[weblacky]

Yo All,
 I've officially started my Indy Nidec PSU rebuild work, I disassembled a unit I know doesn't work but that I've already washed clean (and doesn't need any case work).

I'll be posting images I've taken in our image system in a few days.  But I've finally gotten all electrolytic caps off (one requires you to melt through a lot of adhesive, so more on that trick later).  I need to re-examine my cap collection but I think there is one I have that's missing from this list: https://wiki.preterhuman.net/SGI_Indy#Ca...ly_unit.29

But the rest correctly correlate with my findings, though I'm also tracking manufacturer and family series in order to find the correct upgraded family lines to put in with compatible characteristics and ESR.

But while I need to take a break in the next two days for work-related stuff, I thought I'd post because I ran into an issue identifying parts, and I need the community's help!

While these are parts aren't broken, I wanted to make sure I can find correct replacements for several item.  I've decided NOT to replace the Suppression EMI filter caps because finding new ones with the right footprint has proven very difficult and because the parts measure fine off the board.

So the parts I need help with are (I think), are a varistor (MOV) and an XY Cap?  These are extremely close to the AC line and basically among the first line of defense.

Please see attached pictures.  They MAY have been made by muRata.  While technically they seem to be fine.  I wanted to try to find replacements as they are there for safety and I thought A. perhaps I'll run into a PSU with MOV damage and need to a new MOV and B. Perhaps a new MOV and safety caps would be safer/faster/better?

So anyone you can help ID these, please help!

I've tested nearly all the ICs on the main PCB, I've gotten datasheets for nearly all the ICs (excluding small transistor like items) for the daughter card (where the opto-couplers live!).  So that needs to be tested as well.  I also need to start recording some of these test values for my test system.  

So far, things are going good, there was zero leakage of anything, but I will say, there is ZERO chance of you doing all the caps without a desoldering gun and even then I had to re-solder and add and desolder a few times to penetrate the entire PCB via for caps, transformers, and semiconductors. The vias are smaller than you'd expect on this board.  So most cannot be cleared with simple desoldering braid easily (many attempts would be needed).

However layout is OK, and there are only a few items that aren't accessible unless you desolder the majority of parts (like hidden titanium caps and resistors right under FETS).

I'll be posting more, don't worry.  I want to race ahead so I can place a DigiKey order and work for a few days while it arrives.  So I'm eager to find the answers.  Also, I'm cheating slightly on this PSU as I ran across a small tidbit of advice years ago on common failures (I will repost that, since it's not from my research).  However, so far none of the common components they flagged tested as failed. So this thing might just be Caps-only.  But I need to clear the basics on the daughter board before I make such a claim.


Thanks for any help and I will be continuing this post with more pics, I just don't have the time to prep them now, I will later, promise!

[robespierre]

The precise parts aren't that critical, but you would want to match the values, voltage ratings, and the safety class if relevant.
A component tester should tell you if the capacitor is within its tolerance (90 to 110 pF). Testing varistors requires a megger (insulation tester).
These were the only components of their type at the input? Normally input filtering uses either a potted filter module, or a discrete filter made of an X2 cap across L and N, two Y2 caps across L-G and N-G, an inductor, and sometimes additional X and Y caps. It depends how much filtering the device needed to pass its RFI tests. Surge protection is also often of the three-MOV type, across L and N, L-G, and N-G.
One of the problems with MOVs is their tendency to heat up and explode, so a modern version with an integral fuse would be nice. This MOV is rated for 472V, which happens to be the maximum peak voltage plus a third. That's a good value to keep to if it will ever be used in 240VAC countries.

[weblacky]

What you see is what I have. I know that 101K is a capacitor rating marking?

I see the manually written voltages on them. But I don't have any other info. For for the XY cap, there are two of these in parallel with the ground (I don't see an X and Y cap separately!). For the MOV, there are two in parallel on the PCB as well. From what I can see I have a large (transformer-ish) coil, two different EMI and filtering suppression caps (metalized film in plastic boxes type) and two of each of the above posted parts. I do have a single very small ceramic disc cap buried in there but I haven't established if that is safety or functionality at this time.

It looks like they just put suppression on then use a pair of each of the above parts in parallel with each other and put one end into the live area and one into ground on PCB (soldered to track that attaches to screws in case.

I don't have enough info to find replacements, could you provide any more advice to at least track down the correct safety ratings on replacements? I'm not sure how to track down these safety-rated compliant caps in the like Digikey (jaron/terminology issues, I guess).

Wish I owned an insulation tester (on long wish list), so that's a no...I can't have everything (or so I was told).

[robespierre]

OK, so there are box-type (RIFA?) RFI caps, a choke, and single-layer ceramics, all part of the filter circuit. That's a typical topology (look up "pi filter") that has a sharp response.
When capacitors have one leg on ground, they are being used as Y-class noise suppressors. Any replacement must have a Y2 or Y1 rating. They normally come in pairs (L to ground, N to ground).
If they are "across the line" (one leg on Live and one on Neutral), they are being used as X-class suppressors. A replacement with either an X2 (or X1) or a Y2 (or Y1) rating is safe to use, although the filter is generally designed for best noise performance when X2 is used. It will be a single cap instead of a matched pair.
The filter capacitors before and after the choke can have different properties (as with your box film caps on the inlet side, and ceramic discs after the choke). You can think of it as a three-stage filter that needs to have a very low high-frequency impedance in its first stage (box cap), then the choke acts as a low-pass filter that smooths off medium-frequency noise, then the ceramics of the third stage have wider bandwidth to shunt low-frequency noise. It works in reverse too, to reduce interference being emitted from the PSU into the IEC cable.

Sorry if I wasn't super explicit about the ratings, I thought "101K" was mostly self-explanatory. The digits represent picofarads and work just like color bands on a resistor. So 10 x 10^(1) is 100 pF and K means ±10%. The impedance of these 100 pF Y-capacitors at 60 Hz is the cause of ground leakage current in switch-mode PSU devices (if you have ever plugged a computer into a GFCI socket and wondered why it trips, that is why). Naturally a higher capacitance has a lower impedance, and causes more ground leakage.

If you search in the "single layer ceramic capacitor" category you should see criteria for X2 or Y2 ratings. Also remember to specify the lead spacing if you want components that fit the board.

[Irinikus]

These might help as examples for the components question: ("KC" no longer seem to be a valid safety rating code though!)

Here's a link to the complete PDF: http://j5d2v7d7.stackpathcdn.com/wp-cont...asheet.pdf

[weblacky]

OK, Yes I understood that 101K should be a marking (just didn't look it back up for the purpose of this post).  It seemed to me they chose a XY so they could just do whatever with it?  I'll retrace the tracks, but both are the SAME part in each pair (double-checked that).

Here are some surface pics (one inside case, rest with caps removed).  You can see the blue parts paired up on the extreme sides.  The "boxes" are suppression caps (as you said).

I'm more worried about whatever KC472P is, (assuming it's a MOV) are MOVs just sold with the cut-in voltage number and that's all you need? Both 101K and KC472P Mention 560-2, XY says ~ (I assume this means AC Volts?) for both 400 and 250...so that's also confusing.  These passive markings don't even locate datasheets, so it's frustrating.

Can you give any advice on the basic specs here?

---

These might help as examples for the components question: ("KC" no longer seem to be a valid safety rating code though!)

Here's a link to the complete PDF: http://j5d2v7d7.stackpathcdn.com/wp-cont...asheet.pdf

I saw that, but since I cannot make any determinations from the markings, nor is their marking system still used...it seems unintelligible to me.  I cannot find equivalence.  That's why I was asking for advice on decoding the meaning on these markings.

[Irinikus]

So I would imagine that the KY402K can be used with a 250V AC rating, when used in line to line application, and 300V AC, when used in a line to ground application, and has a capacitance of 4700PF and a tolerance of 10%.

[weblacky]

I'll back off on this for now, not enough info/knowledge on my part to be confident on these protective caps, I'll leave what's there as they are undamaged.  I've gone ahead and ordered the replacement Nichicon and United Chemi-con Caps I found for the electrolytic family lines.

I ordered 3 sets worth.  I still need to probe the logic daughter card for tests on the ICs.  But it's going to take like 4-5 days to get the parts anyway, so that will give me time to do that checking and recording.

For this PSU, so far I've not found any kind of failure/smoking gun.  For all I know it really was just the caps failing?  If I keep checking and don't come up with any additional info...I"ll be doing a basic current-limited bench power-on.  If it Turns on, great, if not, well, I guess we'll start probing a live board.

UPDATE -> Pics : https://siliconimage.irixnet.org/index.p...y-PSU-Work

[weblacky]

OK, DigiKey has FINALLY shipped (wow, they are normally faster than that...). I took the time to do an in-depth check of the daughterboard (as best I could without a functioning comparator). Initially it appeared (during power off testing) that two of my three optocouplers were damaged (one with resistance in LED and other with shorted transistor output). However, upon removal of all of them and retesting they all acted the same out of circuit, so the issue I observed was just in-circuit measuring issues. I also found a single glass diode (not a zener) that appeared to be short, but when I removed a leg and tilted it out for out-of-circuit testing, it tested correctly. So again, just has been in-circuit oddities so far. Nothing has testing suspect once out of circuit.

In terms of assembled circuit confirmation, until I get a unit fully working, I won't have the luxury to compare one PSU (testing point/location) against another.

I'll keep updating as stuff comes in.

[robespierre]

The whole electronics industry is facing massive shortages and delays. Digikey and Mouser have had trouble keeping their warehouses staffed and they also have supplier problems.