LED driver

Umm, NO. The unit total power is 80 W, not the 5 kV supply. The unit includes much processing power and perhaps fast ADCs, CPU and maybe a FPGA; all with supporting glue components.

@JL have i misunderstood?

to

can hang.

better

Potted modules that output 5 kV isolated from something like 12 V, 24 V, or 48 V have been around since the 1970s. Not necessarily cheap, but available.

?-)

Hi,

What about a module with an input range covering 50VDC to 5kVDC, and fixed 20VDC output? I think if I modify my 40V-600V input flyback by adding a mosfet stack and more primary inductance (and voltage clearance) it may be able to handle it.

cheers, Jamie

Hi,

I originally was thinking the same as Phil Hobbs to use a solar cell, but now I see it makes more sense to use a peltier device heated by a less than 1watt resistor on the HV. Assuming 5% peltier conversion efficiency, that is 50mW of isolated electrical power. Of course as the HV drops the resistor dissipation will go down, and not heat the peltier but once the circuit has started up it can have a reed relay that closes once the voltage goes low enough to bypass the power resistor, and power the LED through a small step down supply etc.

When the circuit starts back up again, the reed relay will be closed, and it has to be able to open again before the HV damages the circuitry.

cheers, Jamie

Hi,

I originally was thinking the same as Phil Hobbs to use a solar cell, but now I see it makes more sense to use a peltier device heated by a less than 1watt resistor on the HV. Assuming 5% peltier conversion efficiency, that is 50mW of isolated electrical power. Of course as the HV drops the resistor dissipation will go down, and not heat the peltier but once the circuit has started up it can have a reed relay that closes once the voltage goes low enough to bypass the power resistor, and power the LED through a small step down supply etc.

When the circuit starts back up again, the reed relay will be closed, and it has to be able to open again before the HV damages the circuitry.

This all assumes you can find a reed relay with 5kV standoff voltage!

I agree it is a messed up problem lol.

cheers, Jamie

Here are some kV rated (up to 70kV) latching relays:

cheers, Jamie

No, Steve's right, I just chose a typical load rather than max.

Shunting the pass transistor with a zener might be enough, or it might not even need that--it's not necessarily hurt by a few mS of 15mA @ Vce(br), or even Vcb(br)--it depends on the source and the load, details we don't have.

John can add those features as needed, or a current-limiter. It leads to other nits and trade-offs we can't make, e.g., a shunted zener worsens line rejection, etc.

Cheers, James Arthur

The circuit I posted uses two transistors and *cannot* latch 'on', because all the positive feedback is a.c. coupled.

It does not latch 'off' because worst-case it idles linear, with positive a.c. feedback gain >1.

Cheers, James Arthur

No, but the control stuff might use a couple of watts. Most of the power will be the HV... ultimately driving a capacitive load that, in pure theory, needs zero power.

We will make or, if we can find the right one, buy the HV supply. It will have the usual current limits and such. We'll build the device driver downstream. I want an LED to indicate when HV is present, if only for me. I like lots of LEDs glowing on my circuit boards.

Have them install a blinking LED to show when the HV is present. Must work down to 48V.

Why not?

Nah, this is where it gets to be interesting. This would be the perfect Zeitgeist project.

Both the UJT as well as this supervisor ciruitry work similar to an NE-2. Except you get to choose whether you want to draw enough current at higher HV rail voltages to have the LED constantly lit or keep it flashing.

Well, I don't. Draw it up and we'll kick it around.

A PIC 10F200, for example, has a sleep current of 100nA and can be woken by its watchdog timer.

Use a 3V 'zener' (a low current bandgap device) from the HV+ with 100nA @ 45V to ground - 450M. So at 5kV that's about 11uA or 55mW.

Connect a reservoir cap across the zener and power the PIC from this high-side 3V shunt regulated supply. Connect an LED with resistor from a PIC output to HV+.

Program the PIC such that every N watchdog timeouts it flashes the LED using some of the stored cap energy, otherwise go back to sleep.

OK, you'd need more current (less resistance) so that you can charge the cap in a reasonable time and flash at a reasonable rate.

Maybe a 48V zener below the resistor to stop it at < 48V, but leakage makes that very tricky. Of course a discrete CMOS solution could work too - oscillator, counter and monostable.

Cheers

Ok, will do, but it'll be a while. Right now I have to debug a switch mode converter for a client, build up a pressure tester and design an engine control module. Then there is that honey-do list, and a friend wants to take me kayaking on Sunday and I don't have any of the gear yet ...

Excuses, excuses!

Going to buy or rent a kayak?

Have you run Cache Creek, north of Sacramento? It's probably bad this year, but it's a hoot when the water is running.

Ok, ok ... just had a Chili Verde burrito and a Negra Modelo, mas grande, at our Mexican place. So here goes:

HV RAIL | \ / 3M \ / | o---------------------------------------o--------o-------' | | | | | \ | | | / | | | \ RLED | | \ / | | / 18M | | | \ | | / 4.7M | | | | V -> | | o--------------/\/\-----------| - | | | | | LED | | | |-------------------| | | | | | | | | . | | | | | | | | | | | | | ||-+ | | o------o IN OUT | | || Going to buy or rent a kayak?

My friend has two kayaks. Maybe also a helmet (he's looking) but no extra life vest. We'll be on the American River, launching near Lotus. I still want to convince him to go mountain biking instead but he can't stand the heat too well.

I haven't kayaked in ages so I've got nothing in terms of gear and also don't remember how to do a proper hip snap roll which supposedly is essential on the American River.

Not yet. I want to see if I like kayaking and how much hassle it all is. Mountain biking is easy, you just strap on your hydration pack, sweat band, helmet, sunglasses, and step on it.

I'd be a little concerned about fig 5 of the data sheet. The supply current spikes around the schmitt transition, and it could eat its own supply voltage. I've used emitter followers to drive uP reset chips (cute way to check multiple supplies) but this one needs to get its peak current from somewhere.

Try this some time:

Family business. They truck you way up the creek, put you in a 2-man rubber raft, give you a 45-second lecture on whitewater rafting, and push you in. The weekend packages include camping and bbq's and stuff.

Make a very low current blocking oscillator to double the 48vots. Single transistor with a small coil. You don't need a secondary winding. select a 100v transistor and get the NE-2 voltage from the collector side. use a 48-50v zener to bypass it so the oscillator will also see this voltage at all times. With that , you can maybe make it down to below 48 volts as an indicator. Jamie

That's one reason why an emitter follower right before the chip can be beneficial. But I've used them in micropower situations and they behaved. Of course, by then you are almost up to the parts count of your circuit in the other thread, except that the NCP302 affords you very precise indicator thresholds.

Very tempting and the prices are fair as well. But it'll be a long drive from here and my wife wouldn't join. She is not a whitewater fan at all.

Like John Wayne would do it? "Now listen up, guys. Don't screw up in any of them rapids and don't lose them thar paddles or you'll be killed dead. Here ya go!".

That's almost how my downhill ski refresher course went. Before it was over the guy said to me and my colleague "You's two, you can go up now!" and pointed us to one of the lifts. I didn't know what a black diamond meant. Afterwards I did and my butt hurt.