Those circuits are brute force, about 35 parts to light up one LED. And high-voltage mosfets come in big packages. I was hoping there might be something more reasonable.
So we'll always have to keep 4 feet away from our own product? That will make development interesting.
Run the current-consumption numbers... TANSTAAFL. ...Jim Thompson
Great, nobody dies as long as the crowbar circuit is absolutely reliable. Just yank the plug and grab stuff.
How go you crowbar 5KV anyhow?
Yeah, a lot of parts for a simple sounding task. But the fets are sot23.
piglet
And one weak blink every 20 seconds. Doesn't sound very safe to me.
Yes. Some fundamental math demonstrates that this is hard to do. Playing with circuits can't overcome fundamentals.
Looks like I need a 20-year lithium battery or a supercap somewhere, to light up the LED during the HV discharge, the time between when power fails and when the caps are bled down below 48 volts.
It's barely theoretically possible without those, but the LED drive efficiency has to be high. That means inductors, or a 24 volt multi-chip LED. But it will still be tough, given about 1 uA to work with.
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
This is what we did. A nice rotating red light beacon thing near, on, or inside the equipment under test. The bigger ones had them installed internally (controlled by a "non-defeatable" door switch, if at all).
The technicians actually doing the work must always follow procedure, but the light provides a handy indication for passers-by to steer clear of. In addition to the yellow-and-black safety chain placed around the work area, indicating the general danger.
Presumably, with a big hunk of metal, spring loaded, so when someone pulls the door too soon, KABANG!!!1, there goes the cap. And now you know it was energized, better pay attention. Also, preferably in a visible location, so the technician can also see that, hmm yep, that thing's pretty damn well shorted across, poke it with the multimeter, hmm yep 0V, good to go.
Seriously, if you have to ask the internet -- not even the internet, the Usenet about this -- you have much bigger problems. Consult your friendly local OSHA offices about electrical safety today.
If you're still making those gradient coil amps, it probably wouldn't hurt to employ some of these procedures around those as well. Lower voltage as I recall, so not as big a deal as industrial 480, or reach-out-and-touch-you high voltage, but prudent all the same.
Tim
"John Larkin" wrote in message news: snipped-for-privacy@4ax.com...
If you're absolutely, positively set on this, you're probably better off with something like
Tim
It looks borderline impossible at even 1% duty cycle.
If I have a power source that is not the HV itself, it's trivial. But if the HV is charged and that aux power source is unavailable, we have a hazard.
Do you know of a high voltage SOT23 mosfet? 600 seems to be the max, so I'd need maybe 10 of those, plus resistors. There are NPN 1200 volt SOT223's, about a wash. Still a lot of junk.
Why? You don't need to go as low as 1%. Assume a 5mA LED which nowadays is really bright. Continously lit that's 25W in the bleeder, too much. At 10% it's only 2.5W. A 5% duty cycle would probably still be feasible and that gets you to 1.25W. Ok, maybe 1.3W because the blinky-blink circuit wants its tax.
There is no aux supply in this case, the blinky-blink circuit and LED is purely fed from the HV. An aux supply would likely not pass muster in front of the agency folks.
You don't believe in capacitors? :-)
All you have to provide is the average current which drops (almost) proportionately with the duty cycle.
BTW, I had something similar to a free lunch on a recent mountain bike ride. Found a major swath of blackberry brambles, nobody seems to ever get there and pick any. So I had a helping of blackberries, courtesy of mother nature. For some strange reason down here at 1400ft elevation they are all still green while up at 2500ft only 80% were green, 10% red and another 10% black and thus ripe. Must be the dry weather that delayed them.
San Francisco is full of blackberries [1], and very few people seem to want to pick them. They dry up, or feed the birds mostly. I have a few favorite patches near my house, like near Douglas Playground, or the north edge of Mt Davidson.
They should be ready in another month maybe. A good blackberry-peach pie is hard to beat.
[1] reviled by the purists as an invasive non-native species. These are the people who girdle eucalyptus trees and rip out pretty stuff to plant poison oak, which *is* a native species.
I can afford maybe 1 watt in the bleeder, which makes the main resistor 25M at
5KV. At 48 volts, that same 25M gives me max available power of 24 uW at 1 uA. Not much to work with.
Some "safety-minded soul" should E-mail a copy of this thread to the SF OSHA office >:-} ...Jim Thompson
Running the LEDs from 5KV is easy.
Are those bells going to work down to 48 volts?
One thing my customer suggests is so many screws on the cover that they take long enough to remove, that the caps are bled down by the time anyone can get the cover off. Assuming the power is removed first. But I want some additional backups, like the LEDs, for my folks, who will manufacture and test these things.
My bigger gradient amps have LED voltage indicators and a pushbutton discharge switch, in addition to the slow bleeders. But neither the indicator nor the discharge button want to work at 5 KV.
And you probably will. You always were a sneaky, slimy SOB.
[...]
Yeah, true. So we need a high-voltage PTC resistor. A buck would be kind of hard to do and get expensive.
My wife turns them into a kind of sloe gin. Delicious.
Oh man, I am super sensitive to poison oak. When bicycling I sometimes have to step on it even if the trail is a white-knuckle ride, just to get up to sufficient speed, raise my legs and then coast through poison oak. Woe if I ever fail to have enough speed and the bike stops in the middle of that.
Have something to add? Share your thoughts — no account required.
Ask the community — no account required