advertisement


4 terminal capacitors

Yes your post #39 selection is OK. Just make sure the lead spacing is correct /matches the originals.

The blue ones are 220nF same as 0.22uF 100 volt caps. Do not think they make (polycarbonate) MKC caps anymore so if you want to change them look at Vishay MKT or EPCOS MKT that fit.
 
Posts crossed while I was working that out, many thanks for that. :)
May I ask a (another) dumb question please?
I can't see any indication of -ve on the PP / PET caps - how do you know which way around to mount?
 
Unlikely to be capacitors at all.... and very unlikely to be those Wima's. The symptoms are of a difficult fault to fix and you will need to be extremely lucky to have a happy ending here I'm afraid.....
 
Jez, I was surprised too, but this is my thinking.
As unlikely as it is that the WIMAs are the fault, at least they are only pennies to replace and more important the board has to come out and Jeddy can inspect the other side really carefully and check previous repairs and re-flow all the joints. If it does not fix it, he has at least eliminated some parts of it.

And I am guilty of exonerating a specific capacitor in the past, because it is 'X' make only to have it take out another set of output devices...
 
snowman_al said:
Jez, I was surprised too, but this is my thinking.
As unlikely as it is that the WIMAs are the fault, at least they are only pennies to replace and more important the board has to come out and Jeddy can inspect the other side really carefully and check previous repairs and re-flow all the joints. If it does not fix it, he has at least eliminated some parts of it.

And I am guilty of exonerating a specific capacitor in the past, because it is 'X' make only to have it take out another set of output devices...
Click to expand...

Fairy nough.... Dry joints vastly more likely. Yes we do see a normally very reliable component fail now and then but some are so reliable that even doing this for a living you see only one failure every few years... and film caps not under HT stress would be one example... I would expect to see 500 dry joints and 100 faulty semiconductors for each faulty film cap over a few years of repairs....

This sounds to me like either "something daft" (that could still be non obvious to find) or a twat of a fault that would give even you or I a sleepless night...
 
Interesting reading, thanks guys.
Firstly I'm thinking along the lines of snoman, I can replace a couple of caps for minimal cost and they are the only lead I have.
I'm guessing here so please shout me down or correct me, but here's a little more information which may or may not be relevant. The comment was made that the B3 / B4 on the PP caps was irrelevant, I'm guessing that if it doesn't relate to the specs then it may relate to the manufacture, in which case I'm curious as to why one was B3 (the bad one on the right channel) and one B4 (the not so bad on the left channel). I bought the amp second hand from a used hifi dealer 20+ years ago, probably not that old at the time which makes me wonder about it's history, maybe it had been returned faulty and repaired? Maybe the original B3 cap was replaced with a B4? Maybe it was a batch of dodgy B3 caps? It's also curious that spraying the B3 with freezer spray makes a lot of noise and the B4 also a bit but not a lot. One thing the guy who I originally took it to for repair did point out was that the fuse had been bypassed! That's corrected now, but it does make me wonder... I also have a white lead from the transformer which has been cut going no-where - I doubt that's normal. Again seems odd. Makes me wonder. Having taken the board off the heat sink I can see there is a tiny component (2mm sq?) nearly behind the above caps, no idea what it is, I don't have a magnifying glass to hand. No way I could replace it anyway, but I guess it is that which could have been affected by the freezer spray.
The amp was also stored for a few years in a loft (yeah, I know).
I guess it's pretty difficult to detect a dry joint in circuit and who knows, maybe replacing the caps will fix the problem because the problem is a dry joint on one of them - but for the minimal cost that would suit me.
Looking at the above flickr link / album you can see the entire PCB and the circuit is pretty trivial and the green caps on the left are pretty much on their own, maybe 1.5cm from anything else (above tiny component excepted), so I'm not exactly freezing anything else. There really aren't many components and I've replaced all the rubycon caps already. Some resistors (I'd guess very reliable), some diodes? I've replaced 4 of the mosfets attached to the heat sink but not the 4 on show.
Given the doubt I'll just pick up replacement green caps from RS just down the road who I see also do them on Tues and see where that takes me. I'm not sure there's much point in replacing caps which don't react to freezer spray yet.
Given the minimal number of components on the board and I've already replaced some I can't help think that unless the PCB is at fault (no idea how likely) it would still be a lot cheaper (and more rewarding) to replace all the components (mabe 50?) than spend £hundreds on a new amp which is anything like as good, even if the fault is a dry joint on one of the components, but yes, maybe re-flowing them comes before that.
 
Well you have to start somewhere...
Jeddy, The B4 or B3 stamp is a red herring, honestly.
I assume you 'froze' all the components on the board individually and not just the capacitors? Please confirm.
Also I maybe should have mentioned a couple of layers of cling film over the components prevents moisture on the circuit, sorry a bit late now.

When you have the board out, please post some pictures of the 'back' side, so we can look too. (Nothing on the net for ROK S1 that I can find). The 2mm square thing could be very significant.

''Given the minimal number of components on the board and I've already replaced some I can't help think that unless the PCB is at fault (no idea how likely) it would still be a lot cheaper (and more rewarding) to replace all the components (mabe 50?) than spend £hundreds on a new amp which is anything like as good, even if the fault is a dry joint on one of the components, but yes, maybe re-flowing them comes before that.'' Well that is one way!
 
Just been listening to your recording of the noise... again. That is quite a thump at the start, then crackles that fade.

I wonder if it is something much more simple - slight arcing as the mains makes a good connection caused by either a 'dirty' mains switch contact or faulty 'kettle' lead or connection.
If you are in a position to try, first check the mains input pins on the amp. Look for arcing. Swap the mains lead for another one anyway.
Not sure if it is practical, but Carefully temporarily link out the mains switch. Or if not practical - leave the mains switch on and use the wall switch to switch off once quiet. Then after the usual delay (over night whatever) switch on at the plug switch. Does it do it then I wonder?
 
Thanks snowman.
B3/B4 fair enough - just grasping for explanations.
Before freezing I did watch a few youtube vids as I wasn't 100% sure what I was doing / how long to freeze for and I basically copied one of those. I sprayed the entire board (noisy), once warmed up again I then started spraying sections of the board to find which section caused the noise, then indivdual components within the offending section. The guy on youtube didn't use clingfilm so neither did I - oh well. I'm not sure how much that helps, any time you freeze anything you'll get condensation from the atmosphere on it anyway, the spray is supposed to not leave any residue itself.
I've added a couple of pics of the back of the board to https://flic.kr/s/aHsmmRqW8t one of the entire board and one of the back of those red 330 caps - in the centre there are a couple of tiny blue components which the photo (camera has better eyes than I do) reveals has 1003 on it, the back of the red caps is just to the right, as before, aside the red caps and the blue 1003 there really is nothing else at that point on the board.

The mains input pins are spotless - look new. I did actually do what you suggested on the above recording, the power switch was on and I turned it on at the mains - the click at the start is the wall switch. I'm going to bottle out of bypassing the mains switch, but thanks for the suggestion. I can however swap the kettle lead.
Just a thought without meaning to, in the words of Alan Sugar 'keep banging on about it' - if the big Roksan caps were faulty could they be not correctly supressing the initial mains surge? I'll believe they aren't causing the crackle, but maybe this is a another problem?

Not sure if I had a brainwave in the bathroom this a.m.. I'm not sure if I can, but if I swap the two red 330 caps around and see what happens.
  1. The problem moves channel = the cap is the problem
  2. The problem remains on the channel = the caps is fine, something else is wrong on the channel
  3. The problem goes away = possible dry joint etc..
I'm not sure if the terminal leads are going to be long enough for me to do that with my amateur soldering, I can't see why not.

I know wima caps are supposed to have a good reputation, but while googling for other information I noticed comments they had a 'fakes' problem too! I'm sure out of Roksan they were fine, but if it has been repaired...

Thanks!
 
Just been having another look at the photos (I've realised I can see better what is going on there than I can with my eyes) and this isn't looking promising. I've uploaded another photo to the album (2018-7-29 11_18_19) around the bottom centre pointed to by the cursor is some 'corrosion' on the circuit. Not sure if that's fixable / bypassable.
 
Jeddy said:
Just been having another look at the photos (I've realised I can see better what is going on there than I can with my eyes) and this isn't looking promising. I've uploaded another photo to the album (2018-7-29 11_18_19) around the bottom centre pointed to by the cursor is some 'corrosion' on the circuit. Not sure if that's fixable / bypassable.
Click to expand...
Just scrape of the green lacquer with a Stanley knife or scalpel Blade, nice and gently, trying to keep as much of the original Copper Track intact.

It must be spotlessly clean. Then use some Bare Copper Solid Wire and Solder over the tracks, touching the solder pads of each component to get a good connection.

I've had this issue with lots of Amps over the years and its usually the dielectric leaking out of one of the caps (I presume you replaced the component likely to have caused the moisture on the tracks) that eats the tracks away.
Be patient and at least that will be another issue out of the way and hopefully closer to fixing the problem.
 
Thanks. Yeah, I've replaced all the leaking rubycon caps (i.e. all of them as pretty much all of them had leaked!). The terminal lead to the left of the cursor is actually one of the replacements. Any particular guage / mm2 copper wire (the track is approx 1.5mm)? Are we talking normal solid copper wire as per domestic circuits (not the braided stuff), I probably have some of that if that's suitable?
 
A single stripped 1.5mm Twin and earth conductor will be fine and handle more current than the track was able to.
 
First the blue 2mm square marked 1003 is a 100K Ohm surface mounted resistor. (1003 = 100 with three zeros, hence 100,000 or 100k Ohms.) It is in parallel with the green 330pF cap. Could be broken or not soldered correctly. That is where the tooth pick is useful.

The corrosion might not of eaten to far into the track. As joley says, scrape the green masking off very carefully and see what is still there. Difficult to see from the pic, but it might be just surface corrosion. In that case polish it and tin it with solder right up to the green mask. If it has eaten the track away - then bridge it with something like a lead from a resistor or capacitor. Again cover the exposed copper with solder.

The picture is not easy to see, but there are what look like 'snail trails' at various places. I think you need to really clean the board of all the residues on there. Certainly all the leaky cap goop. Cotton buds and alcohol (or acetone as long as you keep it away from the components.)
 
That surface mounted resistor could be a problem to detect if it had a problem, it's on the back of the PCB where the heat sink is (the heat sink is basically parallel to the entirety of the PCB and pretty close to it), trying to poke even a cocktail stick down at it with the heat sink attached it will be a challenge.
I'll attack the green lacquer next.
I previously cleaned the PCB with WD40 fast drying contact cleaner which removed the leaked electrolyte nicely, although you have put a slight doubt in my mind as to whether I cleaned the back as well as the front (I think I did as I seem to recall the cotton buds getting caught on the back of the components). No harm in doing again anyway.
As best I understand the circuit (hmmm) at least this is 'before' the mosfets attached to the heat sink, so current is minimal, it's quality I'm after here.
Thanks.
 
Another couple of photos uploaded as above.
It looks like I've got lucky, the track appears to be fine under the green lacquer as per photo. I'm not even sure it needs a coating of solder.
I've also given the back of the board another(?) clean, given the colour of the cotton buds maybe I didn't do it last time.
I suspect I may now swap the caps over but need to give my eyes a rest first (they don't like doing close up stuff like this any more, I used to be quite short sighted which was great for close up work, but the joys of aging mean I can't see close up with or without glasses any more).
Thanks.
 
Track is fine. If you do not want to solder try some nail varnish over the copper. Any colour is fine!

Please trim the leads off the caps closer. The 'snail trails' do not show so much on this picture, but I have marked up some of the 'dirty' areas. Most likely rosin and nothing, but have a look. Some of the joints are a bit 'blobby' if you do not mind me saying. Make sure any where two joints are close together there is nothing 'bridging' the gap.

You are correct, I have to take pictures of some components now to read what they are! Easier than the magnifying glass!

2hoahwy.jpg
 
You must log in or register to reply here.


advertisement


Back
Top