LED driver

Worst case, the bleed resistors bring the 5KV down in 16 seconds.

However, there is a better solution. Make the unit accessible only from the bottom. You have to turn it over to get inside.

Then have a gravity switch short the 5KV to ground. Here is an illustration. The bottom cup can be a parabola like a teacup, or flat bottomed with sloping sides. I had to use a flat bottom due the the difficulty drawing an arc in LTspice.

The lid is a flat cone so the aluminum shorting ball can't sit in the middle. The edges are rolled to prevent flashover. You can get the aluminum balls here:

0.50" (12.7mm) Aluminum Balls Price is for one bag of 50 balls. AL_500 $12.00

Version 4 SHEET 1 880 772 TEXT 104 -224 Left 2 ;'Larkin's 5KV Gravity Switch TEXT 144 -144 Left 2 ;Normal Operation TEXT 152 112 Left 2 ;Tipped on Side TEXT 160 616 Left 2 ;Upside Down TEXT 448 -80 Left 2 ;7/16" clearance TEXT 32 -48 Left 2 ;1/2" Aluminum Ball TEXT -56 -120 Left 2 ;5KV TEXT -56 -72 Left 2 ;GND TEXT 104 -200 Left 2 ;- SW 2014 TEXT 440 128 Left 2 ;Notes: TEXT 440 152 Left 2 ;1. Roll edges to prevent flashover TEXT 440 176 Left 2 ;2. For illustrative purposes only TEXT -96 496 Left 2 ;Ball contacts 5KV and ground TEXT 440 688 Left 2 ;Ball contacts 5KV and ground LINE Normal 208 -80 0 -96 LINE Normal 416 -96 208 -80 LINE Normal 32 32 0 -64 LINE Normal 208 32 32 32 LINE Normal 384 32 416 -64 LINE Normal 384 32 208 32 LINE Normal 176 352 160 560 LINE Normal 160 144 176 352 LINE Normal 288 528 192 560 LINE Normal 288 352 288 528 LINE Normal 288 176 192 144 LINE Normal 288 176 288 352 LINE Normal 208 752 416 768 LINE Normal 0 768 208 752 LINE Normal 384 640 416 736 LINE Normal 208 640 384 640 LINE Normal 32 640 0 736 LINE Normal 32 640 208 640 LINE Normal 448 -64 443 -48 LINE Normal 448 -64 453 -48 LINE Normal 448 -64 448 -32 LINE Normal 448 -96 453 -112 LINE Normal 448 -96 443 -112 LINE Normal 448 -96 448 -128 LINE Normal 456 -64 424 -64 LINE Normal 456 -96 424 -96 LINE Normal 180 -22 169 -33 LINE Normal -6 -98 -22 -114 LINE Normal -3 -61 -15 -67 LINE Normal 155 523 139 507 LINE Normal 419 715 435 699 CIRCLE Normal 240 -48 240 -48 CIRCLE Normal 176 31 240 -33 CIRCLE Normal 208 320 208 320 CIRCLE Normal 227 559 163 495 CIRCLE Normal 176 720 176 720 CIRCLE Normal 415 701 351 765

Thou shalt not goof. But if concerned there'd still be the option of limiting the reservoir capacitor re-charge to a current value that is guaranteed lower than the valley current. The downside is that then the flash frequency would no longer increase with higher voltages on the HV rail.

I would just use the UJT because it's simple, is only one part, readily available and cheap.

[...]

--
Regards, Joerg 

http://www.analogconsultants.com/

If the supply current exceeds the valley current, it latches. The impedance at the G terminal of the PUT determines both the peak (firing) and the valley current, so whatever drives that has to be a compromise. It's a matter of making the numbers work.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

You'd have to either make the re-charge current into the cap smaller than the valley current (current source) or use a SOT23-5 comparator to short the UJT when the cap is below a certain voltage (depleted by LED action). An alternative may be a transistor or FET. Gate tied to the cap via a large value divider. When the cap voltage gets below a limit this transistor or FET opens -> UJT current drops to zero which is guaranteed below valley current. The LED would have to still be drawing enough current to continue current flow at the point when that happens.

Best would probably be to simulate this or just build it and measure. Normally the UJT should behave "digitally", staying in full conduction mode and thus keeping teh LED engaged until this transistor opens and lets it fall below Iv. That should also make sure that the turn-off of the LED is as snappy as the turn-on so it becomes more alerting and visible.

--
Regards, Joerg 

http://www.analogconsultants.com/

Thanks for the comments. The cap multiplier might be a good idea. A 12nF cap would increase the ripple to 50V. The problem now is to bias the emitter follower so it remains linear throughout the flyback cycle.

I suppose a high voltage zener might help. But it will waste a lot of power. Maybe use it as part of the bleed resistor. There should be HV zeners with knees below 200uA.

OTOH, the cap multiplier could help to reduce the ripple in the original configuration without having to increase the value of the HV cap. Now the zener only has to be 10V or so. And you have to provide protection for the emitter-follower when the tech shorts 5KV to ground. That might get interesting at 5KV.

A simpler solution might be to only allow access through the bottom, so you have to turn the unit over. Then an inexpensive gravity switch can short the 5KV to ground through a limit resistor.

Idea #2:

Use a voltage supervisor chip. Those usually have built-in hysteresis but it would be good to feed forward for some more hysteresis. Should be one that live on a microamp of so such as this series:

Use one with a totem pole output, drive a FET with that and this FET drives the LED plus series resistor. Or us a Darlington and a resistor for a more constant LED current.

A resistor divider would connect this to the reservoir cap and when a desired threshold is reached the output of the supervisor chip goes high, turning on the LED. The LED needs to draw more than the current the HV-resistor can provide unless it's ok if the LED is constantly lit at higher voltages.

If the rail voltage is only slightly above 48V or wherever the warning threshold has to be there will be only a few tens of uA. Then the LED will flash less frequenctly. The cap would set the flashing frequency and flash duration under those conditions.

Of course, as usual there needs to be a low leakage angst zener somewhat above the threshold.

--
Regards, Joerg 

http://www.analogconsultants.com/

Hi,

I made a flyback that can operate from about 30VDC all the way up to

600VDC and it can output 20VDC/650mA over that whole range. If I remember right, I tested one of the prototypes down to about 13VDC input when it finally shuts down, but the restart requires about 18VDC I think.

I have the simulation in ltspice using an Ice Components transformer and a depletion mode Fet for bootstrapping off the input voltage. The depletion mode Fet is good as it only draws power for the regulator from the HV while the circuit is starting up, and once the circuit is regulation, the power comes through a third coil on the transformer.

I previously laid it out in Eagle CAD and made some prototypes and they worked.

I just did a simulation with a 20mA LED as the only load, the full circuit draws 962mW off the 400VDC rail, and 789mW off a 48VDC rail, in both cases the output is regulated to 20VDC with the 20mA LED load. No load is still 416mW at 48VDC input. The controller IC is LT1246.

The original circuit is meant for a higher output load, but if it is redesigned for only driving a diode then it would be more efficient.

cheers, Jamie

That's easy. LTSpice says this cuts 40Vp-p @ 20Khz (sine) to ---+----+---. .--' 500k 50W | \ ^ | | ----- | R1 | | === 100K | | | | R3 | +----1M-----+ | | C1 R2 | --- 5nF 25M --- | | === ===

The transistors need short-circuit protection, but that's the general idea.

If the bulk cap at the input is 22nF, we're down to 22nF + C1 = 27nF at 5kV, which bleeds down to 50V in 3s when loaded by R1+R2=25M.

Cheers, James Arthur

--- Version 4 SHEET 1 1072 680 WIRE 240 -608 96 -608 WIRE 384 -608 336 -608 WIRE 416 -608 384 -608 WIRE -112 -496 -128 -496 WIRE -16 -496 -112 -496 WIRE 96 -496 96 -608 WIRE 96 -496 -16 -496 WIRE 160 -496 96 -496 WIRE 288 -496 288 -544 WIRE 288 -496 256 -496 WIRE -128 -432 -128 -496 WIRE -16 -432 -16 -496 WIRE 416 -400 416 -608 WIRE 208 -384 208 -432 WIRE -128 -336 -128 -352 WIRE -16 -288 -16 -352 WIRE 64 -288 -16 -288 WIRE 208 -288 208 -384 WIRE 208 -288 144 -288 WIRE -16 -256 -16 -288 WIRE 208 -256 208 -288 WIRE 416 -256 416 -320 WIRE -16 -160 -16 -176 WIRE 208 -160 208 -192 FLAG 208 -160 0 FLAG -16 -160 0 FLAG -128 -336 0 FLAG -112 -496 Vin FLAG 416 -256 0 FLAG 384 -608 Vfilt FLAG 208 -384 Q1_Vb SYMBOL cap 224 -256 M0 SYMATTR InstName C1 SYMATTR Value 5nF SYMATTR SpiceLine V=50 Irms=0 Rser=0.008 Lser=0 SYMBOL voltage -128 -448 M0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 WINDOW 3 -98 146 Left 2 SYMATTR Value SINE(5000 20 20k) SYMATTR InstName V1 SYMBOL npn 160 -432 R270 SYMATTR InstName Q1 SYMATTR Value 2N2222 SYMBOL res -32 -448 R0 WINDOW 0 -38 42 Left 2 WINDOW 3 -49 81 Left 2 SYMATTR InstName R1 SYMATTR Value 100k SYMBOL res -32 -272 R0 SYMATTR InstName R2 SYMATTR Value 25meg SYMBOL res 432 -304 R180 WINDOW 0 36 76 Left 2 WINDOW 3 36 40 Left 2 SYMATTR InstName RL SYMATTR Value 500k SYMBOL npn 240 -544 R270 SYMATTR InstName Q2 SYMATTR Value 2N4401 SYMBOL res 160 -304 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 1meg TEXT 328 -136 Left 2 !.tran 0 2 0 10u uic TEXT -168 -576 Center 2 ;5KV supply filter

So you don't understand how it works? ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

You "detailed" nothing... you OPINED without backing it up with proof.

This subthread has no intellectual content, so is terminated ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Hi,

Oops, for some reason I thought your supply was 500VDC max, but it is

5kV! :) I haven't tried making a flyback can work from the range of 48VDC to 5kV but the first problem would be the depletion mode fet wouldn't withstand 5kV, but an SCR circuit could do the bootstrapping maybe. I wonder if there is already a DC-DC converter that covers a 5kVDC to 50VDC input range! :)

cheers, Jamie

The PNP/NPN pseudo-SCR/PUT/UJT does not have adequate degrees of freedom, thus my three-part snap/controlled dump/holding current approach. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Which is what the Q6 portion of my...

VoltageIndicator(SED).pdf

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

No compromises.

See my other post regarding degrees of freedom. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

I gave voltages, currents, and the overall situation. I don't know what else you want for "proof." When a circuit has a suspected defect, it's the obligation of the design engineer to prove that it's safe.

Weaseling again! Hiding under the bed!

It hangs up as I explained. Admit it.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

At low currents, say 10 uA, the latch is locked but Q6 is off. It can hang.

Async hairball.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

John, You really should apply some math. However, if you're incapable, simulate it. ...Jim Thompson

--
| James E.Thompson                                 |    mens     | 
| Analog Innovations                               |     et      | 
| Analog/Mixed-Signal ASIC's and Discrete Systems  |    manus    | 
| San Tan Valley, AZ 85142     Skype: skypeanalog  |             | 
| Voice:(480)460-2350  Fax: Available upon request |  Brass Rat  | 
| E-mail Icon at http://www.analog-innovations.com |    1962     | 
              
I love to cook with wine.     Sometimes I even put it in the food.

Thanks for posting the ASC file.

Minor nits: I believe the load is 80W, so RL needs to be 312,500 ohms.

The protection circuits might be tricky. Assuming the 5KV comes up fast, there is also a large voltage across the CB of Q1. So we not only have to consider a short circuit on the output, but the turnon transient.

Here is the revised ASC file with an input rectifier to show the actual ripple with 22nf and 312k load. Works fine.

SHEET 1 1072 680 WIRE 240 -608 96 -608 WIRE 384 -608 336 -608 WIRE 416 -608 384 -608 WIRE 416 -576 416 -608 WIRE 288 -528 288 -544 WIRE -256 -496 -272 -496 WIRE -224 -496 -256 -496 WIRE -128 -496 -160 -496 WIRE -80 -496 -128 -496 WIRE -16 -496 -80 -496 WIRE 96 -496 96 -608 WIRE 96 -496 -16 -496 WIRE 160 -496 96 -496 WIRE 288 -496 288 -528 WIRE 288 -496 256 -496 WIRE -272 -480 -272 -496 WIRE -128 -480 -128 -496 WIRE 416 -480 416 -496 WIRE -16 -464 -16 -496 WIRE 208 -416 208 -432 WIRE -128 -400 -128 -416 WIRE -272 -384 -272 -400 WIRE -16 -336 -16 -384 WIRE 16 -336 -16 -336 WIRE 64 -336 16 -336 WIRE 208 -336 208 -416 WIRE 208 -336 144 -336 WIRE -16 -320 -16 -336 WIRE 208 -320 208 -336 WIRE -16 -224 -16 -240 WIRE 208 -224 208 -256 FLAG 208 -224 0 FLAG -16 -224 0 FLAG -272 -384 0 FLAG -256 -496 Vin FLAG 416 -480 0 FLAG 384 -608 Vfilt FLAG 208 -416 Q1B FLAG 288 -528 Q2B FLAG 16 -336 R1R2 FLAG -128 -400 0 FLAG -80 -496 5KV SYMBOL cap 224 -320 M0 SYMATTR InstName C1 SYMATTR Value 5nF SYMATTR SpiceLine V=50 Irms=0 Rser=0.008 Lser=0 SYMBOL voltage -272 -496 M0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 WINDOW 3 -98 146 Left 2 SYMATTR Value SINE(0 5000 20k) SYMATTR InstName V1 SYMBOL npn 160 -432 R270 WINDOW 0 80 5 VRight 2 WINDOW 3 79 43 VRight 2 SYMATTR InstName Q1 SYMATTR Value 2N2222 SYMBOL res 0 -480 M0 WINDOW 0 -38 42 Left 2 WINDOW 3 -49 81 Left 2 SYMATTR InstName R1 SYMATTR Value 100k SYMBOL res -32 -336 R0 SYMATTR InstName R2 SYMATTR Value 25meg SYMBOL res 400 -480 M180 WINDOW 0 36 76 Left 2 WINDOW 3 36 40 Left 2 SYMATTR InstName RL SYMATTR Value 312500 SYMBOL res 160 -352 R90 WINDOW 0 0 56 VBottom 2 WINDOW 3 32 56 VTop 2 SYMATTR InstName R3 SYMATTR Value 1meg SYMBOL diode -224 -480 R270 WINDOW 0 32 32 VTop 2 WINDOW 3 0 32 VBottom 2 SYMATTR InstName D1 SYMATTR Value 1N6519 SYMBOL cap -144 -480 R0 SYMATTR InstName C2 SYMATTR Value 22nf SYMBOL npn 240 -544 R270 WINDOW 0 80 5 VRight 2 WINDOW 3 79 43 VRight 2 SYMATTR InstName Q2 SYMATTR Value 2N2222 TEXT -16 -656 Left 2 !.tran 0 40m 1ms uic TEXT 56 -680 Center 2 ;5KV supply filter

Here is the PLT file

[Transient Analysis] { Npanes: 2 { traces: 1 {524291,0,"V(q1b)"} X: ('m',0,0,0.004,0.039) Y[0]: ('K',1,800,400,5200) Y[1]: ('_',0,1e+308,0,-1e+308) Volts: ('K',0,0,5,800,400,5200) Log: 0 0 0 GridStyle: 1 }, { traces: 1 {268959746,0,"V(5kv)"} X: ('m',0,0,0.004,0.039) Y[0]: ('K',3,4952,4,4988) Y[1]: ('_',0,1e+308,0,-1e+308) Volts: ('K',0,0,3,4952,4,4988) Log: 0 0 0 GridStyle: 1 } }

I described the hang state. Simulate it yourself, since you have it set up.

This, like most such relaxation oscillators, has overall *negative* feedback, through Q4 and then Q6. Once it enters the hang state, it really wants to stay there. There's plenty of voltage available to keep Q1-Q2 locked up.

Only the delay of C2 makes it work at all, but once it hangs, that doesn't help any more.

--

John Larkin         Highland Technology, Inc 

jlarkin att highlandtechnology dott com 
http://www.highlandtechnology.com

I don't see a way to make that work, either.

This apparently simple problem turns out to be immensely perverse.

We're trying to emulate an NE-2!

--

John Larkin                  Highland Technology Inc 
www.highlandtechnology.com   jlarkin at highlandtechnology dot com    

Precision electronic instrumentation