My Variable Frequency HV Power Supply

It's a very safe power supply -- I assume the 2.2k was intended to be a pull-up, but with the unfortunately placed(?) 1N4007, it'll never turn on.

The freq pot should have a certain minimum resistance in series with it (maybe 1/10th to 1/100th its total resistance), and a maximum in parallel -- this ensures that the 555 never tries to eat itself making too high of a frequency, nor gets stuck in a fixed state if the pot goes open (0Hz isn't much AC).

Not as dangerous in this case, since the 555 is limited by the 10k to pin

Likewise, the PWM control shouldn't go all the way to 2/3*Vcc, but a bit shy of that. So the output doesn't accidentally get stuck on 100% duty. Changing the resistor below the pot to 15k would be about right.

You can also replace the 555 with the other half of the '393, e.g.:

As mentioned, the output stuff is missing a pull up, so at worst, the LM393 sets the output high all the time (PWM control at negative side), and the output just kind of casually drifts up due to leakage (primarily of the top 1N4007). As the MOSFET goes into conduction, it starts to cook, magic smoke releases, and your supply ends up shorted (if it's beefy enough to clear the short, the magic smoke will transition into magic fireworks before everything settles down).

Most of the time, it'll just be switching terribly, because (assuming the

IRF740 is also a poor choice for 12V 5A. If the flyback voltage is low, an IRFZ34N would be passable, but more likely, something like "20N25" would be better (i.e., ~20A, 250V, N channel). You'll still want a heatsink, but it won't be as bad as the IRF740 (which just isn't worth using on 12V).

The TVS diodes are also rather restrictive, and not used correctly. The most likely result (once it's switching, that is) is a cooked schottky diode (it's only rated 40V, and will avalanche at 50-60V for a few pulses before failing shorted).

Here's a more complete example, using a BJT output:

This isn't variable frequency, because the flyback is pretty standard (it's either intended for 15.7kHz from a TV, or 30-100kHz from a monitor). Ignition coils are also fairly standard, given there are a few types to choose from (old fashioned points, modern transistor driven, CDI). You could pick a couple capacitor options for each class of component and leave it at that (frequency isn't as important as PWM, anyway).

Tim

On a sunny day (Sat, 13 Dec 2014 06:05:25 -0600) it happened "Tim Williams" wrote in :

Nice way to remove base charge :-)

I have designed this variable frequency, variable pulse width HV power supply as a bench test device for experimental use.

Here is the schematic:

And here is the Vero board layout:

Can viewers please comment on the above with particular reference to any obvious mistakes and the quality of back EMF protection I have included.

Thank you,

Bob Griffin

Reference designators would help the discussion.

As noted, one diode is backwards. You don't need either diode in the gate circuit, nor the transzorb thing, nor the 100r resistor. The gate drive is maybe a tad weak overall.

You also don't need the schottky.

Auto ignition coils usually have modest turns ratios, and depend on having a lot of primary flyback voltage to get the secondary HV. So the clamp circuit across the primary may limit the secondary voltage.

Inductive flyback type coils generally make fatter sparks with a capacitor across their primary.

CD type coils are different. They need high-voltage primary drive.

(I used to charge an oil capacitor to 600 volts and dump it into an old-style ignition coil through a thyratron tube. Then it was hard to insulate the secondary... the spark would jump out of the coil and over the insulator, back down to the primary terminals. The usual spark plug leads were useless. I had to add an oil-filled extender tube to get the HV out to the world.)

On Sat, 13 Dec 2014 21:51:38 -0800, Bob Griffin Gave us:

Yes. Set your PC's clock correctly. DO IT NOW!

you mean "do it yesterday" :)

My apologies. They have now been added.

Any further advice would be most welcome.

Bob Griffin

Problem: the LM393 is open-collector so doesn't pull up, so when it does go high, Qx and Qy will both turn on, and fry things.

You could do this, simpler:

Or muck the 555 circuit to do variable duty cycle, and drive the mosfet gate directly from the 555 output. Even simpler.

There's probably some ghastly way to use a dual 555, one as the oscillator and the other as a one-shot. I'll let the 555 experts work that out.

Incidentally, "RV" is the reference designator for "Riverboat." Resistors should all be "R."

Thank you for your expert advice. I have revised the schematic accordingly, or as I best understand.

The MOSFET s rated at 300V, 300W. The TVS is 150V breakdown. All as you suggested.

I have not changed the vero board layout yet.

Any further comments would be very much appreciated.

Bob Griffin

Thanks John. I have added that resistor as shown here in circuit version 3. I hope I am getting closer.

I understood the complementary pair would be a better driver for the MOSFET, so I don't mind a little added complexity.

I have seen RV, VR and R variously used to designate pots. Easily confused with recreational vehicles, virtual reality, or R rated movies.

Bob Griffin

Current source into a variable ratio current mirror into a capacitor. Use the comparators in the 555 to short the cap when the voltage hits 2/3 Vcc. There's your variable frequency sawtooth. Throw a comparator on the voltage to get your variable duty cycle.

Disclaimer: I haven't looked at any of the posted schematics.

Complexity = more things to go wrong.

R11 doesn't fix the problem. R6 is unnecessary and makes things slightly worse.

Ultimately, you can't just fiddle parts around and design reasonable circuits. You have to understand the parts and be able to do the math, which means starting from fundamentals. There is no value of R11 that works; that falls out of the math and the part specs.

And I've seen POT3

Easily

Amateurs, especially audio types, make up crazy reference designators.

Varistor, I think.

Cheers

Phil "Roll, black water" Hobbs

That was a different circuit. It needs the pullup resistor.

I explained it! The comparator has an open-collecor output, so when its output is high it basically disappears from the circuit. In that condition, the R7/R8 junction (by symmetry) will be about 6 volts, and both transistors will have about 5 mA base current, so will turn on and fight one another and probably fry.

My simple complementary emitter follower circuit doesn't have that problem; only one transistor can be on at a time. Less parts, too.

As far as being discouraged, if you want to design circuits, you have to start with very simple ones and do the math from the beginning. Start with a battery and some resistors, understand the currents and voltages. Then maybe add an LED and, again, be able to calculate the currents and voltages. And so on.

Electronics isn't like carpentry or mixing cement or things like that. Most people can build a bookshelf or a patio. In electronics, both the arrangement of the parts and the numbers have to be right for a circuit to work. It takes most people years to get this right, usually with some formal EE or physics education.

You can be a hobbyist and build circuits from books or web sites... but many of the web site circuits are crap. And if you try to make changes, without understanding the theory, you're likely to break things.

Some day I may do a video series, Electronics From Scratch.

Thanks for your advice. I removed R6 and R11. I must have misunderstood the schematic you sent that showed the latter.

I am trying to do a competent job on this circuit and learn by doing, so I appreciate your help.

The circuit, as is, is prety much what is available on the net. You seem to suggest it has a fundamental flaw.

Can you tell me exactly what is wrong and how to correct it? I am starting to get a bit discouraged.

Bob Griffin

Most circuits are flawed at first. Experienced circuit designers just find and fix the flaws faster. We have design reviews, where other people check our circuits and help us find errors, or better ways to do things.

Fire it up! If you have an oscilloscope, check the waveforms, especially the mosfet gate, before you apply the +12 to the coil. Start with low frequency, narrow positive pulses on the mosfet gate, 1 ms or less.

The triangle wave out of the 555, into the comparator, might make it difficult to make very low duty cycles, like for instance 1 Hz and 1 ms ON times. An oscillator followed by a one-shot would decouple the width from the rate.

Post pics of giant sparks!

In this latest version, I made all the corrections you kindly suggested, as I understand them. I also used more robust transistors for the complementary pair.

In my original flawed schematic below, I had attempted to provide back EMF protection for the IC's. It was recommended to do this with coil dampening and reverse voltage protection.

Apart from the TVS, these are all gone in the current version. Do you think something like this needs to be added again before testing the circuit?

Thanks for your sage advice on learning electronics. I am sure your videos would make it much easier.

Bob Griffin

Regarding

Replace the avalanche diode (D1) with a capacitor so you're driving the coil correctly. A small amount of parallel capacitance on the primary delays the rise of voltage just enough that it transfers efficiently to the secondary and output load. It wouldn't surprise me if you were losing 25 to 50 percent of your coil's capacity by absorbing the primary's flyback power rather than letting it transfer to the output load.

Here's what it looks like on a 'scope:

There's one other possible problem with your circuit. Gate-Drain capacitive coupling on the MOSFET results in a spike of negative gate current when the flyback voltage falls from ~300V to 0V. I think you'll find that Q1 is momentarily on because it's trying to hold the gate _up_ to 0 volts. In some cases this can cause oscillating in the tens of MHz or Q1 might get hot. A Schottky diode between the Gate and Source will prevent the gate voltage from going negative enough to activate Q1.

I think an auto ignition coil is like a Tesla coil, in that the natural SRF of the secondary is lower than that of the primary, so the cap helps match those frequencies and improve power transfer from primary to secondary. Maybe.