Relay controlled PSU thermistor

[TELVM]

The idea is to switch off the thermistor after a couple seconds, to keep it cool and ready to manage the next inrush current (as opposed to always hot).

From here: Controlling the Inrush Current Required by Large Transformers

What do you think of this circuit? Could it be simplified any more?

Could it work as is for 230VAC?

[mariushm]

Re: Relay controlled PSU thermistor

Simpler than this would probably be to use a single PIC microcontroller, a decoupling cap, a 5v relay, an npn/darlington and a diode. This way everything can run on the 5v standby rail... the relay shouldn't use more than 50-100mA.
The pic would have internal oscillator, brown out reset, watchdog timer etc, resets automatically when power is lost, and gives you the ability to set a proper timer until you trigger the relay.

http://www.digikey.com/product-detai...%2FP-ND/665881

For 0.58$ it's a nobrainer.

With the above circuit, I'd be concerned if there's a half a second or so power loss, the system may reset suddenly and your circuit still overrides the thermistor and the capacitors pull more amps than the fuse.
But in real life, the capacitors probably won't discharge that fast.

In regular power supply, it's not the transformer that pulls current, it's the large capacitors. In linear power supply it's the transformer, especially if it's a toroidal one.

[joshnz]

Re: Relay controlled PSU thermistor

Looks very simple already.
As for it been able to do 230 vac no problem if using a relay with contacts rated for 230v.

This circuit is only one time though, If the ac drops long enough for the transformer to demagnetize and short enough for the relay/12v to not dropout will still get high inrush current from the transformer.

With the above circuit, I'd be concerned if there's a half a second or so power loss, the system may reset suddenly and your circuit still overrides the thermistor and the capacitors pull more amps than the fuse.
But in real life, the capacitors probably won't discharge that fast.

He doesn't mention if for smps or for large transformer.

[TELVM]

Re: Relay controlled PSU thermistor

I'm thinking about applying the idea to a typical PC power supply.

Many middle/high end PC PSUs came nowadays with relayed thermistor/s, v.g.:

=>EXPAND =>EXPAND =>EXPAND

[tom66]

Re: Relay controlled PSU thermistor

Depends on how crucial it is but I would just wire it to +5V standby, and use a large-ish startup capacitor on the 5V standby (might even be OK with original.) Then, the relay would click in as soon as the PSU is ready to run, so it wouldn't cause issues with the computer trying to boot immediately after power is applied (e.g. wake automatically) and draw more current than the thermistor can safely allow (voltage drop etc.)

This is used a lot in LCD TVs, just straight to 5V standby. Works well.

[budm]

Re: Relay controlled PSU thermistor

What will remove the AC input when the unit is not in used? Is there another switch that removes the AC that feeds the 120VAC_IN? As josh had indicated, it will a one time current limiting only.

[redwire]

Re: Relay controlled PSU thermistor

That style of relay time-delay has never worked good for me. As the cap charges up (to it's whopping 1.4V), the transistors gradually turn on and the relay pulls in "soft" and slow. It does not have crisp switching action. Also, I don't think you'll get a few seconds delay with 150uF/100k to only 1.4V. You would add a zener so the cap has to ramp up to, say 6-9V. If the 12VDC supply ever fails to come up, the PTC will get cooked.

Here's a schematic of a (audio mute) circuit that I found works well although it takes a while to figure out. It's not my design but clever enough with one transistor doing a lot of work.
On power-up the S8550 is on (=relay off) until AC is present (which biases S8550 off) AND the 22uF cap has charged up. The mute input is a manual override to turn off the relay and mute things.
The power-up delay is set by the 47k/22uF/75k to a couple seconds.
The hold-timer delay (for AC mains sags and brownouts) is set by the 2.2uF/4k75 pair

[cheapie]

Re: Relay controlled PSU thermistor

Can't you just use a relay that switches on at ~220 VAC with the coil between out and ground?

[eccerr0r]

Re: Relay controlled PSU thermistor

I think cheapie's suggestion is right (and cheap!), instead of using the output of the +12 or +5VSB just have the relay turn on whenever the main high voltage capacitors are charged up to their appropriate voltage - perhaps just use a 220VDC coil relay across the input high voltage capacitors. Once the capacitor's voltage is high enough the "inrush" is over and the relay can just short out the inrush resistor and we save a few watts of power there.

This should work fine if we lose power for a second, depending on how much the caps discharge the relay would open up only if the caps discharge enough and thus need to worry about inrush again.

The question is how much is "enough". This will determine how feasible this solution is. Ideally the open threshold should be fairly high so that we don't have real inrush problems, but not so high that a temporary drop on cap voltage will open up the relay and now the psu needs to draw power through a really resistive inrush resistor... and the computer crashes...

[TELVM]

Re: Relay controlled PSU thermistor

You may well be right eccerr0r.

But it catches my attention that all, or nearly all, OEM PSU thermistor relays I see feed coil with 12VDC.

=>EXPAND =>EXPAND =>EXPAND

I'm just wondering ...

· As the worst of the inrush current is over very quickly, after just a few milliseconds:

· And since there must be (I guess) some time delay in milliseconds from the instant of switching the PSU on until the +12V rail is fully alive and kicking.

· And since the relay takes some milliseconds from receiving the +12V juice to the closing of contacts.


Could it be possible that the 'natural delay of things' is just OK for this task, no extra delay circuit is needed, and that's what OEMs are doing (feeding the relay directly from the +12V rail)?

[eccerr0r]

Re: Relay controlled PSU thermistor

Quite possibly.

I do have to mention that relays with high voltage coils are large and expensive so likely that's the main reason for not using cap voltage directly. I don't think I have any PSUs with the inrush resistor relayed out, if someone could reverse engineer it that would be interesting. Theoretically speaking 12V direct is good enough too, as that drops off and turns off the relay well before the main caps discharge down to zero. Also the nice thing is that there shouldn't be any droop spike that would mistakenly turn "off" the relay... any spikes like that would have been a problem anyway...

Anyone have some real numbers in losses in this resistor in a properly operating PSU?

[TELVM]

Re: Relay controlled PSU thermistor

A crude test: If any forumer has a PSU with relayed out thermistor, he could keep his ear (not too) close to the PSU exhaust fan while powering on, listen for the relay 'click', and give a rough estimate (v.g. 'about instantly', 'about half a second delay', 'about five seconds delay', or whatever).

(I'd do the experiment myself, but unfortunately don't have any PSU with relay).

[TELVM]

Re: Relay controlled PSU thermistor

Seems the 'click' is seriously intriguing the people :

http://www.evga.com/forums/tm.aspx?m...e=1&print=true

http://www.ocia.net/reviews/oczzx850/page4.shtml

http://www.ocztechnologyforum.com/fo...ain-by-itself&

http://www.silentpcreview.com/forums...65250&p=567236

http://www.overclockers.com/forums/s...d.php?t=721437

http://forums.hexus.net/power-suppli...wered-off.html



Judging from this video:

http://www.youtube.com/watch?v=9llD5oXVTg8

the relay clicks in almost inmediately (less than half a second) after pressing the comp power-on button.

Looks like they could be wiring the relay coil directly to +12V rail ...

[eccerr0r]

Re: Relay controlled PSU thermistor

Hmm... tempted to "add" one into my PSU to save a watt or two ... or not...

Will have to measure the thermistor to see how much loss it really is, at least for a lower-power PSU.

[Behemot]

Re: Relay controlled PSU thermistor

From the PSU I am reviewing ATM, it is almost instantly. I have just put the thing back together…but if you would ask really very nicely, I may be willing to open that thing again and do some reverse engineering…or at least try to

[tom66]

Re: Relay controlled PSU thermistor

Another thing it may be for is to save power in standby mode (not always used to short out the thermistor -- check.) Could even use separate bridge and cap for standby supply.

[TELVM]

Re: Relay controlled PSU thermistor

…but if you would ask really very nicely ...

[Behemot]

Re: Relay controlled PSU thermistor

Another thing it may be for is to save power in standby mode (not always used to short out the thermistor -- check.) Could even use separate bridge and cap for standby supply.

Umm, there actually are two bridge rectifiers. You are saying they may not be in parallel, but in separate circuits?? Worth checking

[Behemot]

Re: Relay controlled PSU thermistor

OK. I got this AcBel-manufactured Supermicro server PSU fro end of march, 2010. Both rectifiers are indeed in parallel, nothing special here.

There is relay, 12 V controlled. As far as I can tell, it has NO connection to secondary. Primary and secondary part are really nicely separated in this PSU so it could not even have any. Voltage on one pin of the relay is something over 9 V while being turned on.

ADD// Ok, I am getting something. The non-grounded pin of relay is connected through 161ohm resistor to the same potential as is on Vcc pin of CM6800AG. This pin is the supply voltage for the CM6800 controller of some 13,76 V. It means the relay is actually being feed from the same voltage as the CM6800 is. It seems to be using tiny transformer to produce this supply voltage and small circuit from SMD transistors and stuff.

Hope it helps, I think I have depleted my luck with working on the high side for today and I am mounting it all back together

[TELVM]

Re: Relay controlled PSU thermistor

Thanks for your efforts Behemot . Here, token of gratitude .


What do you think that 161Ω resistor in series with the relay coil is for?