Full A4000 With Ceramic Capacitors + Technique
 

Been a while.

I've finally finished off my A4000 Rev B with full ceramic capacitors using a polyimide heater and a 450W PC PSU.


http://thumbnails109.imagebam.com/35...6355685189.jpg http://thumbnails111.imagebam.com/35...8355685217.jpg http://thumbnails112.imagebam.com/35...8355685255.jpg http://thumbnails111.imagebam.com/35...d355685327.jpg http://thumbnails110.imagebam.com/35...9355685374.jpg http://thumbnails109.imagebam.com/35...2355685402.jpg http://thumbnails110.imagebam.com/35...d355685427.jpg

Polyimide heater won't be cheap, I bought 3 different ones as a lot for a total of $650 USD but I think because they now have the artwork for the heater you would be able to get them for far less.

The copper plate is 3mm thick 35cm x 25cm with the heater stuck to the bottom using the provided adhesive (it's already on the heater when its shipped) and you just have to make sure it's flat and there's no bubbles in the heater.

When you first turn the heater on it draws around 12 amps of current before settling down to around 10 when it gets up to temperature. My first experiments with the heater in open air could only yield around 80 degrees C, but by putting it into the pine frame and putting a layer of alfoil beneath it I was able to get it to 120 degrees.

The wooden surround is good because it gives you somewhere to rest your arm when you are soldering the capacitors in place. You must ensure that you have cleaned up the pads as well as possible or you will have a hell of a time trying to solder the new caps down keeping in mind that your hand is hovering above a board at 120 degrees C!

The small screws holding the stand offs in place under the copper plate won't cause an issue with the board because the component legs / SMD parts keep it above them anyway.

The heater takes about 40 minutes to come up to temperature and it doesn't cause any of the plastic to melt so you don't have to remove anything before you start which is nice.

C181 is made using two 47uF capacitors stacked on top of one another

I'm going finish my A3640 in the same way along with one more A4000 board and then nothing else. If you have any questions feel free to ask via the usual email.

zero242@internode.on.net

It looks great!

It would be a very good solution for the capacitors aging problem. They won't be a problem anymore :)

Congratulations!

No problem to change any polarized capacitor with non-polarized (ceramic)?
It has testing for long time ? just asking..

How much did you spend for your home-made pre-heater?

I remember buying a pretty good one from China for something like £60... With ceramic plates :)

Ceramic capacitors use ceramic material sandwiched between metal plates for capacitance and are polarity neutral. Electrolyitic capacitors use chemicals that depend on having the right voltage polarity applied to work. There is a thread on EAB dealing with capacitors:

http://eab.abime.net/showthread.php?t=74671

There are some potential issues in replacing the A4000 electrolytics with ceramics. There are actually lower capacitance ceramic capacitors in parallel with the electrolytics across the power bus on many locations on the A4000 MB. I don't know the exact reason for this but my guess would be to surpress fast transient voltage spikes while the bigger electrolytics keep the voltage stable.

I choose to replace the electrolytic capacitors on my A4000 with the same ones. I made sure I got high quality Japanese capacitors and not the cheaper ones from China. Commodore got some really bad quality ones when they made the A4000 and those are the ones that leak badly after 20 years. On my latest A4000, it was only the 22uf capacitors. the 47uf and 4.7uf capacitors were all fine. I replaced them anyway just because they were so old.

There are some minor gotchas with ceramic caps:
1. ESR is much lower than electrolytic, so some early regulators designed before large value ceramic caps were popular would not like it. Thankfully, Amiga is very old tech, so there isn't a lot of regulators to worry about.

2. Ceramic cap isn't as good for audio as it has piezoelectric effect. i.e. it could work like a mic and pick up vibration. Use good quality long life electrolytic caps etc. Also dielectric losses is bad leading to distortion. (Tantalum caps are not suitable for audio circuits.) Decoupling is fine, just not for filters/AC coupling. On the other hand, Amiga audio isn't that high quality...

3. Watch out for the different dielectric types. There are a lot of trade offs to make them high density and when you try to use them in real life, you'll find that their capacitance is much lower than you expected due to temperature, DC bias.

For high values ones, use X7R, X5R, avoid Y5V or any dielectric types that promises a large capacitance in small packages until you read their characteristic curves under temperature and DC bias vs your actual usage.

Derate the voltage rating for the cap as their capacitance is dependent on DC bias. Always read the datasheet! This is true even for the X7R/X5R dielectrics.

4. Avoid very large packages (> 1210) as you can have mechanical issues with thermal/board flexing. There are hand soldering issue with thermal stress that can lead to cracks etc.

I'll try to answer everything.

@Anemos: I replaced all the caps on my A4000 board which has become little more than a test bed for anything I want to do to it. The problem I had before was that I would be inclined to leave it off for a considerable time and of the course the capacitors didn't like that. I'll do some more extensive testing when I have finished off the last of the Amiga gear I still have here to repair.

@Jbenam: Being the creative sort I wanted to do things by myself, so I designed and got the polyimide heaters made up with my specifications, with all the design work, shipping etc. it cost $650 USD for 3 heaters, two are very similar 25 x 35 cm and the last one is much smaller. Now that the company has the artwork I could inquire as to how much it would be to get more made up if people are interested.

@Ami GFX: The smaller ceramics are to catch and filter out specific frequencies. I just got tired of replacing caps with more that I knew would fail and cause more damage to the board, in my opinion this is the best solution we have currently and I've read many replies and articles about for and against.

@K.C. Lee: All the capacitors I used are 1210 packages with the 100uF 6.3V replaced with two 47uF caps in parallel. As I am aware of the dangers of thermal shock to ceramics I specifically used the preheater to get the board up to temperature before I started soldering and allowed it to soak. The trick is to make sure the pads are absolutely clean so that when you start soldering it makes it so much easier, I could also recommend that glue could be used if you have a fine enough hand to hold them in place so they don't wick onto the pad when you solder it at one end. It's vital to avoid flexing the board once they are soldered in place and I believe that with the way the A4000 has the daughterboard at 90 degrees that will limit the flexing if you change cards because it's screwed in place. Heat is only moderate on an A4000 so you shouldn't get too much flex on the PCB from temperature variation.

I'll take some screen shots of traces from the audio whilst I tap around with a paint brush or tooth brush to see how microphonic the board is and post the shots here.

Parts used are: 4.7uF X7R 16V, 10uF X5R 25V, 22uF X5R 16V, 47uF X6S 16V

This ceramic solution isn't a magic bullet and requires plenty of preparation, patience and time but in the current circumstances it's certainly better than replacing bad capacitors with more bad capacitors.

Actually, I don't think that the electrolytics are that bad. In Amigas, the failed caps are mostly from bad stock, not because the caps are inherently bad. I've yet to see a failed cap on an A2000. Your preheater recap is interesting as an experiment but not very cost effective for just keeping an A4000 running. Even if it has to be recapped every 20 years or so with new electrolytics, that is not really that expensive or difficult to do.

I work on a lot of old analog gear and one of my current projects is an old Roland synth. It is about 20 years older than the A4000. These synths use electrolytics to keep the control voltage stable on the VCOs and VCFs and VCAs. Out of a dozen or so, only one has completely failed after all these years, in which it was kept in storage for the most part and not turned on at all.

@ Loedown, Although I have my reservations with change polarized to non polarized caps, i wish going all fine.. ;)

I read something interesting here:
http://www.dslreports.com/forum/r269...rtion-Analysis
The spectrum plot is very telling on the harmonic distortion by ceramic caps.

>The waveform with the ceramic cap became almost triangular with a notch at zero crossing.
That sounds like (pun intended) dielectric absorption.

BTW if soldering down 1210 is an issue, then recapping or SMT soldering isn't for that person. :P

There's a world of difference between aluminium surface mount caps and through hole electrolytics, put the two side by side and I know which will fail first.

The only capacitors that have any bearing on the audio path are C433 and C443 apart from the other two 22uF (in / out). Considering the Amiga's grittiness anyway I doubt whether most people would notice the distortion. If people are that worried then solder 22uF NP caps in place of the ceramics.I believe without looking at the schematics that the filter in the audio is using ceramics as standard.

Given that electrolytic capacitors are one of the components most likely to fail over time, the failure rate on both is pretty low. On four A4000s I've had over the last 5 years, the grand total of leaked SMD caps has been 3. Failed caps that just stop working are harder to detect because, in most cases, the Amiga works fine with a missing power supply cap or two. I used an A3640 for years with the three caps with the wrong polarity completely removed.

Commodore hardware is fairly forgiving which is a blessing and a curse (consider timing issues with accelerators).

My main consideration for this project was not so much for the leaking of the capacitors but the damage they do to the board and the effort required to clean it up and get it working again. The battery will do far more damage than the caps but the caps near the audio stage do tend to make a mess when left, more a consideration of that area being the most heavily populated in the smallest space. By the time people even notice a problem it's usually after the electrolyte has eaten away some tracks under or near LF347.

My day job doesn't leave me much time any more but I will post some CRO shots of the audio from the newly capped boards for people to see and compare.

Why do we try to be smarter that people who designed it? There are many reasons to use electrolityc caps instead of other types and they work fine. Otherwise why even bother? A4000 works fine without most of these caps anyway.

I / we are not trying to be smarter than the designers but instead working on a technique to help preserve the boards with current technology, many pieces of equipment get an upgrade as technology changes, take for instance the Vampire 600 project or Jens Schoenfield's work or Ian Stedman's work.

People can do whatever they want with their hardware, this is just an option.

I agree. If there is new knowledge we should take advantadge of it. It has passed more than 20 years from the Amiga era and lots of new things have been improved and discovered and we'll see lots more...and it is necessarily due to try and fail in some cases.

Besides...not always the designers were 100% eficients (A3640 for example, DMA issues,... for sure due a lack of developing time...but there were some mistakes done).

We'll see what the results after a hard testing of this caps replaced motherboard. Is it possible Loedown?

Anything is possible, time permitting isn't always as forgiving.

http://eab.abime.net/showthread.php?t=76376

Edit: the link to a Amibay's thread posted is also here in EAB.

This talks about the sound capacitors....

http://elm-chan.org/docs/cap/ccap.html

Japanese but worth to read (use translator) - pros and cons for typical high capacitance ceramic capacitors.

(IMHO biggest problem is loss of capacity under DC load/bias)