Failed filter inductor? (Antec SP300)

[eccerr0r]

I had neglected repairing the fan after its started whining in my Antec Smartpower 300, and one day it got really quiet... oh good, it fixed it self...

nope... oh no, it SEIZED!

Eventually the power supply and computer got really, really hot. Eventually something fried and the machine had an unexpected shutdown

After some debug I found that the secondary side filter inductor (the big toroid with multiple windings) had shorted out. While the heat due to the fan failing being very suspect, has anyone seen this failure before? I had initially thought (and there were) bad capacitors that caused the failure but ended up being something totally different...

[goodpsusearch]

Re: Failed filter inductor? (Antec SP300)

It is quite common after the fan seizes the toroid gets too hot and cooks itself.

You can replace it with a toroid from a dead psu. I have done it before with success. Just be sure that the new toroid coil is compatible with the one it is going to replace.

[PeteS in CA]

Re: Failed filter inductor? (Antec SP300)

The inductor shorted out? In what sense? A dozen turns of #20 or #18 wire will have very low resistance. Did you have a shorted turn in a winding? Did you have a short from one winding to another? From a winding to the core? One other problem that can happen is that powdered iron cores can deteriorate with heat, causing the inductance to drop and possibly damaging O/P capacitors. If that happened, the core is ruined: do not strip off the wire and try to salvage the core. One other tip. The two most commonO/P inductor core materials are Micrometals -26 (yellow body with one side white) and -52 (green body with one side blue) material. The two materials will give the same inductance with the same number of turn. -52 is better with higher switching frequencies, but can be used with at lower frequencies, and will probably run a little cooler. If your P/S uses bipolar transistors its switch frequency is in the lower range, under 50KHz. If your P/S uses MOSFETs the switch frequency is probably higher, and -52 is the proper choice (-26 will probably work, but it will be hotter, and its life probably shorter).

[eccerr0r]

Re: Failed filter inductor? (Antec SP300)

Yes, I figure the heat may have melted the varnish coatings on the wire and shorted with the other windings on the core (and maybe to the core itself), but I was surprised that coil was first to succumb to the heat versus, say, the transformers or semiconductors. The resistance through it shouldn't be that high though all the secondary current goes through it, I'd thought the semiconductors would pass the same current but higher voltage drop and thus more heat dissipated.

Since this toroid is just used as a LPF versus a transformer, I just replaced it with one that I had whose size was about the same and had similar gauge wire. I had not considered the core materials but yes, I would hope I wouldn't lose too much due to losses which would necessitate replacing it again (ouch. Maybe I should point an IR thermometer at it just for curiosity sake). The "new" inductor had an extra winding I should get rid of, but I was lazy and just heat shrunk some insulation over the ends.

The Antec SP300 is a UC384x based supply and uses a single MOSFET to pull the primary. I've never measured the switching frequency but suspect it's close to the reference designs...

[goodpsusearch]

Re: Failed filter inductor? (Antec SP300)

The inductor shorted out? In what sense? A dozen turns of #20 or #18 wire will have very low resistance. Did you have a shorted turn in a winding? Did you have a short from one winding to another? From a winding to the core? One other problem that can happen is that powdered iron cores can deteriorate with heat, causing the inductance to drop and possibly damaging O/P capacitors. If that happened, the core is ruined: do not strip off the wire and try to salvage the core. One other tip. The two most commonO/P inductor core materials are Micrometals -26 (yellow body with one side white) and -52 (green body with one side blue) material. The two materials will give the same inductance with the same number of turn. -52 is better with higher switching frequencies, but can be used with at lower frequencies, and will probably run a little cooler. If your P/S uses bipolar transistors its switch frequency is in the lower range, under 50KHz. If your P/S uses MOSFETs the switch frequency is probably higher, and -52 is the proper choice (-26 will probably work, but it will be hotter, and its life probably shorter).

This is very informative. Thank you very much

Now, I know why that Konig psu I got some time ago had its toroid coil burnt although the fan has not stopped working. It used mosfets and -26 material.

Thank you again! Bookmarked this post.

[PeteS in CA]

Re: Failed filter inductor? (Antec SP300)

... I was surprised that coil was first to succumb to the heat versus, say, the transformers or semiconductors. The resistance through it shouldn't be that high though all the secondary current goes through it, I'd thought the semiconductors would pass the same current but higher voltage drop and thus more heat dissipated.

The (I)(R) loss of the wire is not the only loss (= heat) in an inductor. Also very relevant is the "core loss", basically the core material's resistance to magnetic field changes. -26 and -52 materials have the same permeability and A(L) (uH per turn ^2). But their core loss vs. frequency is different, -26 having a higher core loss than -52, especially at frequencies greater than ~40Khz (I'm not looking at the relevant graphs, but those can be found on Micrometals' website). Then there's where the inductors are, in relationship to other hot components and airflow, and how much of the core is covered by wire. A relative full core, mounted near the rectifier heatsink, with airflow blocked by capacitors and/or bundles of wires (does that sound like a typical PC P/S?) can get really hot. For -26 and -52 materials, Micrometals recommends, IIRC, keeping the core temperature under 110C or 120C. The deterioration with excessive temperatures takes time - months or even a couple of years - but with a good heatsink, fan and airflow, a rectifier will be able to tolerate a 110C-120C case temperature (I'm not saying it's great, just that it's tolerable).

Please don't misunderstand my point. Micrometals -26 material is excellent when properly used. It's also a bit lower in cost than -52 material. -26 has been an industry standard material since at least the mid-late 1970s, when switch devices were bipolar transistors and switch frequencies were typically in the 20KHz-30KHz range. -52 material was developed by Micrometals (in the 1980s?) when they saw that MOSFETs were going to make higher switch frequencies practical. At a cost a bit higher than -52 material Micrometals has even newer core materials rated for use at 200C.

Now, I know why that Konig psu I got some time ago had its toroid coil burnt although the fan has not stopped working. It used mosfets and -26 material.

If it was running at 100KHz, had hot components nearby, had lots of wire blocking airflow to the core, and poor airflow, a reasonable possibility.

[eccerr0r]

Re: Failed filter inductor? (Antec SP300)

Interesting... now is the IR loss or core loss more significant would you say?
Assuming that the "correct" material is used based on the switching frequency (like, don't use laminated steel for 40KHz ) would IR loss be a larger or smaller factor than core losses?

I was designing and building a buck supply that was pretty much a failure (but learned a great deal from it) - one of the things that I found is that at high load, the inductor indeed got quite hot. But my switching frequency was only 3KHz (don't ask...) and I was using any nillydilly core for the inductor... This may have been why?

(I initially passed this off as IR loss because I just reused the windings on it, which were fairly thin (maybe 22-gauge or so) passing 3 amps. Then again the reason why my circuit was such a failure was because of the 3KHz switchiing frequency -- I needed a *much* larger inductor... (also the fact at the time it got really hot, the PWM was pegged at 100% duty cycle... so basically the inductor is just one big resistor...))

[japlytic]

Re: Failed filter inductor? (Antec SP300)

It is quite common after the fan seizes the toroid gets too hot and cooks itself.

Should power supplies have overtemperature protection in case the fan fails? I think so.

[goodpsusearch]

Re: Failed filter inductor? (Antec SP300)

Should power supplies have overtemperature protection in case the fan fails? I think so.

The good ones should have.

[Behemot]

Re: Failed filter inductor? (Antec SP300)

The good ones have.

[larrymoencurly]

Re: Failed filter inductor? (Antec SP300)

Should power supplies have overtemperature protection in case the fan fails? I think so.

I once asked Ultra about that, and they said theirs didn't (made by Wintech) but that at around 55C the power output would start to decrease enough that there wouldn't be a fire.

Does anyone know what the second thermistor in an old Antec SmartPower does? It's not the fan control thermistor.

[goodpsusearch]

Re: Failed filter inductor? (Antec SP300)

Does anyone know what the second thermistor in an old Antec SmartPower does? It's not the fan control thermistor.

Overtemperature protection.

[PeteS in CA]

Re: Failed filter inductor? (Antec SP300)

Interesting... now is the IR loss or core loss more significant would you say?

The intricacies of inductor design are well over my head. The designer has to pick a large enough core, of the right material for the chosen switch frequency, with enough turns of heavy enough wire.

In your example, the inductor operating at 3KHz, with 22AWG at 3A ... I would say that 22AWG is marginal at best for 3A DC. With AC, another factor comes into play. As frequency increases, current flow tends to concentrate in the outer area of the conductor. This is called "skin effect". The facilities maintenance people where I work - they handle power wiring, among other things - have told me that they have to use larger wire for 400Hz at a certain current than they do for 60Hz at the same current. So at 3KHz AC, 22AWG is definitely too small for 3A. As for the core, well to pick the right size and material you need to know the magnetic field strength and core loss vs. frequency. One can pick the best material from charts one can find on the vendor's website, but it's the magnetic field strength part that gets way over my head.

In your case, I'm going to guess (and it is a pure guess) that the wire loss was the greater factor in how hot it got. Looking at Micrometal's comparison chart, the material performance difference between -26 and -52 (most commonly used in P/Ss) at 3KHz isn't that great. OTOH (speculating), the core size may have been on the small side.

Speaking hypothetically, I'd try ~25KHz with a T157-26 core with enough turns of 18AWG to get ~100uH and see what happened. Realistically, you're stuck with whatever cores you can find, and I don't know how realistic 100uH is. The convenient thing about Micrometals' P/N coding is that the P/N describes the part. A T157-26 core is: a toroid; 1.57 inches outside diameter; made of -26 material. So you can measure the O.D. of whatever you have on hand and figure out what it is from Micrometals' website.

[eccerr0r]

Re: Failed filter inductor? (Antec SP300)

Here we go. The dead coil from my Antec SP300 (shorted between windings). Nice fried glue on it.

The toroid I was using for my 3KHz experimental/homemade switcher looks like it has a good 6 to 10 ft of wire of which all of it needs to deal with IR loss. At 22 gauge I think that's way too long for 3A, so likely it's I*R loss. I wish I could bump up the switching frequency but there's only so much that can be done with the microcontroller... :o

[PeteS in CA]

Re: Failed filter inductor? (Antec SP300)

Yeppers, the glue is well done. Three or four windings, depending on whether the two heaviest windings (gold colored varnish in your pic) are connected to separate circuits or connected in parallel for a single high current O/P. If you want to check whether the core is OK, you can: unwind it; wind 4 or 8 turns using one of the lengths of wire you removed; get the Micrometals datasheet for -26 material, determine what core it is based in the OD; calculate what the inductance should be from the datasheet A(L) (nH/T^2); measure what the actual inductance is. I forget what the A(L) tolerance is, but it should be on the datasheet and is probably +/-20% or even +/-30%.

If you're in a mood to experiment, you can use a 555 timer to make either a fixed duty-cycle or PWMed inverter, with a much higher switch frequency (using a CMOS version might be a good idea).