Relay Power Dissipation Reduction 2

[itsmoked]

I have this 12Vdc relay that I unfortunately have 14V to operate. It can take it but jeez it gets warm.. It's also in a small enclosure that warms up more than would be ideal.

I can stuff in a dropping resistor that will lower the relay's temp but will, of course, do nothing for the box ambient because the same power will be dumped in the box.

So I thought I'd try cycling the power to the relay fast enough that it can't respond. That way I could essentially PWM my way to less average power use in the relay.

So I added a little routine that cuts the power off to the relay after 5ms for 1ms then turns it back on.

Damned if that relay doesn't drop out in 1ms!!! Dang.

Thought you guys would find it interesting.

Here's the relay:
OJE-SH-112DM,000
OJE_series_relay_data_sheet




Keith Cress
kcress -

http://www.flaminsystems.com

 

[MacGyverS2000]

How much power does this relay require to operate such that you're warming up ambient? That's a LOT of power for a pull-in coil...

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

Can you put a capacitor across the coil? Then you could basically modulate the DC voltage across the coil with your SW defined PWM.

Obviously this would add a few ms to the timing.

If there's a spare contact, or even a suitable switched DC line, then you could provide the uC feedback as to the state of the relay contacts, and your SW could adaptively learn the contact drop-out point, and stay something like 20% higher than that for the steady state ON.

 

[itsmoked]

MacGyverS2000; The box is merely 8 cubic inches, made of plastic so it isn't very good at conducting away heat. Add a relay dumping 640mW and it makes a LARGE difference in the box's internal ambient. Something around 20C.

VE1BLL; I'll give the cap a try. Not much room for one.. No aux contacts and a single bounce while the cpu "learns" would have people screaming.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[Skogsgurra]

No auxiliary contact? That's bad. But there's still hope.

DC coils usually drop out at around 10% of nominal excitation. So, if you add a series resistor that reduces current to 20%, you will most likely be fine.

But, theres that big BUT - the relay will never pull in.

BUT, it will if it gets a full 14 V for a few milliseconds. The armature will start moving and when it gets closer to the sealed-in position, it needs gradually less and less current. If there's room for a small resistor and a capacitor parallel to the resistor then I think you are home (almost) free.

BTW: 14V*(full current)*14/12 is a lot more than 14V*(full current)*0.2. So, you will only have around 17% of today's dissipation. I don't think that will overheat the box. Not at all.


Gunnar Englund

http://www.gke.org

--------------------------------------
Half full - Half empty? I don't mind. It's what in it that counts.

 

[Lakey]

Suggestion: Adjust the program so that full voltage (14V!) is applied for enough time to allow the relay to pull-in fully, then apply PWM at > 20% coil volts.

Regards,

 

[mcgyvr]

Can you use something like this?

http://electronicdesign.com/site-fi...nicdesign.com/content/content/63948/63948.jpg

 

[IRstuff]

The datasheets graphs show a release time of about 1ms, but there's in the datasheet, a low-power version of the same relay, with and LM designation

The datasheet also implies that the relay will stay activated down to 0.6V, so you could possibly put a 3-terminal regulator like an LM317 to drive the relay at full voltage and then short out the bias resistor to drop the hold voltage down to the LM317's minimum of 1.2V.

TTFN
faq731-376

Need help writing a question or understanding a reply? forum1529

Of course I can. I can do anything. I can do absolutely anything. I'm an expert!

 

[OperaHouse]

A capacitor can be too bulky or you may run out of PWM pins or not want to slow down a program with an extra routine. Consider driving the relay with two transistors. One would be at full power for a short period and the second parallel with the voltage dropping resistor for hold in.

 

[MacGyverS2000]

8 square inches? That's a tiny spot to work in...

Can you increase the rate of your PWM, or are you stuck with 1ms time slices? Even a doubling in rate should be enough to keep hold-in voltage. I was also going to suggest VE's cap idea.

If you can't adjust PWM rate, what about a simple 555 addition? Obviously needs a touch of power itself, but hopefully less than the 100mW you're burning in the coil now. Maybe a 5-pin PIC chip?

How about using a DC-DC converter to lower the voltage? Here's a tiny one from Micrel:

http://www.micrel.com/index.php/en/...down-internal-switches/article/18-lm2574.html

Here's another I've used from Enpirion:

http://www.altera.com/devices/power/enpirion/regulators/switching-regulators/er3105di.html

Set the output voltage to something like 9-9.5V for a safety margin (the relay is good to 8.4V), and bask in the overall power savings... assuming an 85% efficiency (which is quite reasonable for these things), your power draw should be less than if it was running at 12V.

Dan - Owner

http://www.Hi-TecDesigns.com

 

[VEBill]

I can stuff in a dropping resistor that will lower the relay's temp but will, of course, do nothing for the box ambient because the same power will be dumped in the box.

A series resistors will dissipate heat of course, but the total power/heat from the series resistor and coil should be less than would the coil alone.

 

[itsmoked]

Auxiliary contacts?? This is a SMALL relay in a cost sensitive application.
Ah Gunnar, I get it. Same as mcgyvr's linked schematic. Yeah this should work if I can fit it.

Lakey; That's what I was attempting. I have an interrupt routine that runs at 1ms that was easy to ride this relay PWMing on. Except the relay is so fast it drops in the tiniest period I can present. The two hardware PWMs that could definitely be running smaller periods are pre-occupied.

mcgyvr; Yeah I'm going to try if I can fit it. A 150uA cap is not relatively small. See below.

Ir; I so wish I could use the sensitive version but no one stocks a washable one and I don't need a 1,000 in three months. As you can see I don't have room for an LM317 and the board space needed for heat dissipation for it. I suppose I could add a thru-hole one.

Here's a board shot of the end of interest. The vertical dimension is 2.4"

OperaHouse; I could do your dual run in a second since I have about 8 spare outputs. BUT! in this particular case I am having to run this relay thru a huge opto isolator(U6) because it's running on the switcher's (U3/L1 et.al.) un-isolated output and the processor is running on the isolated output. So your scheme -in this case- would require another large opto unfortunately.

Dan; It's a box of 8 cubic inches, and the board is.. well actually about 8 sq ins too. LOL And yeah I think I'm stuck with about a millisecond slice size. No room for a 555. A 5pin pic would probably fit but having to program it too would run up assembly costs due to the PITA it presents. The switcher mentioned above has only a 33uH inductor to support it's 1Mhz and it uses a bunch of the board so there's no way I can see of stuffing another one in.

VE1BLL Let's do some math on just a single dropping resistor solution.
The coil is 320ohms.
The minimum pull-in voltage is 8.4V

Using 14V -> 14² / 320 = 610mW now.

Using a resistor that will drop 14V down to 9V.
9V/320ohms = 28.1mA
The resistor to back this up (14V-9V) / 28.1mA = 178ohms.
It will dump 28.1mA² x 178 = 140mW or a SM1206 (1/2W) which isn't too big.

Total power will now be 14V x 28.1mA = 394mW. A delightful 35% reduction!
We have a winner! This is what I'll do.


Keith Cress
kcress -

http://www.flaminsystems.com

 

[BrianG]

They say never assume, but I assume your circuit includes an anti back emf diode across the coil to protect the switching device? If it does then I would be surprised if the release time is as short as the 1ms shown, since the spec would normally be measured just for the relay on its own. Usually when a diode is fitted it will increase the release time quite significantly, sometimes double, depending on the relay design.

 

[VEBill]

If you're using a uC, and if you have another spare output pin, then you could have two resistors. One would be chosen to apply the minimum reliable PULL IN voltage, while the other could be chosen to apply the minimum reliable HOLDING voltage. So the power and heat would be related to the lower holding voltage typically. Having the ports in simple wired parallel probably requires Open Collector outputs, otherwise you might need diodes.

You could keep this 2-pin concept in your back pocket in case you need to further reduce the power and heat.

 

[itsmoked]

BrianG; Can you see the picture I provided above? If yes, see the relay?
Look between the top two pins. Yes, I have included the freewheel. Yes, I have heard that freewheels delay opening significantly. I have never witnessed that and this one certainly doesn't! You can ask my dog who's very twitchy and bolts as soon as I pwm the relay with 1ms off periods.

VE1BLL Please see my response to OperaHouse right below the picture directly above.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

Ah, missed that. Carry on then.

 

[itsmoked]

BrianG;

With your comment and Gunnar suggesting my diode was bad or missing since with the diode the relay should definitely stay in longer, I went back to the prototype to check. Because the relay was laid out upside down it had to be soldered in on the 'other side of the board' where it would be on top of the SMD diode. The relay is raised just like it's on top of the diode. But on very close examination the relay is raised because it's on jumper wires - NOT the diode. The diode is missing. BINGO you two!

With the diode scabbed on the 'other side' the relay will stay closed about 3ms after the power is removed instead of the measured 800uS without the diode.

So the PWM gambit should actually work fine - with the diode. There is one caveat I discovered on the bench. If, when you call for the relay you do it with the PWM signal the relay will chatter closed requiring two and sometimes three PWM periods before sealing.

Two ways to combat this are to turn on the relay and after some delay start PWMing after the relay has sealed.

OR

The PWM period must be set up so the ON period is always about 6ms followed by the off period being 2ms for power reduction and to guarantee the relay seals before the first PWM OFF period shows up. Of course you must also set things up so the PWM period starts a fresh the moment the relay is activated and that you don't just randomly gate in a free running PWM.

Keith Cress
kcress -

http://www.flaminsystems.com

 

[VEBill]

If you plan to make use of the diode to maintain the current through the coil with PWM banging away on it, then make sure you double-check the diode specs for such (ab)use.

 

[btrueblood]

As a non EE type, I shudder when I think of the RF noise generated by PWM'ing a solenoid, especially if there is a cap included in there somewhere. I'd use the dc method myself. I presume you've got good shielding, and have tested the function both installed in the box as well as outside.

 

[MacGyverS2000]

Nothing beats a good PWM square wave like an inductor/coil

Dan - Owner

http://www.Hi-TecDesigns.com