Can I get a 150V DC output from the circuit below by using LM317HV-MIL? As the voltage input I will be using a 160V unregulated input voltage from a rectifier.

If so, is there anything that I have to be careful such as power ratings, Vmax, Vce, etc. of the resistors, transistors, capacitors, etc.?

Here is the datasheet: http://www.ti.com/lit/ds/symlink/lm317hv-mil.pdf

Thanks for your help in advance.

enter image description here


Datasheet says, " Since the regulator is floating and sees only the input-to-output differential voltage, supplies of several hundred volts can be regulated as long as the maximum input to output differential is not exceeded, or in other words, do not short the output to ground."

What I did not get is if the input is 160V, I will need C1 have a voltage rating higher than 160V as well as C2 and C3. Is that right? Also, what should be the power ratings of the resistors R5 and R4. Could you help me to calculate the required power rating? For 150V, I calculate it as around 5W. What the R4?


How about this?

enter image description here

  • 3
    \$\begingroup\$ Yes but NOT like this. Vin-out = 60V Absolute max at any time. There is a solution by level shifting the 0V ground to >100V \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Apr 6 '18 at 23:51
  • 1
    \$\begingroup\$ What are you trying to achieve with this power supply? What's the load look like? Is this for a bench supply of some kind? Or are you trying to power a Nixie? What's the context? \$\endgroup\$ – jonk Apr 7 '18 at 0:12
  • \$\begingroup\$ I am going to drive a piezo actuator with this. The actuator has around 4uF capacitance. There is a switching circuit at the tail of the regulator that charges and discharges the piezo actuator. \$\endgroup\$ – CMarine Apr 7 '18 at 0:16
  • \$\begingroup\$ no that circuit won't work. if you turn the R5 down the LM317 will break. C2 could be a problem too. \$\endgroup\$ – Jasen Apr 7 '18 at 1:57
  • \$\begingroup\$ What's the frequency? \$\endgroup\$ – jonk Apr 7 '18 at 5:43

It is possible but not this way. the 0V reference level for ADJ bias current must be level shifted up to say 140V using an "active zener" voltage divider regulator.


simulate this circuit – Schematic created using CircuitLab

Use 140V active Zener + 10V LM317 design = 150V

  • \$\begingroup\$ Thanks!! How can I add some extra protection to limit the current demand of the load? What if I run 30 Amps for 500 micro seconds? \$\endgroup\$ – CMarine Apr 7 '18 at 0:20
  • 1
    \$\begingroup\$ You can follow the App Notes , datasheet but remember if the input short circuits for any reason the chip must be protected with a reverse diode. Design requries thought about ESRC with several caps say from 0.1us to 10 or 50uS transients and then bypass IRs for a Darlington bypass catches up and supplies the current. SHort grounds, low inductance, V=LdI/dt etc EMI control, twisted pair , ground plane etc etc.. lots of design examples on the web \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Apr 7 '18 at 1:03
  • \$\begingroup\$ inductive loads or arcs also need TVS clamps or "OVP" to protect from release overshoot from load side switched. Ferrite chokes, etc ertc . Why not buy a PSU from Meanwell? It's probably cheaper. But the power of 3kW for 0.5ms is 1.5 Joules in energy storage but for 50us = 0.15J =k CV^2 for k being some valuefrom 0.5 to x? But a phase controlled triac or variac supply will drop from no load to full load depends on Pd wasted power and Vmin for LDO so power is a premium for rep rate of pulse. vs cost of big caps. If rep rate is high then string of batteries on floatcharge instead of $$$cap \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Apr 7 '18 at 1:19
  • \$\begingroup\$ Because variac bridge can decay, unless you sync, load pulse during charge peak. RC=8x 0.5ms= 150V/30A * C for 10% sag....etc etc \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Apr 7 '18 at 1:23

In a word - NO. It is rated for 60 volts input, but output has resistors and capacitors to ground. The difference between input and output can be 60 volts minus 1.25 volts. The LM317HV is called 'High-Voltage' because most regulators of that type have a 40 volt limit.

That is why in the application images it shows a output of 50 volts with an input of 60 volts or less.

This is a limit set forth in section 6.1 of the pdf.

Input–output voltage differential −0.3 60 V

Since the output has resistors and loads connected to ground, the device is essentially referenced to ground for voltage adjust and loads. You are limited to the 60 volt MAXIMUM input.

You can use a beefy zener diode to raise ground up 140 volts or more, but this becomes your new 'ground' reference and if it shorts the LM317 will short out as well. It can be done if you accept the risk...

Below is an 'Active' zener diode using a bjt to boost the load current. For high voltage and high current Q1 becomes a bit expensive. It can be replaced with a MOSFET if high power is required.


simulate this circuit – Schematic created using CircuitLab

  • \$\begingroup\$ Thanks!! So, I understand that as long as the input-to-output differential is within the given limits LM193's and 2N905 should be fine in terms of Vce. How can I transform the above circuit such that I can get a 150V DC output? \$\endgroup\$ – CMarine Apr 6 '18 at 23:59
  • \$\begingroup\$ LM317 is a differential regulator so you can regulate 1kV (in theory) as long as input does not vary more than 40V. just active zener must power ADJ bias current. so @CMarine you must define specs for Vin Min:max under load with ESR of source and C bulk storage for RC=8*500us roughly. dV/dt=I/C \$\endgroup\$ – Tony Stewart Sunnyskyguy EE75 Apr 7 '18 at 1:24

First, zener in series with capacitor (C1 and C3, as shown in EDIT2) will not work - where is no path for discharge current until voltage of "-" pin drops to zero. Use caps rated for Vin and Vout, and connect them directly.

Second, in case of output overload, voltage drop across LM317 may become as high as Vin, which will breake it. This is why I think using this regulator IC in such configuration is not a good idea.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.