got bored yesterday and started working again on a Geiger counter design I
came up with a couple of years ago. The high voltage power supply was
mostly cribbed from a couple of other designs I found, since I'm a digital
guy and high voltage isn't my thing. As it stands now, the microcontroller
puts out a square wave that drives the primary of a transformer via a
MOSFET. A voltage doubler on the secondary gets the voltage into the
required range (500+ volts in this case), and a couple of varistors in
series form a shunt regulator.

It's the shunt regulator I'm looking to replace right now. Since the input
to the inverter is provided by the MCU's timer channel, it can be easily
pulse-width modulated. I just need to get some feedback from the HV side so
the MCU knows how to adjust the duty cycle. My first thought was to use a
pair of resistors to form a voltage divider and feed maybe 1/120 of the
output voltage into the MCU's A/D converter. But with any reasonable
resistor values, there's too much load on the supply to maintain the needed
voltage. Higher values might work, but the thick film resistors are
expensive and I'm not sure how much current the ADC needs to function
correctly.

Any suggestions? I'm thinking an op amp might help, with its high input
impedance. That still means using expensive thick film resistors, though.
And again, I'm a digital guy and I'm a bit out of my element here.

I'm really trying to keep the cost and component count down to make it an
easy hobby project, but I'd also like it to not suck a battery dry in 2
hours. If anyone's got any suggestions on how to implement a better
regulator, or maybe a different supply design entirely, let me know.

Current schematic is at: http://n1vg.net/geiger/images/gc1-9v.png. The HV
supply is on the right-hand side - everything from MCU pin 13 to the
Geiger-Mueller tube. The pin header at left is for an LCD display.

My recommendation would be to look at some of the recent portable
electronic photoflash designs. The application is similar and there's
a great emphasis on economy and long battery life. Unless you need a
very closely controlled voltage, I'd probably look at a flyback supply
rather than a square wave with voltage doubler.

Norm Strong

 

[Joerg]

Hi Frank,

The resistor in the collector has to be low enough so that hfe * leakage
is of no importance - and there was a 470K resistor in series with
the voltage doubler ouput, so I can't imagine frying anything, really.

You are right, the transistor won't fry in this application. But it
would still be a mushy regulator because you are relying quite a bit on
hfe. Voltage control is a lot less mushy.

A comparator is okay, but sometimes they are giving away analog inputs
for free, almost. A few weeks ago I selected a PIC16F88 (7 multiplexed adc
inputs, 10 bit). Another version, the 16F87 is identical but only has
digital inputs. The 16F88 is $2.41, the 16F87 is $2.26. That's 15 cents
for the adc, not bad I didn't need the adc, but since I only need
to build 50 pieces, I choosed the 16F88, in case I get a crazy idea.

$2.41 is a lot of money for a uC. But it seems that Microchip isn't
charging such a premium for the privilege of having an ADC. That is
different with other brands. The MSP, for example, goes from $1 to $2 if
you want a converter.

Regards, Joerg

 

[Rich Grise]

.

I'm really trying to keep the cost and component count down to make it an
easy hobby project, but I'd also like it to not suck a battery dry in 2
hours. If anyone's got any suggestions on how to implement a better
regulator, or maybe a different supply design entirely, let me know.

Current schematic is at: http://n1vg.net/geiger/images/gc1-9v.png. The HV
supply is on the right-hand side - everything from MCU pin 13 to the
Geiger-Mueller tube. The pin header at left is for an LCD display.

I vote you don't need to regulate it at all. Find a very-high-impedance
meter and measure the voltage, then just adjust your pulse width so it's
500, and be done with it. A geiger tube doesn't take very much current at
all, even when it fires.

I'd also look at the transformer polarity, since it is kind of a flyback
design. I saw one very much like this in a NASA tech brief, which they
called a "forback" since it's kind of a morphodite of forward and flyback
converters.

Good Luck!
Rich

 

[Terry Given]

I'll take a look.




Isn't it? It's using a transformer with isolated windings. Doesn't that
make it an isolated flyback converter? Or does that not count because it
feeds back into the +9v rail through the shunt?

Scott

Its not isolated because the primary and secondary circuits share a
common "ground"

Cheers
Terry

 

[Terry Given]

Indeed it does... draws about 40 or 50 ma, with < 10 of that being the MCU
and display. Didn't know that about MOVs - one more reason to find another
design.

little dutch timebomb, tick-tock-boom.

Thanks,

Scott

Cheers
Terry

 

[Terry Given]

Robert Baer wrote...



Given the device physics, that would have to be made from 89
5.6V zener dies in series. :>) (If a zener reference was
desired, it would make more sense to attenuate the HV down
to 5.6V, compare and regulate.)

#define zener avalanche diode

I confess, I do it too.

Actually, I once debugged some code that contained:
#define x a
#define y b
#define d a
#define q b

It had some interesting side effects.
(three co-ordinate systems - xy, ab, dq. And a programmer who didnt know
about unions)

Mind you, that code also had a giant static array of constants which was
generated by macros (ie at compile time), including generating the
variable names (it was a table of variable parameters - max, min,
default etc). So a text search on the variable name cannot find the
table entry. That was incredibly frustrating - so much so we ended up
running the pre-compiler, keeping the output as our new source code, and
throwing away all the macros.

Cheers
Terry

 

[Robert Baer]

Robert Baer wrote...

Given the device physics, that would have to be made from 89
5.6V zener dies in series. :>) (If a zener reference was
desired, it would make more sense to attenuate the HV down
to 5.6V, compare and regulate.)

I use a dual 8V unit giving 17V after (some) TC.

 

[Joerg]

Hi Terry,

little dutch timebomb, tick-tock-boom.

They might not go boom in this app because there is little energy
available but they'll just eat themselves away.

Question: Why is this one 'Dutch'?

Regards, Joerg

 

[Terry Given]

Hi Terry,




They might not go boom in this app because there is little energy
available but they'll just eat themselves away.

Question: Why is this one 'Dutch'?

Regards, Joerg

Hi Joerg,

I have no idea where I first heard that saying, but thats how I heard
it. Working for 10 years on high power electronics with a dutch engineer
made it very appropriate

I saw a guy use a MOV as a leakage clamp in a 50W smps once. He ignored
my sage advice, and consequently suffered a 100% failure rate -
catastrophic failures at that, enough to write the whole PCB off (a
dozen units). The re-design used an avalanche diode (which ran cool).

Cheers
Terry

 

[Frithiof Andreas Jensen]

I'm really trying to keep the cost and component count down to make it an
easy hobby project, but I'd also like it to not suck a battery dry in 2
hours. If anyone's got any suggestions on how to implement a better
regulator, or maybe a different supply design entirely, let me know.

Why not a Bang-Bang controller? How god does this have to be, anyway?

Basically, one uses a comparator with hysteresis to switch the HV Supply
on/off, between two set limits.
The filter capacitor will make sure that the total duty-cycle of the power
supply is very low, thus power
consumption will be low too. Easy.