Low current HV inverter: choice of caps/diodes etc.

On a sunny day (Fri, 13 Nov 2009 08:48:52 -0800 (PST)) it happened Julian Bunn wrote in :

1N4007 is not very fast IIRC, better get some fast diodes if you use a high frequency. 50 nF points to a high frequency...

How much ripple can you tolerate?

As frequency goes up, the caps get less time to discharge between periods, so a smaller value will then do.

Maybe some sort of flyback switcher would give better results... What does your circuit look like?

Bunn

gh frequency.

Right now, I'm running at around 50kHz, but I can easily use a lower or higher frequency. I did wonder whether I should be using a much lower (more like a mains) frequency, since perhaps the transformer is not designed for such high frequencies.

A lot: this is for driving a single indicator tube.

,

I've tried several (!) but right now it looks like a 555 timer with a power NPN Darlington's base attached via a 390Ohm resistor to the clock output. The 6-0-6 winding of the transformer is attached between the collector and +9V.

On the output windings I have a standard doubler: a couple of .05uF caps and a couple of 1N4007 diodes.

I'd like to avoid special transformers like flyback if possible. My impression is that my problem is not there anyway, it's with the rectifying/ doubling.

Many thanks, Julian

On a sunny day (Fri, 13 Nov 2009 09:15:00 -0800 (PST)) it happened Julian Bunn wrote in :

Yes, could be you are just warming up the iron in the transformer... Did you measure the voltage on the 9 V battery while it was running? Due to possible losses I would expect it to be less then 9 V.

Q = C x U = i x t, so at 50 kHz t = 20 uS. C = 50 nF, If i = 1 mA then the riplle will at least be U = i x t / C = 10^-3 x 2.10^-5 /

5.10^-8 = 4 V. For a 2x more current 2 x more ripple, etc.

There a 2 basic ways to double the voltage, one like this: |\\| --- ||----------| |--------- + ||( | |/| | ||( --- | ||( / \\ === | --- | | | | /// /// ///

Or:

T sec |\\| ---------| |----------- + ||( | |/| | ||( | | ||( | === ----)---------------| | | | === | | | |/| | ---| |---------| |\\| | /// It may make some difference depending how the waveform is.

If you are using some old filament transformer, these might be good for a few hundred Hz at most.

Many older output transformers in tube amplifiers suffered a lot from stray capacitances at frequencies below 20 kHz.

frequency.

I would suggest dropping the frequency considerably and increase the capacitor size correspondingly.

Paul

Linear Tech has some photoflash app notes.

John

ian Bunn

I d

high frequency.

Thanks Paul, Jan, John.

I'll try a lower frequency, larger caps, and the other doubler circuit suggested (the lower of Jan's schematics).

Thanks for the tip about the Linear Tech app notes!

Julian

First off, if you are only talking a few mA, you won't need to store much energy, so your multiplier caps can likely be smaller.

If you do not step up the secondary so much, you could use two multiplier stages, and get a bit more control over your final output capabilities.

After the multiplier stage(s), you can place capacitance across the output to store a bit more, and smooth what you have a bit more.

The diode choice is fine.

So, there is no reason not to get the multiplication other than it must be wired up wrong.

Also, you cannot use polarized caps. They should be ceramic or poly, which, have a band marking, not for polarity, but to show where the outer foil side is wrapped (terminated) to.

Ceramic is easier. Shorted HV caps, however, can appear as 'good' via normal testing modes. They only show failed when they 'fail' to work in an HV circuit. Typically, they fail appearing as a short to the circuit above a certain voltage, and can hold a charge below it. The dielectric has been pierced.

You must have gotten the circuit wrong, or one of the caps is blown.

Maybe one of the diodes could be bad, but less likely at these voltages. Though you could try a fresh pair in your thoroughly audited assembly. :-)

Cut off the right hand side, unless you want a pulse generator and regulator. Transformer is on ferrite, maybe 20uH primary.

Tim

I posted a zener tester circuit on abse a week or so ago. It produces about 170 VDC at about 8 mA, using a 9V battery as the source, so it should work for your app. You won't need the output transistors if you don't need current regulation, but you may need to add a voltage regulator.

Ed

Thanks for all the excellent advice.

I replaced the caps (disc shaped, orange in colour, not sure of the construction), by some more modern types (polyester, I think: yellow brick shaped things from Mouser, rated at 630V) and the voltage shot up to over 200VDC! I also reduced the frequency to around 500Hz - it seemed to improve stability.

So, problem solved :-)

Julian

Part of your losses will be in the sharp edges you are putting onto the transformer. Part will be into the storage time of the 1N4007s. Some will be in the leakage of the 1N4007s.

You may be better off with 1N4946 (IIRC) fast rectifiers intended for

1A or less service.

You can reduce the energy that is lost into the core from the high frequencies if you put something like this in the center tap leg.

L1 Supply----+----))))))-------+-------------Center tap ! ! ! =3D=3D=3D C1 ! D1 ! L2 ------!

"MooseFET" Julian Bunn

--------------------------------------------------

** The leakage inductance of the tranny the OP has chosen to use is HUGE !!!

So much so, that efficient *sine wave* operation is possible only to about

1 kHz.

** Hopeless at anything like 50kHz. ** Absolutely microscopic at these voltages. ** The OP has solved his problem by reducing the frequency to 500Hz - where core losses are tiny and leakage I is not too serious.

..... Phil