Comments on: Lowering power consumption tip #3 – Using Relays

By: Nigel Jones

Nigel Jones — Sun, 10 Jan 2010 12:55:09 +0000

Michiel:Thanks for your informative comment.I'm skeptical that EMC is an issue. Many embedded systems contain switching regulators that are switching a lot more current at a lot higher frequencies – and often in discontinuous modes of operation. Now granted many of these designs use shielded coils which help to dramatically reduce emissions. However, I've also seen many designs that don't use shielded coils.As to the demise of the UC3702. I think the explanation is simply that people are using microprocessors to perform this function and hence the demand for the chips is drying up.I took a quick look at the Maxim solutions. They seemed unnecessarily complicated to me. Certainly they didn't seem to offer a compelling advantage over PWM.Regarding the Tyco recommended switching frequency. I can't say I disagree with them. However this is definitely a technique where you have to experiment to find the optimal setting. Switching at 20 kHz will work. However your switching losses will be ten times higher that switching at 2 kHz. Thus it's perfectly reasonable to start at 20 kHz to prove the technique, and then start dropping the frequency in order to cut down on your switching losses.Based on your comments it appears that you are researching this technique for a new project. If this is the case then I'd appreciate it if you'd post your results here as and when you implement whatever technique you end up using.

By: Michiel

Michiel — Sat, 09 Jan 2010 14:06:50 +0000

I find it suspicious that there used to be some integrated solutions for PWM driving a relais, e.g. the UC3702 from texas instrument, but the chip is obsolete for some time and not recommended anymore in new designs. I found no replacements for these chips. Maxim has some relais drivers with power saving functions, but they lower the voltage with a linear regulator instead of using PWM (which is less efficient). Since the technology existed, but is no not used anymore, the cause is maybe EMC? In http://relays.tycoelectronics.com/appnotes/app_pdfs/PWM_application_note.pdf, a minimum PWM frequency of 20 kHz is recommended, although they also warn for EMC problems.

By: Nigel Jones

Nigel Jones — Sat, 07 Nov 2009 00:55:55 +0000

I'm sure that's possible – but I haven't observed it. If vibration was an issue with a product I probably wouldn't use the technique. There again if vibration was an issue I'd probably not use relays at all!

By: Jonathan Jones

Jonathan Jones — Fri, 06 Nov 2009 22:16:32 +0000

Presumably reducing the current through the relay coil will also reduce the force holding the contacts together.Could this result in the contacts running hotter and affect the reliability of the end product ? Ditto reduced resistance to vibration.This is more likely to be an issue if the relay contacts are run near their specified rating.

By: Nigel Jones

Nigel Jones — Fri, 06 Nov 2009 13:03:08 +0000

I've never experienced an EMI problem using the technique. I suspect that the emissions are trivial compared to interrupting the load that the relay is switching. I have used two microprocessor ports – but not in the way you described. What I've done is generate a PWM signal that is the PWM source for multiple relays. Then through judicious external logic I can feed either a 1, 0, or PWM signal to any given relay.

By: Kasper

Kasper — Thu, 05 Nov 2009 16:16:09 +0000

Regarding insignificant overall power savings: the relay is usually actuated by 5V or 12V. As Nigel said, the relay usually dominates the power budget. This means that a more expensive regulator might have to be chosen because of continuous output power requirements.Nigel, is there possibly an issue with EMI when holding a relay with PWM? What if two microcontroller outputs were used? One would control the operating, the other would control the holding, using a series resistor to limit the hold current.

By: Nigel Jones

Nigel Jones — Tue, 03 Nov 2009 12:08:57 +0000

I suspect that most readers will agree with you. However (as usual) I'd like to offer a contrary point of view. The 200mW you mention is of course insignificant in terms of the overall power consumption. However, it is unnecessary power consumption, whereas the 2000W heater is presumably necessary power consumption. Now clearly, one system saving 200 mW doesn't change much. However, if at any given time around the world there are a billion relays turned on by an embedded system, then the power savings would be 200 MW. As the saying goes, a billion here, a billion there and pretty soon you are talking real money!

By: Kepa Diez

Kepa Diez — Tue, 03 Nov 2009 08:15:36 +0000

A nice pice of information, as usual.But, I think that in most designs relays are used, when activated, to turn on a high power load (say, a 2000W heater element). In this case, is it worth to save 200 mW, when this only accounts for a 0.01% of the whole?There is no doubt that this technique can be very useful indeed, especially taking into account all the new low power and standby regulations, however I am afraid that it won't fit most projects.