LED driver

A normally closed crow bar shunt will solve your problems.

If you loose power to the unit the crow bar shorts the output and there is no need to worry about the LED not reporting HV.

During normal operating that includes having power, you simply employ a sensing circuit from a V divider off the HV lead. Even with that, that type of equipment usually still employs a crow bar.

So no matter how you look at it, I think you'll need a crow bar bleeder.

I suppose you can have it by pass the resistor for a more aggressive shunt in case power drops out, it'll shorten the life of the contact but it'll drain it fast. oil filled contacts are good for this.

As far as I know, HV test gear requires a mechanical crow bar when power is off or lost.. You may want to look into that.

All of our units we use have them and we have many at work, all have crow bars. Normally closed!

We have a couple that will test the possibility of a non working crow bar at power up, small leakage voltage and if that is the case, it forces a main breaker trip, via a breaker option package.

Jamie

That's probably what I'd do IRL. A slow passive bleeder, plus a fast active one that engages on primary power loss; the latter being the primary, and the former the redundant safety.

The idea being to eliminate the danger, rather than simply indicate it.

But that's not nearly as fun as a 48V-5kV to LED switcher!

Cheers, James Arthur

As soon as you _truly_ address safety and use crowbars to disable the HV when mains are off, the "blinker" part becomes trivial.

Since the mains are on during test, use an auxiliary supply of 5V (which probably already exists), and then use this circuit...

VoltageSensitiveBlinker.png

on the S.E.D/Schematics page of my website.

Constant blink rate for just below 48V on up to 5KV ...Jim Thompson

Maybe not a pure circuit design problem, if interlocking of mechanisms is added in.

How about a circuit that runs LEDs off a battery, that is interconnected with the HV power source so that it delivers a power-OK battery check result as part of the power-up sequence? Realtime clock in the HV power source that tracks ON time and insists on replacement schedule?

Or, how about a mechanical shunt that shorts the HV when/if the HV warning signal goes dark? That would allow a static voltmeter, which doesn't sense low-ish voltages, to be the primary indicator, with a 'powering up' secondary indicator circuit so you can open the shunt when powering up.

"Jim Thompson" wrote in message news: snipped-for-privacy@4ax.com...

"555C" = CMOS? If not, I don't recall that the RST pin has low enough leakage to change state as shown. Of course CMOS would be fine, it's not like it needs anything fancy.

Tim

Yep, "555C" = CMOS. I have both CMOS and bipolar 555 Spice models.

The reset is "soft"... I thought about adding some snap, but Larkin wouldn't like it anyway ;-) ...Jim Thompson

My unit will be about the size of a cigar box. A Kilovac HV relay, in addition to being expensive, wouldn't fit inside.

Sure, it's trivial if you have a 5 volt supply. In fact, it degenerates to an LED and a resistor across the 24V input. But I want to indicate charge on the HV caps whether 24V (or 5V) is present or not.

What I need is a breakdown circuit, something that draws almost no current up to, say, 30 volts, then shorts out until the current drops to 100 uA or something. Like an avalanche transistor, but lower voltage. And *simple*.

A sidac almost works, but their trigger currents tend to be high.

Some transistors will do this, used upside-down, but I don't think I could trust them.

How does that 100uA meet your

How do you light an LED on very low currents?? Adequate enough to be "attention-getting"? ...Jim Thompson

An avalanche SCR, i.e. a diac, is the obvious candidate. They're available at lower currents than the protection-device (SIDAC) items. Historically, the early Shockley diodes were of this general type.

But for real HV, the best solution involves vacuum tubes. You won't like it.

For HV sensing without drawing inconvenient current, though, use a phototube as a modulator! Hamamatsu has 'em (R12290) up to 3kV (it'd be easy to go higher, and lots less expensive if you were willing to home-build).

Put a little modulated LED light into it, amplify and synchronous-rectify the output current, and you can easily sense the applied bias, while only drawing a few picoamps (nanoamps?) from the HV source. Just dim the LED until 48V is the sense threshold.

It uses a vacuum tube, needs a light-tight box, gets flaky around the work function voltage, and is kinda fiddly, but the alternatives included moving parts and trusting 25 Mohm metering resistors.

Can you cut the stored energy? I.e., use a cap multiplier to remove ripple, rather than brute capacitance?

That would help.

Yeah, we need a 5kV solid state switch. Even a crowbar needs it unless it's mechanical. (And if it's mechanical, ick.)

Cheers, James

So use a super-cap charged from your 5V... that will run the blinkety-blink until the charge bleeds off of the HV.

...Jim Thompson

That's tiny

Some of our stuff I can walk inside..

Jamie

How about if the HV bleeder chain had a bunch of LEDs shinning on photodiodes.. (This may be Phil H's upstream idea.)

That could power the LED blink circuit. Maybe too expensive?

George H.

It's company policy to not build anything that would break your foot if you dropped it.

"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...

So the crane-lifted transformers in those... gradient amps I think, or whatever they were, must've been subcontracted?

Dur...

Tim

It's a new policy.

addition

Hunh? I have seen plenty of small Kilovac relays that would fit within a deck of cards package. Maybe those are really small cigars? Plus they have competitors in that size and rating range.

?-)

Then modify my stupid circuit and have the 5 Meg resistor drive the base of a high beta (>100) transistor with LED in collector. High efficiency / brilliance white LEDs have visible light from sub-microamp region (not attention getting), to mega-amp region (one shot *PAFF* very attention getting).