I tested again today with a new LM317 (the previous one had turned into a 3 Ohm resistor). Before testing I added 2 pcs 15 Volt zeners in series across the LM317. The voltage across the LM317 never came above 33 volt, turning the variac. However later loading a capacitor across the output killed the LM317. At least the zeners worked…
A 30 Volt pre-regulator is not an option, since I need a higher output voltage (150 - 400 Volt). The 165 Volt was just for testing.
Conclusion: this is not a valid option for me to limit the current in a “higher” voltage supply.
Regards, Gerrit
A 30 Volt pre-regulator is not an option, since I need a higher output voltage (150 - 400 Volt). The 165 Volt was just for testing.
Conclusion: this is not a valid option for me to limit the current in a “higher” voltage supply.
Regards, Gerrit
Last edited:
Dear Gerrit
For what it is worth, check page 17 of this specs sheet for LM317: https://www.ti.com/lit/ds/symlink/lm317.pdf?ts=1733197740345
I know from experience that this circuit works for higher current needs. But yes, if you go beyond 40 volts of drop, even for a mili-second, the LM317 will die. SS devices are not anywhere near as forgiving as tubes/valves...
Ian
For what it is worth, check page 17 of this specs sheet for LM317: https://www.ti.com/lit/ds/symlink/lm317.pdf?ts=1733197740345
I know from experience that this circuit works for higher current needs. But yes, if you go beyond 40 volts of drop, even for a mili-second, the LM317 will die. SS devices are not anywhere near as forgiving as tubes/valves...
Ian
If the circuit works as a limiter, not a current regulator (which has a stable, accurate and predictable value), a MOSFet based limiter would work reliably, provided the MOS is usable in linear mode.
Something like this for example:
Something like this for example:
Ian, I only need max. 100 mA.
Elvee, This could work of course, but when the load is a tube audio stage, we cannot mess around with audio ground like this. At least I wouldn't do this.
Of course it can be used in a positive rail as well with a DC input.
Regards, Gerrit
Elvee, This could work of course, but when the load is a tube audio stage, we cannot mess around with audio ground like this. At least I wouldn't do this.
Of course it can be used in a positive rail as well with a DC input.
Regards, Gerrit
You can always reverse the polarity of the semi's, or rearrange the elements like this for example:
That's because the capacitor acted as a short circuit on the output the moment it was connected (or – hopefully – the moment the power was turned on). This means the current flowed through the zener diodes and into the capacitor without any current limit. That likely blew the zener diodes, which then caused the LM317 to 'see' over-voltage and die.
I suggest that you simulate the circuit rather than keep blowing up parts. Specifically, you need to look at the voltage across the LM317 and also at the current through any protection device across the LM317 on start-up and power-down. When you run the start-up sim, set the initial condition on the output capacitor to 0 V and let the transient simulation use the initial conditions.
Beware that the simulation does not show the complete picture, especially when parts fail. You can have 1 kA flowing briefly in a zener diode without the simulator complaining, but in reality that diode will disappear in a bright flash of light.
Tom
So here is a question: What should happen when the current limiter engages? You say that you want this circuit to work at higher input voltages (400 V). Your original post says that you want a 62.5 mA current limit. That's 0.0625*400 = 25 W dissipated in the current limiter when it engages. How are you planning to get rid of the heat generated from this?
What are you trying to accomplish with the current limiter? Could what you're trying to accomplish be accomplished better with a different circuit?
To work around the power dissipation, you could look at fold-back current limiters. They reduce the output current to near zero once the limit has been exceeded. The drawback of fold-back current limiting is that they have two stable operating points and you need to make sure that the limiter doesn't start up at the wrong point or trip accidentally during a momentary glitch during operation.
Tom
What are you trying to accomplish with the current limiter? Could what you're trying to accomplish be accomplished better with a different circuit?
To work around the power dissipation, you could look at fold-back current limiters. They reduce the output current to near zero once the limit has been exceeded. The drawback of fold-back current limiting is that they have two stable operating points and you need to make sure that the limiter doesn't start up at the wrong point or trip accidentally during a momentary glitch during operation.
Tom
The above circuit has a problem with self-generated noise at higher voltages. Walt Jung explained this in this PDF and I can confirm it from my own tests.
https://refsnregs.waltjung.org/ax_High_Perf_Current_Regs_Revisited_0409_052921.pdf
https://refsnregs.waltjung.org/ax_High_Perf_Current_Regs_Revisited_0409_052921.pdf
Absolutely correct.
But the title of the thread is about LM317 configured as a current source for high voltage.
Not how to make a CCS for HV with the lowest noise.
In the latter case, one can simply replace the LM317 by a suitably low noise depletion MOSFET or JFET.
LU1014 in triode cell comes to mind, other than DN2540, BS129, etc.
Patrick
But the title of the thread is about LM317 configured as a current source for high voltage.
Not how to make a CCS for HV with the lowest noise.
In the latter case, one can simply replace the LM317 by a suitably low noise depletion MOSFET or JFET.
LU1014 in triode cell comes to mind, other than DN2540, BS129, etc.
Patrick
Last edited:
gerrittube mentions a tube audio stage so I assume he doesn't need an additional noise source in the power supply. Otherwise, noise at 200V is audible enough on the speaker so it doesn't even need to be measured.
Let's say an IXTP1R6N50D2 would be enough for this application without much complication.
Let's say an IXTP1R6N50D2 would be enough for this application without much complication.
The path to guaranteed success is to build a true current source circuit including enormous heatsink, which is able to continuously dissipate 40 watts. Now it's nonfatal to ground the output, putting 400V across the current source while conducting a constant 0.1 amperes. Pay for a big heatsink, pay for a large diesize pass device (either MOS or BJT) whose Safe Operating Area includes 400V @ 0.1A (maybe N devices in parallel plus ballasting), pay for control circuitry which does the right thing under fault conditions and also under normal conditions, pay for the PCB area and chassis volume to include these. Then: Guaranteed Success.
Or just pony up for a depletion mode mosfet of proper voltage and current rating. A source resistor and stopper resistor is all you need. To set the output current. You'll probably need to do some fiddling around with a pot to set the proper source resistor value (they ain't all alike).
An IXTP08N50D2 will get the job done - I use them all the time in lower voltage applications. The SOA curve includes 100 mA, but you might need a largish heat sink.
An IXTP08N50D2 will get the job done - I use them all the time in lower voltage applications. The SOA curve includes 100 mA, but you might need a largish heat sink.
Depends on what dynamic impedance you want to aim at.
All described in the Jung articles.
Patrick
All described in the Jung articles.
Patrick
@gerrittube : you should give some elements of context, to understand the purpose of the limiter.
The answers will differ very much depending on the actual use: do you need a simple protection limiter, a high perf CCS having infinite impedance, frequency response, compliance, accuracy, etc or something in-between?
The schematic of the circuit where it will be implanted would be very useful
The answers will differ very much depending on the actual use: do you need a simple protection limiter, a high perf CCS having infinite impedance, frequency response, compliance, accuracy, etc or something in-between?
The schematic of the circuit where it will be implanted would be very useful
That all goes by the wayside if you apply enough voltage to the LM317 to outright blow it up. The original poster was talking about 165V, quite enough voltage to do the deed.
Elvee,
I was looking for a current limiter (say 100 mA), regardless of the applied voltage (above a certain value of course). The intended range would be from 150 to 400 Volt.
Perhaps a 10M45S will do, I haven’t tried this yet.
Regards, Gerrit
I was looking for a current limiter (say 100 mA), regardless of the applied voltage (above a certain value of course). The intended range would be from 150 to 400 Volt.
Perhaps a 10M45S will do, I haven’t tried this yet.
Regards, Gerrit
Limiter means a safety feature, thus neglecting other aspects like output impédance, noise bandwidth, etc.
The (crude) circuits I proposed fit the bill, as would other options. Complicated and elaborate optons may have better properties, but they come at a cost.
Basic principle is KISS. If you add useless features, that's your problem and responsability......
The (crude) circuits I proposed fit the bill, as would other options. Complicated and elaborate optons may have better properties, but they come at a cost.
Basic principle is KISS. If you add useless features, that's your problem and responsability......