300w + 300w = 600w?

[Since87]

Oh, one other issue.

Do you have an explanation for the signal shown below?



It looks to me like that's about 2.5 Ap-p at around 111 Hz. Were you hooked up to a load that draws current like that? (A DC to AC inverter perhaps.) Or, is that the paralleled supplies oscillating?

[Yo-DUH_87]

That is due to the cheap signal generator driving the car amp I was using as a test load (for lack of something better). It has had a problem all along, I'm just too lazy/cheap to fix it

I would have used something else as a test load, but I didn't have anything handy...

It doesn't do that when you use the circuit to drive something that is not the test load (such as a large fan, but that didn't draw enough current to use for testing).

[bigben2k]

Originally posted by Yo-DUH_87
When ben and I were discussing the driver, I was expecting ben to have the extra boost from the zener diode connected to the source of the mosfet. That would have worked out quite nicely (as it did in my circuit). But, alas, ben had to bring up the "small" issue about the zener possibly allowing current to flow through the zener and power on the mosfet when you didn't want it on, so the quick solution was to connect it to ground...

@ben, I'm working on figuring this all out. Hopefuly I'll have some time this afternoon

Thanks, looking forward to it!

[Yo-DUH_87]

Ok, I did the testing, and current does flow backwards through the mosfets, without much power loss

So, it is in need of a slight redesign. I think that I can use the parts ben has, with a few additions (shouldn't have to replace any parts, I don't think). Sorry about that, ben!

Good news it that I don't think it will decrease the efficenty of the circuit any, and should make ben's life safer.

[bigben2k]

Sounds good.

I'm ready to:
-replace the MOSFETs for the 12v lines
and/or
-build a seperate driver for those 12v lines

I still have to order the extra zener, to raise the limit above 16v, so whatever you got, throw it at me



As for the driver current at startup, there might be a way to fix that: I'll investigate further. Worst case, it imposes a delay, so I'd need a clearer explanation of the issue here, because I don't see how a delay has any impact. If it's an overload condition on the driver, then I'd need to address that.

[bigben2k]

I was going over Brian's PWM article, when it hit me, about why a driver is reffered to as being high-side or low-side.

So I found this interesting chip: Maxim's 1614 (PDF, 8 pages). I'll review it later, to see if it could drive the MOSFET of the 12v line.

edit:
High side driver is for when a MOSFET is hooked up between the high voltage (i.e. +12) and the load.

Low side driver is for when the MOSFET is hooked up between the load and ground.

Since the ground is common in a PC for all voltages, it's not possible to setup a low-side MOSFET.

[Yo-DUH_87]

Ben, I am hoping to be able and use the parts you have, with perhaps some slight alterations. I just need to get the design out of my head and onto some transmittable form of communication

[bigben2k]

Originally posted by bigben2k
...
PSU#4 has been chopped up, and will be my prototype, for the wiring fit. I'm pondering wether I should use the remaining wiring to connect the PSUs to the Yo_Duh circuit, or if I should use my 6 gauge copper "wire" instead. There are a lot of wires!
+12 : 5 wires
+5 : 12 wires (including a special one, see below)
+3.3: 5 wires (also including a special one)
GND: 18 wires (!)

The +5 and +3.3 lines include a seperate wire, with the same color, that was wired directly into the mobo header, on the same pin as the regular voltage lines. It looks like they were some kind of additional filtering or voltage stabilizer: they're linked to a cap and resistor.

I guess that this additional stabilizer line will be fed to the Yo_Duh circuit, since I can't run it to the mobo anymore.:shrug:
...

Going over some specs that Yo_Duh posted in another thread:
http://www.formfactors.org/developer...12V_PS_1_1.pdf

I found out / confirmed that those extra lines on +3.3 and +5 are indeed voltage stabilizers. In fact, it's the actual "remote sense" line that the PSU circuitry depends on.

[bigben2k]

Here is Yo_Duh's proposed circuit, to prevent the PSUs from feeding each other, when one of them's off:

In short, it involves adding a 50 Ohm resistor, and using two MOSFETs, back to back, as well as a couple of diodes. Comments?

[Yo-DUH_87]

Told ya I was tired when I smacked that up for you...

The resistor should be on the other side of the zener, like this:

[Yo-DUH_87]

And the resistor can be removed, as long as the zener is able to sink the ammount of current your FET driver produces.

Sorry about that, I should have waited until I was more alert to send that

[Since87]

This design has a MOSFET used 'backwards', and I don't see the point of having a second 'forward' MOSFET in series with it.

I don't see any point to the 50 Ohm resistor. It will slow the speed with which power 'comes up' at power up. But, is there a benefit to that?

Once the 'forward' MOSFET is gotten rid of, the zener protection needs to be from gate to source of the 'backwards' MOSFET.


Edit: punctuation

[Yo-DUH_87]

Originally posted by Since87
This design has a MOSFET used 'backwards', and I don't see the point of having a second 'forward' MOSFET in series with it.

The body diodes block internal flow.

I don't see any point to the 50 Ohm resistor. It will slow the speed with which power 'comes up' at power up. But, is there a benefit to that?

That is just in case you are using a more powerful driver than speced. The resistor can be eliminated if you are using a standard FET driver

Once the 'forward' MOSFET is gotten rid of, the zener protection needs to be from gate to source of the 'backwards' MOSFET.

The "protection" is just keeping the driver from providing too much voltage to the mosfets. No more

[Since87]

Originally posted by Yo-DUH_87
The body diodes block internal flow.

The body diode of the 'backwards' (top) MOSFET blocks reverse current flow.

You still haven't explained the point of the 'forward' MOSFET.

Originally posted by Yo-DUH_87
The "protection" is just keeping the driver from providing too much voltage to the mosfets. No more

Generally the zener is used to prevent transient conditions from exceeding the "Absolute Maximum" spec for Vgs:

VGS Gate-to-Source Voltage ± 20 V

Is the gate to source voltage of the 'backwards' MOSFET adequately clamped?

[bigben2k]

It's a little radical, but I'm thinking of going this route:

-use the MOSFET in its normal "forward" fashion, the way it was meant to.

-In order to prevent any current from feeding back into a PSU that is off (sinceit's not an issue when all PSUs are On), I would add a power diode.

Of course at this point, it's actually a step backwards, but...

-add a relay, to bypass the power diode, where the relay would be powered by each PSU's "Pwr_Good" line. This would be done to get around the voltage drop of the power diode, in its normal forward direction.

The problem I'm afraid, is finding a relay with contacts that have a minimal resistance, and that can handle the potentially high current load.

The relay would otherwise be wired, just like pHaestus did, for wiring a pump to a PSU.

[Yo-DUH_87]

Umm, no offence, but the pwr_good is on all the time, it is just a 5v ttl logic signal saying that "all is fine here now, thank you for asking."

Instead, why not power it from the PSU's 5v line? If you realy want to go that route.

I don't know, but IMO I think that purchacing the extra mosfets would be cheaper than getting all those relays.

Since87, the reasons the 2nd mosfet is there is a. the internal body diode, and b. to prevent reverse current while the gate is saturated.

[bigben2k]

Actually, www.allelectronics.com has an AZ971, available for $1.25 ea., in qty 10. The site says the contacts are rated for 30 amps, but the manufacturer says it's rated for 40:
http://www.azettler.com/pdfs/az970.pdf
(PDF, 2 pages)

The problem is that they don't give out any contact resistance, except for a "less than 100 milliOhm, initially", under what they call a "24V, 1A voltage drop method".

The "PWR_GOOD" line isn't always on: it only turns on when all output voltages are OK. What I'm not so sure about, is the power off sequence.

[Since87]

Originally posted by Yo-DUH_87
Since87, the reasons the 2nd mosfet is there is a. the internal body diode, and b. to prevent reverse current while the gate is saturated.

a. Which MOSFET is the "2nd mosfet"? When you say, the reason is "the internal body diode", are you referring to the internal body diode of the second MOSFET or the first?

b. If one MOSFET's gate is saturated, then both will be saturated. There is nothing preventing reverse current flow when both MOSFET's are 'on'.

[Since87]

Originally posted by bigben2k
It's a little radical, but I'm thinking of going this route:

-use the MOSFET in its normal "forward" fashion, the way it was meant to.

I think you had it right when you were talking about using a single MOSFET 'reversed'. (Although the gate to source protection Zener should go between the gate and umm... source.)

I really don't think the MOSFET will vary enough from its specs to matter, when used that way.

[bigben2k]

Originally posted by Since87
a. Which MOSFET is the "2nd mosfet"? When you say, the reason is "the internal body diode", are you referring to the internal body diode of the second MOSFET or the first?

b. If one MOSFET's gate is saturated, then both will be saturated. There is nothing preventing reverse current flow when both MOSFET's are 'on'.

Right, but if the driver is operating from the PWR_Good line, it should shut off, when the PSU is OFF, and with the additional diode, it should prevent the MOSFETs from seeing any voltage. With both MOSFETs, there shouldn't be any leaks back into the PSU that's off, because the internal diode of the backward MOSFET.

The problem I see is the potentially unpredictable behavior of the MOSFET, when used backwards. And I still need a better driver for the +12 line.


But back to what you're saying: when the gates are saturaded, what prevents current from leaking back?

Given a 4 milliohm resistance, I have to assume that the voltage drop across a single MOSFET would be around 0.16 volts (@ 40 amps) and 4 millivolts (@ 1 amp).

The MOSFETs "thermally stabilize", as that's the balancing principle here, combined with the trim pots to fine tune the voltage differences (That's the way I understand it :shrug: ).

I don't know!

[bigben2k]

Originally posted by Since87
I think you had it right when you were talking about using a single MOSFET 'reversed'. (Although the gate to source protection Zener should go between the gate and umm... source.)

I really don't think the MOSFET will vary enough from its specs to matter, when used that way.

Ok, I think it's time for me to run some tests on this.

I'll try assembling one circuit, put a dummy load on it, and measure some voltages and currents (where I can).

[bigben2k]

Going over the SSI specs, the maximum delay between the PWR_ON line commanding a PSU to turn off, and the PWR_OK (previously referred to as PWR_GOOD) dropping, is max 50 millisecond.

So the relay would have to respond within that time frame.

[edit] Specs for the above relay indicate a response time ("Release time") in the order of 5 millisecond, so I think this will work.

[bigben2k]

I ordered a technical CD from http://www.vicorpower.com/ , to help me figure out, hopefully, some of what's used now, in the PSU industry.

I should start putting together the first circuit shortly, for some tests.

[bigben2k]

This might be a foolish idea... What if I took the 12V and 5V regulating lines from the PSU, and hooked them up together, passed the MOSFET?

I'll be pondering this today.

[Since87]

Originally posted by bigben2k
This might be a foolish idea... What if I took the 12V and 5V regulating lines from the PSU, and hooked them up together, passed the MOSFET?

I'll be pondering this today.

I'd written a fairly long response to this earlier, but due to some BS that Hotmail does, I lost it.

I assume you mean "What if I hook the output sense lines up to the common output power bus of the system?"

I see three big problems with this.

The first and biggest, is whether the PSU's will start up correctly with this kind of setup. The first PSU to come up will be pulling the sense lines of the other PSU's up. Nasty things might happen because of that. I wouldn't even guess what is likely to happen without seeing a schematic of the PSU's.

The second problem would be, that with all of the sense lines tied together at the output bus, all of the PSU's might be trying to regulate that point to a slightly different value. It's quite possible that oscillation would occur as different PSU's alternated outputting more or less power trying to regulate the output to their individual setpoints. You could use an RC so that at high frequency each PSU is sensing its own output and at low frequency each PSU is sensing the output at common bus. This would eliminate the possibility of oscillation at the cost of degrading the systems response to load transients. Added capacitance on the output bus could make up for the degraded transient response, but it would be somewhat costly.

Finally, sensing at the output bus pretty much defeats the load sharing capabilities of the setup. The sense line effectively makes the MOSFET resistance a nonfactor, but you were relying on the MOSFET resistance to ensure sharing of the load.