Help with PSU (Temp control fan, load bank, & PWM circuit)

[ronv]

What you might try is 2 of those TVS diodes in parallel. Normally you can't parallel diodes, but with the TVS I think you could. Might just blow 2 of them or maybe it would be enough to keep them from popping.

 

[ronv]

Wow, can we trade messages!
Thanks for the heads up on the mail. I guess I have to delete them.

 

[()blivion]

I think power supplies are rated in watts. ()blivion probably knows for sure.

They are rated in watts. But what they will and won't actually do is entirely dependent on the exact unit you have and it's topology.

Most generic PC PSU's nowadays are voltage doubled offline half bridge buck topology devices. They use current sense regulation with comparators or small SCR's for the over current shutdown. They some times use quad comps for over/under voltage shutdown and the power good signal. They have a shared inductor/transformer for all rails + an output toroidal inductor. They do not employ synchronous rectification. They use a pulse skipping technique for very light loads, and typical PWM for regular loads. The TL494 and it's clones are about the most common controller IC's found. The regulation is almost always only sensing one single rail, making every rail follow that one through it's different load conditions. This is important information because it's quite normal for loading of one rail to trip the under voltage of another without taking the first rail out of spec. Or loading one rail without loading another to trip the over voltage. This can confuse some who are not aware of this quirk, and lead to misdiagnosis of the problem.

*HERE* is a link to a small collection of typical PSU schematics.

From what I have seen most all these type of PSU's will succeed in putting out approximately the same voltage regardless of the current. In other words, the over current shutdown is separate and doesn't care about the voltage, even if that exceeds rated wattage. But once you start to reach the max wattage the voltage will in fact sag by quite a bit, presumably because there is only so much energy available. Also note they will break down if you exceed the max watts regardless of how you exceed it. It is almost always the offline power switch that dies from such things. Short circuiting of the mains power to the low voltage side of the system is exceedingly rare. One in a million. Extreme voltage spikes that don't completely destroy the device are also very rare. The inductor/transformer usually saturates causing the switch to melt from quick massive over current long before the output ever sees much more than 1.5x it's rated output voltage.

And again, for most PSU's the nameplate information should be in the fiction section of the library :/

 

[jocanon]

Wow, you really know PSUs. I think the high end server supplies are fairly accurate regarding the power rating. I imagine consumer grade PSUs in typical desktop computers would probably not live up to their specs though.

 

[()blivion]

I think the high end server supplies are fairly accurate regarding the power rating.

Probably.

 

[Dusey52]

Just thought I'd chime in with a little info about these power supplies.

- Single high power 12 volt output rail.
- Low power -12 volt output.
- Low power 5 volt standby output (always on when A/C is applied).
- They have a current share input/output for connecting units in parallel for load sharing.
- They have remote voltage sense inputs (+S & -S) for point of load voltage regulation.

The sense lines are connected internally to +12 and DC ground through a voltage divider. If they're connected to the main output lines as intended, the voltage out will match the rated output (what's on the name plate) of 12.15 volts. If the sense lines are left open, the voltage will go up to about 12.5 volts. A lot of the R/C guys like to 'over volt' them for the reasons jocanon outlined, so they leave them open or even bias them to drive the voltage even higher.

 

[jocanon]

Thanks Dusey52 for chiming in. Good to hear from over here!

 

[ronv]

I still don't understand the overvolting thing. It seems like with 24 volts you can only charge a 5 cell battery --- 4.2 volts each X 5 = 21 volts needed. Current is limited only by the battery type and the charger,
So if the supply is rated at 575 watts to can draw 47.44 amps at 12.12 volts or 46 amps at 12.5 volts. Seems like you would want the higher amerage not the higher voltage. What am I missing? Or does the charger need more than 3 volts overhead to work well?

 

[jocanon]

Hmm, Dusey52 may need to chime in again, but what I know is that to get the max out of the PL8 you need higher voltage. I am charging 6 4 cell lipos in parallel, so I belive that is basically the same as charging one large 24 cell lipo battery.

 

[Dusey52]

For 2,3,4 or even 5 cell batteries 24 volts is usually good enough (depending on the charger). But when you move up to the bigger packs, 6, 8 or 10 cells you start running up against the input current limitations of the charger and the only wany to get around that is to up the voltage.

The typical charger we're talking about has a built-in buck-boost regulator. When charging a 6 cell or larger battery at 30-40 amps the input currents can get quite large, which is why people with the these batteries & chargers are switching from 12V to 24V (or more) power supplies to power them. All of these devices have input & output limits on voltage, current & watts. So bumping up the voltage in some cases means you get full power out of your charger vs not getting full power out if you don't. It may seem trivial but a lot of folks want every last watt out of their charger as it reduces the time to charge and increases the time to fly.

Dean

 

[ronv]

Ahh, Didn't know they were buck boost. I can understand if it might run out of soup trying to charge 6 or more.

 

[jocanon]

Put the 28v TVS in and it doesn't get warm at all like the 26V one did, so I think it's good with the 28V.

 

[()blivion]

Yeah working with tolerance and parts that are so close to each other is a PITA. I think 50 Volt transistors would have been a better fit for this project. But it's much too late to change it now.

 

[ronv]

Good. That makes me fell better!

 

[()blivion]

Good. That makes me fell better!

NO NO NO. Sorry. Things are 100% workable with the parts you selected. I'm just obnoxiously picky. IDK, anyway....

I forgot what parts are being used, are they avalanche rated MOSFETS? If so then doesn't that mean they would intrinsically survive the odd over voltage or two? Certainly would not survive the total load currents that we are looking at or a short to mains, but like... back EMF spikes and such should be fine right?

Spikes are all the TVS's are going to save us from anyway. They would instantly ablate at mains power levels, (as will the rest of the system.)

In any case, Going with the TVS's... we should put a bunch in parallel and put a slow blow fuse up stream of that. This way if the PSU gets evil and shorts mains to the 12V lines, the TVS's will trip short, this will almost instantly blow the fuse and open the whole circuit. *KINDA* convoluted... but opening the input "automagically" during catastrophic over voltage is a great safety feature. And this is a very problem free way of going about it IMO. Should also save TVS's in exchange for destroying 50+ amp fuses instead.

Good idea?
-()b

 

[ronv]

No, you didn't hurt my feelings. I'm just glad the voltage (with ripple) isn't above the 28 volt one.
I suspect the supply overcurrent will save the load box as long as it fails under load. Don't know if the supply has overvoltage protection but it would take a ton of TVS to blow a 50 amp fuse. Maybe a crowbar circuit?

 

[jocanon]

It does have OVP. It shuts off at 13.82v.

 

[jocanon]

After testing things out, I think I should put the thermistor for the temperature control in the PSU on the yellow transformer (circled in the picture below). There are a few main reasons for this:

1. It is much easier to attach. If I put it on the heat sink I have to glue it on with epoxy since tape just does not want to stick to the heat sink but electrical tape or any strong kind of tape sticks well to the yellow wrapping around the transistor. I want a quick and easy way to install the circuit since I will be selling these to other people and want it to be as user friendly for them as possible. It will make it a lot more convient if they do not have to mix epoxy and mess with all that, just quickly tape it on and they are done. Since it is enclosed the tape won't be moving around at all and I think it should hold over time.

2. It is more responsive to changes in load. The heat sink works well too, but it is a little bit slower in changing temperature when load increases/decreases whereas the transformer revs up pretty quickly in temperature once the load increases and decreases pretty quickly in temperature once the load decreases. Since the fan will always be on at min speed, I don't think the fan slowing down too quickly will cause any problems and the min speed of the fan should be enough to cover any cooling needs if they just let it idle for a long time with no current.

Please let me know if you agree with or dissagree with any of the above.

 

[ronv]

The transformer might be better as the heatsink might be at a high voltage.

 

[jocanon]

You are right. The heat sink is at high voltage, don't ask me how I know