The LNDs seem to run around 1.6 mA and are pretty consistant, but hardly precise. OK if you don't care about the absolute current.
Is it safe to series them? That might be useful in a low-current, high-voltage regulator I'm considering... 800 volts maybe. I could parallel each one with a 390 volt zener, available for a mere $32 each.
Or I could just cascode an IXTY01N100D (1KV depletion fet) with an optocoupler as the series pass element.
It would be great if I could series a couple of MOC8204 optocouplers (good for 400 volts each!) but that might not be safe.
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Hah. Yes, it's totally safe. You use resistors setting each current to 0.5mA, or some such. The lowest-current one of the series avalanches first, say at 600V, and dissipates 300 mW, which is fine. The next one takes over. And etc.
But wrong. Ring-of-two was more usually used to describe an arrangement whe re two voltage references, one referred to a positive rail and the other re turned to a more negative rail, each defined a constant current sources tha t provided the current to excite the other voltage reference.
If your needs extended to buying one 1N829 temperature stable 6.2V +/-5% re ference voltage (5.89V to 6.51V)
formatting link
you got 5ppm/C stability if the current through the diode was exactly 7.5mA .
The stability could go down to +/- 10ppm/C if you let the current through t he diode range over 7.5mA +/-1mA. Without trimming, a ring-of-two could off er +/-8ppm/C stability. With trimming you could pad a 768R 0.1% E96 15ppm m etal film resistor up to anything from 785R to 868R with a 26 turn trimming potentiometer, to get exactly 5.76V over the 768R resistor. You'd have two of them, so you'd have to set one, then the other, then correct the first one a little to reflect the effect of setting the second one.
formatting link
I know you only need 100R but 500R was all that Farnell stocked in that ser ies.
You'd use a two pairs of dual, matched transistors - NPN and PNP - to turn the voltage references into current sources, perhaps with Wilson cascodes i f you were being particularly picky
I'd sketch it in LT Spice, but I let my copy up-date itself recently, and n ow I can't edit the component data - I can bring up the zener diode symbol, but I can't put in 1N829 as the component value - even when running the pr ogram as an administrator. Removing it and re-installing it didn't cure the problem, nor did playing with permissions on the program file. Very odd.
On Monday, 13 April 2015 12:36:17 UTC+10, Bill Sloman wrote:
now I can't edit the component data - I can bring up the zener diode symbo l, but I can't put in 1N829 as the component value - even when running the program as an administrator. Removing it and re-installing it didn't cure t he problem, nor did playing with permissions on the program file. Very odd.
It turned out that I had to let LTSpice identify it as a particular real di ode, and only then could I edit it to represent the diode I wanted.
Here's the .asc file.
Version 4 SHEET 1 880 680 WIRE -96 -464 -320 -464 WIRE 336 -464 -96 -464 WIRE -96 -400 -96 -464 WIRE 336 -368 336 -464 WIRE -96 -288 -96 -320 WIRE -96 -176 -96 -208 WIRE 336 -176 336 -304 WIRE 176 -128 -32 -128 WIRE 272 -128 176 -128 WIRE 176 -48 176 -128 WIRE 336 -48 336 -80 WIRE 336 -48 176 -48 WIRE -96 48 -96 -80 WIRE 80 48 -96 48 WIRE -96 80 -96 48 WIRE 336 80 336 -48 WIRE -320 112 -320 -464 WIRE 80 128 80 48 WIRE 80 128 -32 128 WIRE 272 128 80 128 WIRE 336 208 336 176 WIRE -96 272 -96 176 WIRE 336 320 336 288 WIRE -320 448 -320 192 WIRE -144 448 -320 448 WIRE -96 448 -96 336 WIRE -96 448 -144 448 WIRE 336 448 336 400 WIRE 336 448 -96 448 WIRE -144 512 -144 448 FLAG -144 512 0 SYMBOL zener -80 336 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D1 SYMATTR Value 1N829 SYMBOL zener 352 -304 R180 WINDOW 0 24 64 Left 2 WINDOW 3 24 0 Left 2 SYMATTR InstName D2 SYMATTR Value 1N829 SYMBOL res -112 -416 R0 SYMATTR InstName R1 SYMATTR Value 768R SYMATTR SpiceLine tol=0.1% SYMBOL Misc\\xvaristor -112 -304 R0 SYMATTR InstName U1 SYMATTR Value 0R to 500R SYMBOL res 320 304 R0 SYMATTR InstName R2 SYMATTR Value 768R SYMATTR SpiceLine tol=0.1% SYMBOL Misc\\xvaristor 320 192 R0 SYMATTR InstName U2 SYMATTR Value 0R to 500R SYMBOL voltage -320 96 R0 WINDOW 123 0 0 Left 2 WINDOW 39 0 0 Left 2 SYMATTR InstName V1 SYMATTR Value 12V SYMBOL npn 272 80 R0 SYMATTR InstName Q1B SYMATTR Value BCV61C-2 SYMBOL npn -32 80 M0 SYMATTR InstName Q1A SYMATTR Value BCV61C-1 SYMBOL pnp -32 -80 R180 SYMATTR InstName Q2A SYMATTR Value BCV62C-1 SYMBOL pnp 272 -80 M180 SYMATTR InstName Q2B SYMATTR Value BCV62C-2
I need a high voltage linear regulator, actually 8 on a board, low current requirement, so I was thinking of something like this...
formatting link
The resistor above the phototransistor could set the current limit.
The 1KV Ixys depl fet is a big DPAK, but this circuit is so clean and simple, it's worth it.
Or I guess I could do this...
formatting link
less fun.
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
Hard to say. Somebody whose circuit diagrams don't include values for any of the components used isn't putting enough effort into communication to be amusing or informative.
Not only does it glow in the dark, the current should also be dependent on any illumination falling on the LED. Could you re-measure your circuit in the dark and in sunlight? I would be interested to see how many ppm (or %) it changes. This might only be a problem if you allow flickering lighting to fall on a low-noise circuit, which might cause hum.
Yes, thanks Bill. You are correct. The "ring of two" is the two mutually biased current stabilized zener reference circuit. I think I came across the name wrongly applied to the vbe current sink in some audio amplifier text.
I'd expect that effect to be really small, sub-PPM, likely unmeasurable.
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
The devil is in the details, as they say. Speed is an issue, but with the right opto-coupler, and with base-resistor bypassing, etc., it should do, for a power supply anyway. :-)
Some unintended fun, because either the depletion-mode MOSFET runs at a high quiescent current, or you can't get it to pump itself up for high-current loads. Our Figure 9.110, page 696, is similar, but we don't try to avoid using a gate pullup.
My particular application needs low output current, way below 1 mA, and low quiescent power dissipation. The big power hog will be the feedback sense resistor, which I can make 20 megs or so. The output cap can be big enough (10 nF or so maybe?) that the loop can be slow.
But the photon-coupled thing looks best. It's remarkably economical of parts.
Another photon-coupled circuit would use a PV coupler and an enhancement fet...
formatting link
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
It's a DAC-programmed, constant-voltage, current-limited regulator.
--
John Larkin Highland Technology, Inc
picosecond timing laser drivers and controllers
jlarkin att highlandtechnology dott com
http://www.highlandtechnology.com
I've seen these diodes delivered with a calibration certificate, and it was a lot better than 5% for those parts. The 7.5 mA requirement was easy - just regulate with an op amp bootstrapped to the reference diode
The 1N829 isn't a zener diode; it's a REFERENCE diode, there's a forward-biased diode in series with a zener, to provide temperature compensation. It's really an IC. The venerable uA723 regulator has the same kind of reference, which is very useful (and easier to find).
It might. I just measured the output of a noname T1-3/4 red LED in my lab (pointed at the ceiling fixture from a few feet below) and got about
100 nA (a bit over a volt, measured with a Tek DMM916 on the 4V scale, whose shunt resistance is 11.11 Mohm).
Cheers
Phil Hobbs
--
Dr Philip C D Hobbs
Principal Consultant
ElectroOptical Innovations LLC
Optics, Electro-optics, Photonics, Analog Electronics
160 North State Road #203
Briarcliff Manor NY 10510
hobbs at electrooptical dot net
http://electrooptical.net
That information isn't shown on the data sheet, and isn't available to prod uction until the diode has been delivered. In reality - with that informati on - you'd use one or two "select on test" additional resistors to customis e a particular example of the circuit to match it's particular diode, rathe r than a trimmer, which some clown (graduate student) can always re-adjust to satisfy their particular prejudices.
Everybody knows that. Trimming for exactly 7.5mA with a +/-5% reference dio de makes life a bit more complicated - and if you want exactly 10.00V out i t gets a bit worse. Been there, done that.
The ring-of-two configuration is pretty old, and was probably first realise d with valves and neon-style gas-discharge references.
nd now I can't edit the component data - I can bring up the zener diode sym bol, but I can't put in 1N829 as the component value
Not necessarily. The poor man's 1N823 is a gold-doped transistor in which y ou use the collector-base diode as the forward diode, and the reverse based emitter-base diode as the zener. Transistors aren't integrated circuits.
There's a short paper on doing exactly that in Journal of Scientific Instru ment in the 1960's. I think I've referred to it here, back when I could rem ember the transistor part number. The paper talks about trimming the curren t for minimum temperature coefficient - which happened at around a couple o f mA - by sticking a soldering iron on the transistor's metal can. My memor y says TO-5, but it's not to be relied on.
ry
There are many better voltage references around these days, some of them tr immed at manufacture to tight tolerances. My current favourite is the LTC66
55BHMS8-5.
formatting link
which I've been told has a very low 1/f noise content. I've got a couple th at will end up in my low distortion sine wave oscillator circuit if I ever actually put it together.
The ring-of-two that I posted as an LTSpice file wasn't all that practical. In reality you'd want only one 5ppm/C stable 6.2V +/-0.31V voltage source, and the second reference diode wouldn't need to be nearly as good as 1N829 , nor need an exactly 7.5mA +/-0.1mA reference current, so one of the trimm ing resistors becomes redundant.
As usual, the real circuit is less symmetrical and less cute than the the l ecture example, and takes more work to get right.
This form is appealing for its simplicity, with the resistor-programmed CL, etc., but what can it do? One limitation is limited output capability, all the current has to go through the opto-coupler. Looking in AoE-III, Figure 12.85, page 845, for candidates, there's D., the photo-Darlington. But examining the 4N32 and H11B1, neither is appealing. We're limited to 10 to 20mA, and switching speeds as slow as 20us.
E., photodiode + two transistors, a 6N139, could work but the PD needs a + supply. It could be returned to the depletion MOSFET's source. Fairchild's datasheet shows output currents to 50mA, with 5mA drive, and somewhat faster speeds. That's a good possibility.
F., an LTV-352, looks like trouble, with a threshold, and it looks miserably slow.
All this forces us back to ordinary single-transistor opto-couplers, B. in our figure. The 4N35 appears to be limited to about 10mA, but it may be fast enough, especially with a say 1k base resistor to help turn it off. The CNY17-4 boasts higher CTR, 160% minimum. With 20mA drive from the op-amp, John's very simple regulator could work to 30mA, not too bad.
Table 3.6, page 210, Depletion-mode MOSFETS, has 1kV IXYS parts to consider. A high-current part like the 3A IXTP3N100D2 would be good, because its -Vgs ~ -3V, would be high enough to operate the opto-coupler.
But is its high Ciss = 1020pF capacitance an issue? If we estimate gm at 10mA to be 20mS, we can calculate fT, which is its frequency response in cascode mode. fT = gm / 2pi Ciss = 3MHz. OK, phew, good, clearly that's not a problem. What about power dissipation? 30mA at 800V is 24W. Ouch, that's a big heat sink! We could limit this to say 12W with a foldback current limit, by adding a resistor from the FET's source to +800V. If Vgs is 3V then John's source resistor would be 100 ohms, and a roughly 47k resistor would be called for. But, oops! that'd have to be a 13W resistor. Ouch. OK, forgettaboutit, use a big heatsink, maybe with a fan.
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.