I think I found it in AoE. 1500 volts in a DPAK is impressive, or maybe crazy.
Right, small parts are generally faster. I sometimes parallel several small mosfets or phemts, which has the extra benefit of reducing lead inductances.
Here's an isolated (transformer-coupled) 100 volt pulse into 50 ohms, using four SOT23 fets in parallel, each with its own gate driver. Fets are often much faster than their data sheets suggest, if you can drive them hard enough.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Yep, fun stuff!
--
Thanks,
- Win
The 2sk4177s look like they will work well in the flyback converter and the various regulators, but they will have a fair amount of switching losses in the output pulser if the customer pushes up to around 1200 volts and 10 KHz. So your ndfpd1n150c (TO220), or maybe an IXYS IXTY02N120P (DPAK) would be better there; much less output capacitance. I'll get some and try them on the breadboard.
(That's my $24 Amazon 2KV scope probe on the right.)
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
Your storage-capacitor chain doesn't need larger capacitors?
--
Thanks,
- Win
The C-W string? The voltage drop in the 33n caps is pretty small. My load current will be in the 5-8 mA range, I hope.
The HV amps, on other sheets, have 220 nF film caps, as reservoirs. I'm using some Kemet X1-type film caps, rated at 310 volts AC but 100% tested at 1500 DC. They fail at about 3KV. Some bigger units, rated
760 volts AC, seem happy at 6KV, but they are, well, big. Huge, actually.-- John Larkin Highland Technology, Inc lunatic fringe electronics
OK, I have the IR2213 driving a pair of stacked IXTY02N120P fets, running at 1200 volts and 10 KHz. Rise/fall times are around 25 ns, some insane slew rate. The unloaded supply current, IC+fets, is a mere
490 uA, which equates to about 40 pF total equivalent switching capacitance. Just connecting the scope probe makes the supply current go up about 60 uA.And nothing has blown up yet!
The ndfpd part would have even less c, but it's not surface mount.
Yup, fun stuff.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
d.JPG
Figure being able to drive 6ft of coax (180pF) and reasonable electrode loads (20pF), now you're at 3.5 watts and 3 mA on the HV supply. It'd be nice to run at 50kHz, pushing you to 18 watts and 15mA. Now a pair of TO-220 parts doesn't look so bad, with a serious heatsink. BTW, John, don't forget a 50-ohm low-inductance output resistor to forward match the coax.
--
Thanks,
- Win
The PLZT device looks like just under 50 pF, and the connections will be just wires, probably bundled and tie-wrapped or something awful like that. I'm suggesting that the pulsed wire be inside a plastic tube, to reduce c and add some HV insulation. These foreigners tend to not listen to me. With my 40 pF equivalent internal c and an external
100 pF, 1200 volt pulses at 10 KHz should need 1.7 mA of supply current per channel, and it looks like I can tweak my supply up to at least 6.That is NOT nice!
No coax; I'm thinking a couple of K ohms in each leg of the output connections, to soften things up a little. Fortunately, we never need to run above 10 KHz.
The customer was going to use Apex high-voltage opamps. Their quiescent power dissipation would have been over 20 watts per channel, times 6 channels, which would have needed a rack and fans and expensive stuff, and only swing 800 volts.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Is there a rule of thumb regarding the capacitance required in each stage for a given output voltage and current?
Or an equation. Or something. I DON'T KNOW WHAT I'M DOING.
Each cap has to furnish the load current for the period of the flyback oscillator.
C*V = I*T (charge=charge)
where I is the load current, T is the period (1/F, where F is the switching frequency) and V is how much voltage you are willing to lose across this cap. Farads, volts, amps, seconds.
In my case, F is fixed at 200 KHz by the LTC chip. So at 5 mA load, the voltage drop V=I*T/C across one 33 nF cap calculates to 0.75 volts peak, no big deal for a 1400 volt supply. 33 nF is the biggest affordable 500 volt cap that I can get in 0805.
You can also just Spice it and measure the voltage drops. Spice shows just about 1 volt p-p across each of my 33n pump-side caps at 4 mA load, close enough.
I do that a lot: guess, Spice, and fiddle, rather than doing a bunch of arithmetic.
-- John Larkin Highland Technology, Inc lunatic fringe electronics
Well, I figured you'd like to add general-purpose usage to your design, and that calls for coax.
In my early days of designing instruments that other people use, I quickly learned they often change their minds after they start using it, and realize they'd really like to "turn it up", or some such. For five decades of instrument design I've tried to outperform the stated specs by at least a factor of two if I reasonably could. This helped my designs to have a longer life and fill more applications than originally envisioned.
BTW, 2-watt inductance-free 50-ohm resistors: Ohmite WNC51RFET, $1 qty 100 at Digi-Key.
--
Thanks,
- Win
Er... but there's N stages, so N capacitors in series, so the drop is
0.75V * N... and the rectification is half-wave, so isn't it T/2 that you want?There are full-wave C-W multiplier designs, and with triangle-wave drive, they have droop times much below T/2.
I did say that the calculated drop is per capacitor. 0.75 times 4 is still trivial.
and the rectification is half-wave, so isn't it T/2 that
The delta-V across a cap depends on the average current and the rep rate. The conduction duty cycle washes out.
I don't need fancy features here, and the flyback converter does what it does. The actual (well, Spiced) capacitor current is fairly short pulses, at the rising edge of the flyback output.
-- John Larkin Highland Technology, Inc picosecond timing precision measurement jlarkin att highlandtechnology dott com http://www.highlandtechnology.com
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.