Coolmax V-500 quality power supply

[kaboom]

Re: Coolmax V-500 quality power supply

I removed the pc board from the case and have been tracing the circuit. The primary winding of that transformer connects to the pwm chip U3843. Each of the two secondary windings of that transformer connects to a FET. Unless the transformer shorts (which is possible) the transformer provides isolation between the pwm chip and the MOSFET switching transistors.

Shall we argue some more!

Okay, they're driving both MOSFETs with that GDT. It's also possible to drive the low side one directly and just use a GDT for the highside, which is what I assumed.

Of course, the former saves the 384x if the MOSFETs explode; the latter most certainly does not.

If properly designed and insulated, as in double-layer-insulated gate windings, the GDT won't allow a direct connection between its windings. That insulation looks like a heavy orange-teflon jacket over the magnet wire.

[kaboom]

Re: Coolmax V-500 quality power supply

This is why.....This one was made in October, 2013. The label claimed 450W. Only one cap per rail, even the 5VSB!!! First time I've ever seen that.

It won't be the last time you see it, either. Probably 600-800 mV ripple on 5VSB with a 1A load. Which is also what I got from a "cheep tablet" supply with it's one output cap- at 1A also.

450mW, with the ripple in spec.

You're worried about the 5VAC SB?? That's the least of the trouble.

+3.3 will become 5 or so when that MOSFET in the linear regulator shorts- see the nice heatsink?

No seperate output inductor and mag-amp regulator choke, so they're getting 3.3 from 5, in "giant LM317" fashion. All supplies that do so have much hotter secondary heatsinks (or whichever heatsink has the pass transistor)- doing so with those "snap-able" things-that-look-like-heatsinks is playing Russian roulette with the MB's chipset and RAM regulators.

Then there are minor things, like a 10A fast acting fuse, instead of a more sensible 4 or 5A slo blow. Zero input filtering = zero radio reception for 50-100 feet of any power wires feeding the thing, and acting as antennas.

All 20AWG output leads, and questionable caps overall. Looks like the same brittle PCB material that POWMAX used to use.

The only things missing are "diodes on a bracket" for +12.

[everell]

Re: Coolmax V-500 quality power supply

Kaboom, thanks for that informative discussion on wire color codes. I have been checking those IMicros more carefully and find that some have black on hot, white on neutral. Others have white on hot and black on neutral. Even worse, some are switching power on and off on the neutral side!

[kaboom]

Re: Coolmax V-500 quality power supply

Kaboom, thanks for that informative discussion on wire color codes. I have been checking those IMicros more carefully and find that some have black on hot, white on neutral. Others have white on hot and black on neutral. Even worse, some are switching power on and off on the neutral side!

I figured- often, cheep supplies with black and white wires are wired wrong.

Be careful with those. If you get a short from line (the actual hot, not the "reversed neutral") to GND, like a bad/fake Y cap, you'll be in for a surprise- perhaps along the line of my forum name.

And with the switch wired wrong, it can happen without you even knowing.

There were some power supplies sold by radioshack that were recalled due to fires caused by this. Some "utility" SMPSes do not connect, or bond, the negative output terminal to the case or EGC, apparently for controlling ground loops and/or interference when used with CB/HAM equipment.

I'm not sure what actually failed, possibly the transformers were poorly insulated, perhaps there was a hotside to coldside Y cap that failed.

But the biggest factor was the lack of secondary bonding to the EGC. Remember, line and neutral were reversed, so the fuse was useless (in the noodle) and the actual fault path was from line to the secondary low-side/common. It wasn't a ground since it wasn't bonded to ground! At least in a safe place.

What happened was these supplies would break down, causing the entire "12V" output to become elevated to line voltage. The supply, taken by itself, when this happened, appeared perfecly safe. The metal case was bonded to the EGC, but, again, not the DC common. Both binding posts were insulated from the chassis.

What would happen when either a HAM/CB radio with outdoor antenna, or "car audio" amp, with its connections to other grounded devices, was connected?

Let's start with the HF radio and its antenna's metal mast. Yes, dipoles can be used, with both elements insulated from ground, but I am trying to keep this concise. Current from the AC line goes to ground thru the antenna mast, coming from the power supply's fault, via the "hot" DC common, into the radio and out/up the coax shield.

Some radios float their chassis via ceramic caps, to avoid ground loops in automotive installations, but they will not withstand 120V AC, and fail the same way a bad/fake Y-cap does- by smoking and burning.

In those radios that directly connect their chassis to circuit common, this provides a low-impedance path from the "hot" DC common, directly to the coax shield. Please note, only this node of the resulting "circuit" is low-impedance; overall, it's still too high to result in current sufficient to trip the breaker. Now the DC common, radio chassis, and coax shield are live, with between 5-15A, depending on soil conditions, passing over the coax. The actual voltage-to-ground, as measured from the floating common, will vary depending on total path impedance.

This will heat the coax up, and any concealed runs become/became fire hazards.

Now take the audio amp supplied from such a supply. The same breakdown occurs, except this time, the "fault return" path is into the car amp, out and up the shield of the audio cable, hopefully into a grounded mixer. Either the audio cable, or the isolating/loop breaking resistors inside the amp burn up. A real shock hazard exists, if the mixer is not grounded, as consumer audio equipment tends to be.

If the shield of that audio cables burns open, then the front end opamps of the amp fry and die, since they then become a fault path. Floating DC common->opamp supply common->backfeed thru opamp, fault feeds out of the input pins->current travels up the still-intact signal conductor(s). This also kills the mixer outputs- its output opamps end up breaking down too.

This is why there's a big deal about how and where the ground/EGC is terminated. It must be able to hold during a fault, even to the point of instantaneously tripping the branch ckt bkr!

The EGC is more important than where the fuse is- while we don't want to fuse the neutral, the presence or lack of the EGC is a game changer. Take the above example, with a fault current below the rating of the fuse. That supply would still sit there and shock. Now, the fuse would blow if external paths were returning the fault to ground, possibly even saving the cables if the fuse were small enough.

By bonding both the chassis of the supply, as well as the DC common (turning it into a DC ground) to the EGC, any faults inside that supply return via its power cords green wire, keeping most of the current away from the external connections of devices powered by it.



Now your unit doesn't even have Y-caps, but let's say it did and that "ground" wire was hanging on by a solder ball or wire strand. And the underrated ceramic "Y caps" short or go leaky. The ground burns clear. At this point, two things happen- the fault clears through other return paths, like up the VGA/DVI/HDMI/USB, or audio cable(s); or the entire case sits there at 120V, like during a "box only" memtest or burn-in where no monitor or other grounded external device is connected.

Even if external bonded/grounded devices are connected, the small signal circuits could be damaged by either induced voltage (shield to signal wires within cable) or ground potential rise- we're not talking about a few amps of normal return current here- it takes upwards of 80A to instantaneously trip a typical 15A ckt breaker.

If you want to fool with those Imicros, you can switch the wires around, fusing the hot. Now there should be some kind of boss for a ground terminal- you can use that instead of going thru the board for the EGC. You might have to drill a hole and use machine hardware if the threads don't hold- the original holes should take the smaller case screws- the ones for floppy and optical drives.

Be sure to clean it well, and use either 2-26 or dielectric grease to keep corrosion from affecting your ground. Also, if you've got large fuses in them, put a 4 or 5A slo blo in. The whole "10A fuse no matter what" is a leftover from the no-thermistor and "diodes on a bracket" days.

That was bad- back in '98 or '99, the PS fan in the my first "real" computer became noisy. I thought "Gee, I'll order one from tigerdirect." Hehe... this gutless supply had no AC line thermistor, diodes on a bracket for +12, who knew what kind of caps, and the fan didn't appear to be well-oiled, putting right back where I started. Did I say no thermistor? Plugging it in with the switch on gave quite a snap as those caps charged up! After 5 or so "hotplugs," the end of its plug looked like one from an old vacuum after many hotplugs.

Since this wasn't any good for the power switch (AT system), I just used this for playing with the fan- how to take one apart and oil it. I took the fan from the original supply, oiled it, and put it and the supply back to work. I still had that old supply up until late 2007 when the motherboard got too cranky- a silly Amptron with questionable specs and better avoided... wasn't worth recapping.

But that fan's still around.

[momaka]

Re: Coolmax V-500 quality power supply

This is why.....This one was made in October, 2013. The label claimed 450W. Only one cap per rail, even the 5VSB!!! First time I've ever seen that.
https://www.badcaps.net/forum/attach...1&d=1425951558

Never trust a piece of work from the devil
His helper, the green phantom known as "Logysis" is no good either.

[everell]

Re: Coolmax V-500 quality power supply

Here are the circuit boards from two recent Coolmax V-400 power supplies. I took the wiring harness off and both heat sinks off so that we could see the difference in the circuit design. First and very impressive is that they have moved away from the old TL494/KA7500 design and moved to the SD6109 pwm chip. Next, the two transistor 5vsb design has given way to an integrated circuit UTC7608D, No line filter, and on one of the boards there are no jumper wires to wire around the line filter. This is because the circuit traces on the bottom of the pc board short out the filter! Obviously they never intended to have a primary filter. Four small diodes rather than a bridge, I could not make out the markings. The Main filter capacitors were Canicon 470 uF/200 volts measuring only about 70% of marked capacitance. The switching transistors were the same ole 13007. The output diodes were two S20C40C and a SFR12S20T. The output filter capacitors were (one per rail) AsiaX 1000Mfd/16 volts (12 volt rail, 5 volt rail, 3.3 volt rail, and 5vsb rail). And that line up will run 400 watts?????

The top board had a heat problem. The fan was stuck! Notice that the color of the yellow painted group regulation toroid core has turned a light brown from the heat. The glue between it and the 3.3 volt coil has turned black. Also the white sleeving on the main switching transformer has turned brown from the heat.

My guess is that it might hold 200 watts. So what do you think?

[stj]

Re: Coolmax V-500 quality power supply

i'm sure the missing filter is because it has a passive PFC choke.
i'v got a few psu's like that - the filtering is moved to the back of the power inlet.

[c_hegge]

Re: Coolmax V-500 quality power supply

It will probably manage 250W without blowing up, but with no PI coils and only one cap per rail, probably nothing in spec.

[everell]

Re: Coolmax V-500 quality power supply

At least the pwm chip they are using, the SD6109, is a big improvement. Here is a data sheet on the SD6109.