I've got a 30V 2A isolated DC/DC supply. I need to be capable of outputting 1A for short periods of time (3 sec, exception condition only), so with temperature derating that's what I picked.

However, in normal operation the current draw is only 20-60mA. The poor efficiency of this is really bothering me. (It's like 10%)

How much trouble would it be to use two 30V supples, one for say 150-300mA and the other this 2A supply, and switch between them as the current needs change? How to best handle the transition? If using FETs, should I swap the FETs at the same time and depend on caps to carry the current during the transition?

I can handle maybe 500mV droop as it swaps. Limiting at 150mA for a few hundred milliseconds while the 2A supply comes up is also okay.


Okay, so here's a bit more detail. I'm controlling a network of devices. Depending on the number of devices, this can draw anywhere from 20-300mA. Most configurations would probably be around 60mA.

If there's a short somewhere on this network, the devices can disable sections of the network to work around the short... but they require about 1A for 3 seconds to detect this condition. This test would reoccur periodically to detect when the network is restored. The devices are someone else's product and so the behavior can't be changed.


Sorry, input voltage range 9-36V nominally 24V.

  • \$\begingroup\$ are any of the parameters of the dc/dc converter under your control, i.e. did you design it yourself and can change operating mode and frequency? also, first you write that your current draw is 20-60 mA then you want to swap between 150-300 mA and 2A. 300 mA is much higher than 60 mA. which one is it ? :) \$\endgroup\$ – Bjorn Wesen Jun 30 '11 at 15:13
  • 1
    \$\begingroup\$ Can you explain more about what this thing is thats pulling power? It may be easier to approach it from having 1 power supply for the high power part of the system and then a separate power supply for the low power portion. Like if it was a micro powering a motor, you would have 1 rail for the micro and another for the motor. \$\endgroup\$ – Kellenjb Jun 30 '11 at 15:19
  • \$\begingroup\$ @Bjorn Wesen the way I read it, the reason he said 150-300mA was so that it could still provide a more then the 20-60 mA typical for use during the transition stage. \$\endgroup\$ – Kellenjb Jun 30 '11 at 15:26
  • \$\begingroup\$ @darron - You say a DC/DC. What's the input voltage? \$\endgroup\$ – stevenvh Jun 30 '11 at 15:27
  • \$\begingroup\$ @steven - He also said 30 V. \$\endgroup\$ – Kevin Vermeer Jun 30 '11 at 15:32

As already asked - do you really care about the lost energy? Cost wise it's not liable to matter. Thermally it may.

You can easily enough achieve what you want, not by switching between supplies but instead by adding in a second supply when required. Your acceptable response time of "several hundred milliseconds" is 10+ whole mains cycles so (assuming this is mains powered), so response time from a "standing start" should be easily enough achievable.

You could feed the second supply to the load via a diode to allow it to be turned on and off "somewhat gracefully". You don't say how precise you need the voltage to be but with proper design you should be able to achieve "take up" by the larger supply as the smaller supply droops somewhat or have the voltage rise 'just slightly' as the large supply cuts in.

If you don't mind a relatively minor piece of custom electronics then a custom regulator and switch would be relatively easy to implement. Large supply can then be able to make somewhat more voltage than needed at say 2A and the regulator/switch drops the voltage difference to ensure that the 30V output is maintained.

This could almost be done with just a linear regulator (eg LM350) and a diode but the regulation would be imperfect (maybe a few tenths of a volt variation across zero to full load due to the LM350 voltage sense being somewhat decouples behind the output diode). Adding a single opamp section and a low cost voltage reference would allow the 30V rail to be maintained "rock steady" once the large supply had reached design output level.

BUT! - your bricks have an unacceptably bad no load rating. Modern supplies are able to do far better than what you are quoting. Wattages under 1 Watt idle are typical for supply designs targeted at consumer appliances where standby power is now considered a major issue. (Rule of thumb: 1 watt always on typically costs ~~= $2/year - obviously depending on your power charges)

As has been noted, you can certainly buy supplies that come much closer to what you want than what you have now. BUT the admitted 'kluge' of a custom off the shelf supply optimised for very high efficiency at say 150-300 mA plus a custom "pick up the load" system would allow a very quick and easy and overall very efficient system. If you do this you MUST ensure that the large supply is never required during normal operation. ie if the low Wattage supply can supply up to 300 mA at acceptable voltage, normal operation MUST NEVER exceed 300 mA or the large supply will be cutting in and out semi randomly. The custom large wattage supply needs some means of detecting the need for high current and switching in for a finite period. This could be based on droop of the small system under load or by the test system providing a "wakeup" signal. An off the shelf power supply may be available to do this but the requirement is specialised enough (even though simple) that making your own is probably preferred if the capability is available.


Is efficiency when drawing 20-60mA really that important? You only say it's "bothering" you. Perhaps the answer is simply to get over it?

60mA at 30V is 1.8W. You say the supply is 10% efficient at that current, but that would imply a total input power of 18W. I find that hard to believe.

It is certainly possible to make or get DC power supplies that have good efficiency at low currents too. Instead of making a kludge to switch between two supplies, get the right 30V 2A supply.

If your existing 30V 2A supply really is drawing 18W in for 1.8W out, then it will probably draw most of that even with 0 power out. In that case switching between two supplies isn't going to help since the 2A supply is still sitting there wasting the power.

  • 1
    \$\begingroup\$ @Olin - the 10% efficiency is pretty bad, but for any PS it would become worse at lower loads (until it's 0% without load). I presumed the 10% is at the lower 20mA. But 6W still isn't very good... \$\endgroup\$ – stevenvh Jun 30 '11 at 15:47
  • \$\begingroup\$ Yeah, it's about 5W at 20mA. \$\endgroup\$ – darron Jun 30 '11 at 15:53
  • \$\begingroup\$ The supply has an enable line that seems to work pretty well. \$\endgroup\$ – darron Jun 30 '11 at 15:55
  • 1
    \$\begingroup\$ "Bothering" in the sense that I have a whole lot more power to dissipate out of an enclosed box. I can add more thermal management to the more extreme configurations. \$\endgroup\$ – darron Jun 30 '11 at 15:57
  • \$\begingroup\$ @darron: 5W at 20mA isn't very good. That particular supply was apparently not optimized for efficient low current performance. That's not necessarily bad since there are many applications where that's just fine. However, yours isn't one of them. Look around and find a supply that is efficient at low currents, or roll your own by using a pulse on demand strategy, for example. \$\endgroup\$ – Olin Lathrop Jun 30 '11 at 15:57

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.