I'm trying to combine several SMPSs each having a ~350VDC output @100mA each.


Purchased an unused batch of small DC->DC SMPSs in good industrial quality. Unfortunately they're too weak for my purpose.


  • Input 24-30VDC
  • Output adjustable 75VDC-400VDC @ 40W max load.
  • Non-isolated ( Input and Output shares GND. )
  • Switching circuit operating ~ 75KHz.
  • Measured: 350V @ 100mA => Ripple < 3V.

The SMPSs have built-in soft-start + short-circuit protection, but apart from adjustable voltage, they have no other control-options.

I was hoping to combine say 7 of these SMPSs using "simple" non-foldback current-limiters and connect their outputs in parallel. (The current-limiter in mind is almost similar to the ones described in this topic.)

In my case one dedicated SMPS is used to drive the voltage (emitter follower.)

The "driver" will deliver the desired voltage and the remaining SMPSs should be in current limiting mode, sharing the load equally between them.

At least, that's the result I'm aiming for. ;-)

One Voltage Driver" and 7 current limited "Slaves"

There should be a total of 7 or 8 current sources. ( Only 3 is shown here ).
Resistors / Capacitors / Voltages has not been calculated.

  1. Is the constant voltage driver + several current limiters a good approach?
  2. Do I need diodes ( D2 - D8 ) at the output of each current limiter?
  3. Is it necessary to have a resistor ( R1 ) or a "current limiter" in series with the output of the CV-mode "driver" ?

Having a "driver" capable of delivering relatively high current should eliminate the need for Darlington BJTs as serial regulators.
However this also means that the current-limiting "bypass" BJTs need to deal with higher currents.

  1. Can I use normal BJTs ( TIP50G or MJE18004G ) or are Darlington BJTs a requirement?

Can I use 2 standard BJTs in a Darlington setup?

  • \$\begingroup\$ You will also have to make sure that the supply can provide enough current for all of them when they happen to operate in phase. I have a feeling it will not be good for the supply, unless it is current limited (and possibly capable of tolerating a 75KHz pulse overcurrent situation). \$\endgroup\$
    – Indraneel
    May 22, 2019 at 16:50
  • \$\begingroup\$ Thanks for bringing this issue up as well. :-) In my case the 24VDC is supplied from an array of lead-acid batteries ( +1000Ah total ). This power source is "unregulated" so I'm sure they will be able to handle potential "in-phase currents", even if these should exceed the average current by a factor 10. \$\endgroup\$
    – Smeden
    May 22, 2019 at 17:35
  • \$\begingroup\$ Constant current is the way to go here. What circuit are you proposing? \$\endgroup\$
    – winny
    May 23, 2019 at 6:07
  • \$\begingroup\$ @Indraneel Given adequate output capacitance, that switching frequency is not an issue and “pulse overcurrent” won’t be a thing, especially if Smeden adds external constant current regulation. \$\endgroup\$
    – winny
    May 23, 2019 at 6:10

1 Answer 1


Consider what happens to your current limiter circuit under overload conditions. If we look at one circuit alone and assume it has resistor values chosen to limit the current to 100 mA, what happens if the input is 350 volts and the load is 1 kOhm? Since the current is limited to 0.1 A, the voltage on the load is 100 volts, so the drop across the circuit is 250 Volts and the circuit must dissipate 250 volts x 0.1 amp or 25 watts. So one of your circuits will fail quickly whenever an overload occurs, followed by the others.

You do need the diode if driving loads with any capacitance, because if you turn the drive circuit off or lower it, the result will be a momentary reverse voltage on the base-emitters.

You do need R1 for some current limiting because otherwise Q1 and Q2 are simply diodes if they become saturated and the left-hand supply will see the entire load.

Good luck!

  • \$\begingroup\$ Thanks for yor valuable input. The D2,D3,..D8 diodes will stay. :-) <br/>What will be a good choice for R1? I know that the V1 "driver" Voltage will have to be cranked up as well.*<br/> Do you think the "Emitter degeneration" resistors R3, R5,...,R15 are necessary? *The calculations would be a lot easier without them. <br/> I will try to implement fold-back when the basic problems are solved. ;-) <br/> Thanks again. \$\endgroup\$
    – Smeden
    May 23, 2019 at 22:13
  • \$\begingroup\$ The combination of R1 and R3 set your closed-loop gain. I wouldn't try calculations, but would use a SPICE program; many are free and they work well for these types of analyses. Your concern is how the circuit will react to a change in load or upon start-up. I wouldn't be surprised if the circuit oscillated or had to ring down without R3. Also, there is no problem creating your own Darlington pair from individual transistors. Good luck! \$\endgroup\$ May 24, 2019 at 13:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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