What would be the minumum PWM frequency to hold a relay?

[blueroomelectronics]

I'm assuming a low frequency square wave at lets say 40% duty ~ 75Hz would keep a small DC relay energized. Any problems with this approach?

 

[dknguyen]

Why not just use an on or off?

 

[blueroomelectronics]

Relays once energized don't require much current to stay that way. Keeps the power requirements down and everything runs a little cooler.

 

[Papabravo]

It depends entirely on the mechanism. Current is the thing that pulls the relay in. Usually there is a spring return. If the current drops too low the spring will cause the contacts to open. what you will get is a chattering contact. You're just going to have to try it to see what happens.

Now a latching relay or a double throw relay might be a better choice.

 

[blueroomelectronics]

A latching relay generally requires double the I/O, hmm but an interesting idea non the less.

 

[dknguyen]

What about a PTC?

I guess you coudl use the PWM, but you'd have to almost trial and error while increasing the frequency to find out when the coil relay starts filtering it well enough. And when does the switching losses going through the flyback diode outweight the power saved?

 

[blueroomelectronics]

A software PWM is a pretty near zero cost option, it's for controlling a series of relays (8). I'll build it and test it.
Basically I'm using the OE line on a 74HCT595 + ULN2803A to run 8 relays. On powerup I've got a pullup on the OE line to keep the relays off till I have a chance to set them. After they're running I plan to toggle the OE line while I'm updating an interrupt service routine. 50% is super easy but I've worked with hardware PWM around 1kHz in the past with solenoids and it puts much less stress on the parts (nice and cool).
A typical 5V relay might only need 2V to stay energised. Why burn energy if you don't need to.

 

[dknguyen]

Why burn energy if you don't need to.

Cause it's cheaper (in time, for this case specifically).

 

[k7elp60]

Like you said once a relay is energized the voltage necessary to keep it energized is lower. I realized this a numeber of years ago. I did a experiment to see how the lower voltage affected the contact resistance of the relay contacts. On a 12V relay and a load I checked the voltage drop across the closed contacts with a 25A DC load the voltage across the contacts was 86.7mV with the normal 12V on the coil. With the coil voltage reduced to 6V the voltage across the contacts was 88.6mV. I figgured this was due to the decrease in pressure created by the magnetic field.

This experiment was done over 10 years ago and the motivation was to compare the power comsumed in relay contacts to that of the Rds of power mosfets. I did build a regulator circut to apply the normal 12V to the relay to energize the coil and then a short time later reduce the coil voltage to 6V to keep the relay engergized. At the time my circuit consumed less power than a power mosfet.

In todays power mosfets with extremly low Rds the power consumption is less than the relay.
Perhaps you might try reducing the voltage shortly after the relay is energized.

 

[k7elp60]

I'm assuming a low frequency square wave at lets say 40% duty ~ 75Hz would keep a small DC relay energized. Any problems with this approach?

After my last post I did some more thinking the subject. 75Hz the period is
13.3mS. It might work but you may have to adjust the period as the operate and release time my be different and might cause a problem, as the operate and release time varies on different relays.

 

[Oznog]

It's not so much a reduction in the voltage required to keep it latched, but the current.
No one can say how low the PWM freq can be, for that we need to know the inductance of the coil, the coil resistance, and the forward voltage of the flyback diode. BTW, you know you need a flyback diode in there, right? That's not just to suppress high voltage spikes but it actually sustains the magnetic field in the inductor and reduces the required period.

If you get the latching current too low, it may be prone to unlatching if bumped around since the coil's only putting out slightly more force than the return spring.

 

[mvs sarma]

A polarized relay (with a permanent magnet incorporated) latches by itself without any current. Only it needs a reverse pulse to release.

 

[blueroomelectronics]

The flyback diodes are the ones built into the ULN2803A driver.

 

[kchriste]

Yes, 8 single coil latching relays like mvs_sarma described will work with the addition of only one IO pin. Something like the TQ2-L-5V should work:
http://pewa.panasonic.com/pcsd/product/sign/pdf_cat/tq.pdf
with it's 20ma coil current. You would drive them in the same fashion as you would an LCD display directly. The extra IO pin would be connected to the common of all relay coils (backplane equiv) and would be driven out of phase with any of the other 8 IOs to set/reset a relay and in-phase for no change.

 

[Boncuk]

Add an electrolytic capacitor wired parallel with a current limiting resistor. Connect both in series with the relay coil. This is a reliable and cheap way.

If the switching transistor is turned off the cap will charge to supply voltage. As soon as the switching transistor connects the coil to GND the cap will give the extra current to energize the relay. When it's done the current will decrease thus saving power. The values of R and C depend very much on the relay you are using. Miniature relays hold if the supply voltage drops about 30 to 40% of the nominal voltage.

I do not recommend to use that for power switching since the pressure on the contacts will decrease too.

For relays in a DIP package 200Ohms and 470uF are good values to start with.

 

[Boncuk]

Here is an example of the circuit. The coil restance is 240Ohm and the current without limiting resistor is 56.1mA. With limiting resistor the initial current is high until the cap is discharged. Thereafter the current drops to 37.3mA.

Coils with higher restistance require a higher value limiting resistor for optimum power saving.

To hold the relay the capacitor has no influence.

 

[blueroomelectronics]

I could double the frequency to 150Hz or 300Hz.

 

[Nigel Goodwin]

I could double the frequency to 150Hz or 300Hz.

I still really fail to see why you're wanting to complicate things by driving the relays using PWM? - relays are designed to be permanently powered - the only reason would be to reduce current consumption, but if it's battery powered you wouldn't be using relays anyway?.

 

[Diver300]

On a 12V relay and a load I checked the voltage drop across the closed contacts with a 25A DC load the voltage across the contacts was 86.7mV with the normal 12V on the coil. With the coil voltage reduced to 6V the voltage across the contacts was 88.6mV. I figgured this was due to the decrease in pressure created by the magnetic field.

Most relays don't work like that. There is a spring so that the armature can move further after the contacts have closed. If the relay is holding in, the armature will usually be in contact with the core of the coil, so the pressure there will vary, but the spring will keep a constant pressure on the contacts.

 

[blueroomelectronics]

I still really fail to see why you're wanting to complicate things by driving the relays using PWM? - relays are designed to be permanently powered - the only reason would be to reduce current consumption, but if it's battery powered you wouldn't be using relays anyway?.

It's not mandatory, the design simply supports it. 8 relays drawing 100ma each is 800ma, a simple software toggle of the enable line would cut this in half. (heat) I thought the hardware PWM was overkill and wanted to use it elsewhere.