A600 Capacitor Replacement

[Loedown]

Originally Posted by Retrofan

Which are then the especifications I must search for C234 and C334? I have a 1D1 mobo, but I suposse that for the audio section they must be the same for all the A1200.

C324, C334

22uF 16V

[Retrofan]

But as Cosmos has told me, the originals are also polarized.... so then I won't change them again.

[Loedown]

Originally Posted by Retrofan

But as Cosmos has told me, the originals are also polarized.... so then I won't change them again.

The choice is yours, Commodore screwed up the design, it's a design flaw. Using polarized parts is ok but they will wear out and fail far sooner than non polarized parts would. Electronics is a funny industry, things can fail without the slightest provocation.

[Retrofan]

Originally Posted by Cosmos

Yes, but SMT non-polarised are hard to find...

First search (Google search, 0,09 seconds):
http://www.hbelects.com/Capacitor/El...Ratial-12N-099

Are this ones?

Or this:
http://nl.mouser.com/search/ProductD...wCD8u6nQ%3D%3D

[Eclipse]

Originally Posted by Retrofan

First search (Google search, 0,09 seconds):
http://www.hbelects.com/Capacitor/El...Ratial-12N-099

Are this ones?

Or this:
http://nl.mouser.com/search/ProductD...wCD8u6nQ%3D%3D

The first link, correct description but wrong picture. Usually a generic picture is used so I suspect it's correct. (If you want to use normal radial and not SMD)

Second are ceramic caps, so I'd say no.

You're going to have a hard time finding 22uf 16v(or higher) electrolytic non-polarised.
My personal advice is unless someone on here has any or knows where to get them cheap, stick to using a low profile radial.

[8bitbubsy]

Originally Posted by Eclipse

You're going to have a hard time finding 22uf 16v(or higher) electrolytic non-polarised.
My personal advice is unless someone on here has any or knows where to get them cheap, stick to using a low profile radial.

http://futurlec.com/CapElectroNonPolar.shtml


22uF 100V Non-Polarised Crossover Capacitor $1.10
(sadly only 100V available, thus it is a huge cap...)

Shipping worldwide is only $5 for batches under $30 on futurlec. No costs for the picking job etc!
These capacitors are huge, but with some trickery you can easily fit them. Be creative, I've installed two of these in my A1200 and A600.

[rkauer]

Also remember the audio sections swings from -12V to +12V (that plus the tolerance: ~10%) means 24 x 110% = 26,4V.

25V parts are out of the question: way too tight on voltage tolerance. Next available (theoretically) voltage for capacitors is 35V.

Nothing stop the user to put higher voltage capacitors instead, but remember the bigger the voltage of the capacitor is, the bigger the capacitor is for the same capacitance value.

[Loedown]

Originally Posted by rkauer

Also remember the audio sections swings from -12V to +12V (that plus the tolerance: ~10%) means 24 x 110% = 26,4V.

25V parts are out of the question: way too tight on voltage tolerance. Next available (theoretically) voltage for capacitors is 35V.

Nothing stop the user to put higher voltage capacitors instead, but remember the bigger the voltage of the capacitor is, the bigger the capacitor is for the same capacitance value.

The voltage swing is + / - 12V if the amp is rail to rail, I checked, it isn't.

The voltages are in respect to gnd, so 16wv should be more than adequate, based on further consideration.

When I recap my A4000 or I recap someone elses I'll take some pictures, I'll be honest Bubsy, your solution is practical but aesthetically displeasing, why would you use an axial part when radials are more common and easier to place?

Originally Posted by Eclipse

You're going to have a hard time finding 22uf 16v(or higher) electrolytic non-polarised.
My personal advice is unless someone on here has any or knows where to get them cheap, stick to using a low profile radial.

If you guys are a little more patient I will buy a reel of them through Nichicon and ship them off wherever for a small fee.

[Retrofan]

I found this article:

Amiga Audio Circuit Capacitor Repair

by James Jackson
This article contains graphic content dealing with the disassembly and repair of the Amiga 4000 and may be considered too graphic by some readers.
Or put differently: This article requires the Amiga 4000 to be opened to make changes. I make no guarantees of success nor can I be held responsible for any damage incurred while implementing the changes described herein.
Please be sure to read through this document in its' entirety before attempting any repairs.
Legacy Capacitor Problems


In my last article, (Amiga A3640 CPU Board Repair), I discussed the problem of the reversed electrolytic capacitors on the Amiga 4000 A3640 Daughterboard. Many of you are probably heaving a big sigh of relief. You own Amiga 500 or Amiga 2000 computers and do not have to worry about this problem since you do not have the CPU Daughterboards in your computers.
Well, before you get too smug, let me mention that while investigating the reversed capacitor problem, I stumbled onto another capacitor issue that has existed for a long time. It exists on the Amiga 4000, the Amiga 500, Amiga 2000 - and these are just the computers that I had schematics for. I'm not a betting person, but if I were, I'd wager that this issue is on the Amiga 3000, 1200, 600 and probably goes back to the grandmother of them all, yes - the Amiga 1000.

Just what is this problem? Is it serious? Can it be corrected?
The problem lies in the audio circuitry, which all Amiga computers possess, and is on the motherboard. (See Figure 1.)
It is not serious, and yes, it can be corrected.
First, an explanation of how the audio output circuitry works.
The audio output amplifier is connected to a bipolar voltage source. That is, it is connected to +12V and -12V. When an amplifier is connected this way, its' output can swing from a positive potential through zero volts to a negative potential.
Normal connections on the output of an amplifier like this are as shown in the figures. It is either a direct connection to the output, or through a non-polarized capacitor.
If an amplifier is connected to a uni-polar supply, it is connected to a source like +12V and Ground. Then the output of this amplifier would swing from Ground to near +12V on the output. With an amplifier like this, we can use a capacitor on the output. The capacitor could be a polarized capacitor as the voltage potential on the capacitors' positive lead would always be positive.
What evidently happened at Commodore was that someone did not realize the affects of designing a polarized capacitor into a circuit like this. The first designer made the mistake, and every other designer after that just copied the circuit which propagated the error. Or, if it was the same designer responsible for all of the Amiga designs, he did not bother to check his previous design.
What was needed was a capacitor coupled output on the output of the Amiga. This is okay. However, the way that it was implemented in the Amiga could use some improvement.
Will this capacitor incorrectly designed into the Amiga damage the computer? Well, the answer to that is - Yes - and - No. You see, electrically, there will probably not be much difference between the polarized and non-polarized capacitors. This is the 'No' part of the answer.
In the Amiga 4000, they are using surface mount (SMT) components, including SMT capacitors. For some reason, I believe that these capacitors may have been stressed during the soldering phase of assembly. Either that, or this particular value of capacitor in the SMT version is a dud. In either case, on my Amiga 4000, the left channel capacitor had leaked its' electrolyte out onto the printed circuit board (PCB), and corrosion had set in on a few of the surrounding components. This is the 'Yes' part of the answer.

I have removed the defective SMT capacitors, cleaned the area surrounding the capacitors, and replaced them with a through hole variety non-polarized capacitor.
I recommend that those of you with at least an Amiga 600, 1200, 3000 or 4000 check your own machines. These are computers that Commodore converted over to SMT production, and are most likely to exhibit this problem.
If you choose to ignore the potential problem, here is what could happen. As the capacitors leak their electrolyte onto the PCB, the electrolyte flows out onto the board covering any circuitry and component pads that it comes into contact with. Over a period of time, the fluid, being corrosive, begins to corrode the things that it has come into contact with. It's just like automotive battery acid, on a smaller, but no less dangerous, scale. If left alone long enough, eventually the corrosive electrolyte will eat through the copper traces, vias, and even integrated circuit (IC) leads. The result will be a major repair job.
The way to correct the design error that exists in our Amiga computers (short of redesigning the audio circuitry), is to use a non-polarized capacitor. When choosing a non-polarized capacitor, use the same value of 22uF, and be sure that the voltage rating is high enough to withstand the largest output from the audio amplifier. I chose a 35 volt capacitor (See Figure 2).
A non-polarized capacitor would cost a bit more and many times a more costly item will not be used in the design - so that a few pennies can be saved.

[Loedown]

What people are forgetting is that the amplifier can only swing into either positive or negative but not both at the same time. The LF353 is not a rail to rail amplifier meaning it cannot have an output of + / - 12V, the datasheet I saw quoted minimum of + / - 12 V with Vcc of + / - 15 V and typical 13.5 V @ 15 V. 12 V is one fifth lower, so the values of 9.6 V minimum to 10.8 V typical, with some consideration for variances in chips, power supplies on the Amiga etc.

[Eclipse]

@8bitbubsy,
Sorry if I confused anyone.
I meant SMD 22uf non-polarissed are difficult to find, normal radials are still fairly easy to get hold of.

[jaycee]

Cor, only just spotted the audio capacitor problem! Paula outputs a signal using its 5V supply with a 2.5V bias. This bias is removed first by the first two opamp halves using "AVREF" that is 2.5v from a potential divider. Then the signal is amplified by the second halves and the filter implemented as a Sallen-Key filter in this stage. Because the opamps are powered from +12/-12V the output capacitors should indeed be bipolar units.


16V rated parts are adequate as the LF357 does not swing rail to rail. Normal radial through-hole parts can be soldered to the pads. In the event that bipolar capacitors cant be obtained, you can make them - simply use 2x47uF capacitors in series with the negative pins tied together.

[8bitbubsy]

Originally Posted by jaycee

[...] In the event that bipolar capacitors cant be obtained, you can make them - simply use 2x47uF capacitors in series with the negative pins tied together.

Is this just the same as having a bi-polar cap? Except that it makes 23,5µF instead of 22µF?

[jaycee]

Yep it's the same. In fact, it is exactly how nonpolarised capacitors are constructed internally.

[Stedy]

Two questions to ask yourselves

1) Does the Amiga audio output go negative?
2) Do you really need 35V parts for a 1V audio signal?

I can answer question #2, no you do not need 35V parts. Any 6.3V or above part should do as line level audio is only 1V. You are primarily concerned with the voltage across the capacitor, 1V + any DC offset which will be small, allowing for s short circuit to ground.

To answer question #1 it would be necessary to measure the outputs of the LF357/TL084 op-amps with an oscilloscope. I suspect the signal will be +/- 0.5V around ground.

As an aside, the main killer of electrolytics is non-use in properly design equipment. There is a MIL-STD that covers this but they need to be powered periodically to retain their dielectric properties. Will be re-capping an A500 soon as it's caps have failed from lack of use. Unit worked fine when purchased 6 years ago, has been powered 3-4 times since then and now will not boot.

A supplier I use, ESR (www.esr.co.uk) list some 22uF non-polarised electrolytics, order code 776-226 and cost £0.91 each but they are 100V parts!

Found this on Ebay, good price, right physical size.
http://cgi.ebay.co.uk/22UF-16V-NON-P...item1c11675cd5

Ian

[jaycee]

1) Yes, it is an AC output with no DC bias. Check my description of the output circuit in post #72. The reason for the capacitor coupling is to remove any residual DC offset present on the output of the LF357 opamp, and also to provide some isolation - this is pretty standard on audio outputs.

2) It is probably only 1V RMS output, but still best to use a 16V part. 35V is massive overkill.

I noticed on the A1200 schematic, the 22uF output caps are actually labelled "NP" but the symbol shows a polar capacitor. This is probably simply that they used the right footprint in Mentor but the matching symbol assumed a polar cap. In the A600, they omitted to add "NP" so I guess they got manufactured with polarised capacitors.

I'd probably use CPC for the parts (they're the small-arms part of Farnell) and IIRC they stock Panasonic capacitors, however I dont think they have nonpolars either. I'd probably use the 2 capacitor method if i couldn't be bothered to get proper NP caps.

[8bitbubsy]

Originally Posted by Stedy

Found this on Ebay, good price, right physical size.
http://cgi.ebay.co.uk/22UF-16V-NON-P...item1c11675cd5

Ian

Smooth!
I just bought a pack of those. Time to make my A500 and A2000 get the non-polar caps as well, without the big cap mess.

[Eclipse]

Thanks Ian, I've decided to use them in place of the 2 conventional parts on the A500 and on the A600+A1200 instead of SMT parts.

Just a tip though, pre-form them properly. Don't be tempted to just splay the legs out. I'll upload a picture of what I mean later.

[e5frog]

Originally Posted by Eclipse

Just a tip though, pre-form them properly. Don't be tempted to just splay the legs out. I'll upload a picture of what I mean later.

Still waiting.

[kipper2k]

Originally Posted by e5frog

Still waiting.

lol... he did say later