Lookie lookie – An (almost) all-ceramic capacitor A1200

[amigasith]

Cheers, fellow miggy lovers! I learned so much in this forum while getting "on board" with Amiga again that I decided it's about time to give something back – here are my humble experiences with the much-loved A1200 capacitors. I did not find any similar experience report somewhere in the net and so I really hope that this is of some use for anybody. What did I do? I replaced (almost) all of the capacitors on my miggy with ceramic ones and guess what – it works like a charm! But first things first...

Here is my board (Rev 2B) before the operation:


I read so much about Commodore's poor capacitor choices that I decided to check the caps on my A1200 also. While most of them looked very nice (there was no electrolyte spill yet either), one of them (C239) had already dark and dull solder joints. Unfortunately, my multimeter cannot measure capacitances, but I tried a very basic method (found on the net somewhere) to check whether C239 still had some capacitance left: I measured its resistance. Why does this work? Well, while measuring resistance, the multimeter uses a very low current to do its job and thereby charges the capacitor. When measuring the resistance of a capacitor, this leads to a resistance of close to 0 at the beginning of the measurement and then it goes up to (theoretically) infinity, in practice simply very high resistance values. But C239 always had resistance 0... Of course, I observed polarity and just to make sure I tried with other caps and they all behaved like expected (close to 0 at the beginning and very high Ohm values after a couple of seconds). So I knew that C239 was already done. I really hoped to get around replacing the caps, but after this little experiment I knew that I had to do something about it.

Here is the section with C239 before the operation:


My general idea was: If I really have to undergo this "surgery", then I want a solution that "lasts forever" or at least close to. I really didn't want to do all this again in a couple of years. And so I sat down and read a lot about the different types of capacitors. While ceramic ones seem to have the best characteristics in general (with the most important aspect being that they cannot leak), they are definitely pretty different from the electrolytic ones used on the A1200 boards. In addition, I read Stedy's comment (http://eab.abime.net/showpost.php?p=828328) about higher inrush currents for ceramics, and so I was really unsure whether my idea of using ceramic caps would work. I really searched the net a long time but couldn't find anybody who reported that he successfully replaced all the caps with ceramic ones. However, I gained some confidence while watching a Youtube video here: [ Show youtube player ]. Beware, it's in German but the guy does a real nice job there. Nevertheless, he doesn't show a working miggy either after the surgery...

Soo - I replaced the caps one by one, checking my A1200 after each new cap. Although I only did a quick test (booting into Workbench), the whole procedure took me quite long (2.5 hrs), but it was well worth it. Otherwise, if I had replaced all caps in one shot (like the guy in the video did) just to find out that it doesn't work… Would have been not so cool The only ones that I did not replace with ceramics are the through-hole capacitors. For those, I used very nice electrolytic ones from Panasonic (3000 hrs at 105 degrees Celsius). Those should also do it for a while. While the twist-and-turn technique in the video looks quite evil, it works out very nicely in practice. I tried to find some other electronics board to practice, but couldn't find one with SMD electrolytic caps. So my first attempt was for real and to my surprise it was really easy. Just don't use any force. I thought about it for a while but I did not want to use hot air, since this too can do quite some damage to the surroundings (plastics, etc.).

Here is the section with C239 after the operation:


After the surgery, I ran a whole bunch of tests with an ACA-1231/42 and my Indivision AGA Mk2 plugged in (to provoke potential timing issues and the like). I chose 1024x768 as the Workbench resolution and my miggy did not exhibit any problems even after a couple of hours! Of course, I had it turned on during the last days also from time to time, but still no random freezes / crashes etc. So I think I can safely say that everything worked out nicely. My board is a Rev 2B, just to repeat. Since I always like to be prepared as much as I can, I bought a spare A1200 board before the surgery - just in case. In the meantime, I recapped this board also and it too works like a charm. This is a 1D1 by the way. Again, I plugged the ACA-1231/42 and my Indivision AGA Mk2 in for my tests and everything worked smoothly without any freezes / crashes and the like.

Here is my board after the operation:


And a close-up of C822 as an example:


In case you want to perform your own experiments, here is a "shopping list" for Rev 2B boards:

Ceramics:
2 x 10 uF (http://www.reichelt.de/index.html?ARTICLE=107456)
5 x 22 uF (http://www.reichelt.de/index.html?ARTICLE=107457)
2 x 47uF (http://www.reichelt.de/index.html?ARTICLE=107458)
4 x 100uF (http://www.reichelt.de/index.html?ARTICLE=107459)

Electrolytic:
2 x 470 uF (http://www.reichelt.de/index.html?ARTICLE=84668)
2 x 1.000 uF (http://www.reichelt.de/index.html?ARTICLE=84694)

For the Rev 1D1 board, you need 2 more 10 uF, but the rest of the list is the same.

Phew, that was a long entry. Hope somebody made it down here And of course, I am really interested what people think about this! Looking forward to your posts.

[DDNI]

Nice job!

[Retrofan]

Hi!.

Originally Posted by amigasith

Looking forward to your posts.

... I see you have a 2B board and you are using an Indi MKII. I can see also you've made timing fixes. Two questions:

- Have you made also the ferrite beads fix? http://eab.abime.net/showpost.php?p=...postcount=2065

- Have you got any problem after running for a while HighGFX?

Thanks.

[prowler]

What DDNI said , well done and thanks for the comprehensive instructions too!

[amigasith]

@DDNI

Thanks

@Retrofan:

- I only applied Stedy's fixes from here: http://www.ianstedman.co.uk/Amiga/am..._mobo_fix.html but skipped the extra wire links. So in total, I only removed E123C and E125C, changed R118 to 220 Ohms and added this 470 Ohm resistor as described. No ferrite beds added as replacements for the resistors, although my board also lacks the capacitors (E121C and E122C) described in the ferrite bed fix. I also noticed that my ACA1231 is not running as hot compared to before the fix - I guess this is due to one of the resistors...

- No problems... HighGFX is running nicely. I used it during my stability tests. But I like SuperPlus better - to me, it simply looks nicer than 1024x768.

@ prowler

Thanks to u also

[amigasith]

@prowler

Thanks for merging my post - editing the old one simply didn't come to my mind sorry

There is something, however, that I forgot to mention in my original post: It might only be imagination, but I have the impression that my miggy's sound is a bit clearer than before, especially the heights. But this could be due to the capacitor change in general, not necessarily due to the ceramics.

[Retrofan]

Originally Posted by amigasith

@DDNI

Thanks

@Retrofan:

- I only applied Stedy's fixes from here: http://www.ianstedman.co.uk/Amiga/am..._mobo_fix.html but skipped the extra wire links. So in total, I only removed E123C and E125C, changed R118 to 220 Ohms and added this 470 Ohm resistor as described. No ferrite beds added as replacements for the resistors, although my board also lacks the capacitors (E121C and E122C) described in the ferrite bed fix. I also noticed that my ACA1231 is not running as hot compared to before the fix - I guess this is due to one of the resistors...

- No problems... HighGFX is running nicely. I used it during my stability tests. But I like SuperPlus better - to me, it simply looks nicer than 1024x768.

@ prowler

Thanks to u also

Ok, thanks. So when I can I will see if I have the E121C and E122C caps... http://eab.abime.net/showthread.php?t=64629

[prowler]

Originally Posted by amigasith

@prowler

Thanks for merging my post - editing the old one simply didn't come to my mind sorry

No problem, mate. It just helps to keep the threads tidier.

Originally Posted by amigasith

There is something, however, that I forgot to mention in my original post: It might only be imagination, but I have the impression that my miggy's sound is a bit clearer than before, especially the heights. But this could be due to the capacitor change in general, not necessarily due to the ceramics.

Yes, changing the capacitors in the audio circuits is a recommended remedy for poor audio quality.

If you have noticed an improvement in the audio performance, then it confirms that ceramic capacitors are certainly equal to the task. Just think what those old capacitors might have been doing to the video signals!

[Loedown]

Indeed a really nice job, I would think too that the ceramics are of larger size they would be less inclined to suffer the fate of the very small ceramic capacitors, going short circuit when the board is flexed too much. It also stops the issue of having to use non polarized capacitors in some parts of the audio stage.

[amigasith]

@Retrofan:

I actually stubmled across your thread while doing my research. Sorry to read that you had so much trouble with your Indi I would guess that most of the troubles were caused by the short, however. My Indi (and also Alice and the other custom chips) are not running really hot - just hand-warm. One last remark: If you're looking for better stability by using ceramic caps (from your comments I thought that maybe you do, if not simply forget the following), I doubt that this will work. For me, it was exactly the other way round: My Rev 2B board ran stable before the surgery and I feared that I would lose stability by replacing the caps with ceramics. Fortunately, the stability remained.

@Loedown:

Thanks And yes, using larger ceramics is safer... Not only from a board flexing point of view, but I would also guess that they survive higher currents in case something else goes wrong. For me, the 1210 type series is a good compromise between size and functionality.

[Loedown]

Seems TDK is doing multiple stacked capacitors so if the problem with replacing the 100uF caps is the capacity you may want to check them out. Further investigation shows that some companies are doing mega stacks for PSU and specialty applications. With all this happening I would dare say that I will be offering a ceramic SMD solution in the very near future as aluminium SMD caps are crap.

[Loedown]

I have pondered this some more, all of the parts are available in 105 degree types and the through hole parts don't suffer as badly as the SMD aluminium ones.

I really can't see a reason why all Amigas could not have their capacitors replaced with these types, all I can see is benefits.

1. No more polarity issues with 3640 etc. where the mask was screwed up, or sound stages in most Amigas

2. No chance of leakage

3. Very rare for ceramics to go short, seen one in 20 + years of repairs

I reckon this thread should be stickied.


Even with the capacitor values rising it won't be long before the through hole parts can be replaced by a viable ceramic solution, 10 x 100uF @ 16V gives you a CD32 main filter cap for instance.

[john1979]

@amigaisth

Could you tell me what desoldering/soldering equipment you used to perform the job?

I've decided to do the eventual cap replacement myself rather than send to Amigakit (don't like the idea of my miggy on a delivery van). But I'm just starting out, having only done basic soldering with a cheap hand iron (megadrive ntsc mod).

[amigasith]

Cheers All

@Loedown:

I'm with you - my idea was mostly to replace the crappy SMD electrolytic caps, since those tend to leak much more often than their through-hole counterparts. The 1D1 board I did was really bad - while desoldering the caps I had the impression of working close to a harbor, if you know what I mean. It smelled badly like old fish - this was the spilled electrolyte of the stinky SMD caps. If you use decent through-hole electrolytic caps, those should last really long. And I'm not so sure how good very high capacitance ceramics like mega stacks really are as of today. All the hints that I found on the net said that you should better stay away from them until companies have gained more experience. Might be rumors only though...

@john1979:

I only used really basic stuff No kidding. I only have a regular soldering iron (a 20 year old Weller), not even a soldering station. As already said, I did not use hot air since I feared that it would cause more damage than benefit. I used a normal flat nose pliers to "twist-and-turn" out the SMD caps. Sounds scary, but works very well. Watch the Youtube video (link above) and just don't use any force. They come off very easily. Some practice with another old / broken electronics board helps of course. Besides this, I only used soldering wick to clean the board after the cap removal and some lighter fuel to remove the flux. Other people use isopropyl alcohol, but I didn't have IPA at home, so I used lighter fuel. The cleaning is not only for optical reasons - the soldering of the ceramic caps is much easier if the board is free from solder leftovers. For removing the through-hole caps, I used a solder sucker. Again, watch the Youtube video and you will see how this works - the guy did a real nice job in my opinion.

[Stedy]

Hi,

Nice job on the ceramic capacitor replacements, you might want to uprate a few of the components though.

The capacitance of a ceramic capacitor, with an X5R, X7R or X8R changes with applied voltage. The 1210 size, 6.3V X5R 100uF capacitors from Murata, connected across a +5V supply, will only have a working capacitance of 37-39uF. I use Kemet Spice as it is one of the best free, capacitor spice tools, to obtain these figures.
A better choice would be to use 2 x 47uF 10V capacitors running on +5V (50% bias against 80% for the 6.3V) as this yields an effective capacitance of 74-78uF over the capacitive frequency range. Electrolytic have much better stability of capacitance with applied voltage a 100uF capacitor will be 95-100uF, typically.

Lastly, be vary careful with how you solder some multi-layer ceramics. You must not hand solder them. The thermal shock of a soldering iron, on a part at room temperature to >400C can fracture the part. This has led to the device shorting out, in one instance, the parts got so hot, they de-soldered (>183C) and on another board, they caught fire and destroyed the PCB. Neither of this instances affected the aircraft systems. To solder larger ceramic capacitors by hand, the PCB and component are heated to >110C, then soldered in the normal manner, the assembly is then left to cool naturally. This reduces the thermal shock. Refer to the final pages of the Murata manual linked on the reichelt website.

One of the biggest cause of electrolytic capacitors failing, is lack of use, certainly the case of an Amiga left in storage for many years. In the Mil/Aero market (where I work) we restrict use of electrolytics (mainly use Tantalum and Ceramics), as any parts in stores, have to be taken out every two years and tested/conditioned. There is a MIL-SPEC for this but also look for capacitor reformers.

There is nothing wrong with using ceramic capacitors, the PowerPC Graphics system I developed @ work uses around 1000 ceramic capacitors, values from 10pF to 47uF and works well, but I did check all component values over the temperature and voltage ranges Just be aware of the parts you choose and their limitations. I could easily write an essay on capacitors, types and operation in circuits, because I spend too much time dealing with circuits.

As with all things electronic, your mileage may vary, I hope you have no issues with your new set up, I am not trying to alarm, just to educate.

Ian

[amigasith]

Cheers, Ian!

Crap, this is exactly the kind of feedback I was looking for Thank you very much for your insights – you really must have spent a lot of time just for the measuring of all the components over the years. Hats off

I have two questions for you:

- Which 47uF 10V capacitors would you recommend? And where can you get them? At least reichelt.de does not have any with those specs…
- Do you think the other ceramics (10/22/47 uF) are okay? Or should I try and get others for those also?

[rkupka]

Thank you for these valuable insights Stedy.
May I ask you to give us some recommendation about (SMD) tantalum capacitors ? Is it safe to use them as a replacement for alu electrolyte capacitors in our Amigas ?

[Loedown]

Originally Posted by Stedy

Hi,

Nice job on the ceramic capacitor replacements, you might want to uprate a few of the components though.

The capacitance of a ceramic capacitor, with an X5R, X7R or X8R changes with applied voltage. The 1210 size, 6.3V X5R 100uF capacitors from Murata, connected across a +5V supply, will only have a working capacitance of 37-39uF. I use Kemet Spice as it is one of the best free, capacitor spice tools, to obtain these figures.
A better choice would be to use 2 x 47uF 10V capacitors running on +5V (50% bias against 80% for the 6.3V) as this yields an effective capacitance of 74-78uF over the capacitive frequency range. Electrolytic have much better stability of capacitance with applied voltage a 100uF capacitor will be 95-100uF, typically.

Lastly, be vary careful with how you solder some multi-layer ceramics. You must not hand solder them. The thermal shock of a soldering iron, on a part at room temperature to >400C can fracture the part. This has led to the device shorting out, in one instance, the parts got so hot, they de-soldered (>183C) and on another board, they caught fire and destroyed the PCB. Neither of this instances affected the aircraft systems. To solder larger ceramic capacitors by hand, the PCB and component are heated to >110C, then soldered in the normal manner, the assembly is then left to cool naturally. This reduces the thermal shock. Refer to the final pages of the Murata manual linked on the reichelt website.

One of the biggest cause of electrolytic capacitors failing, is lack of use, certainly the case of an Amiga left in storage for many years. In the Mil/Aero market (where I work) we restrict use of electrolytics (mainly use Tantalum and Ceramics), as any parts in stores, have to be taken out every two years and tested/conditioned. There is a MIL-SPEC for this but also look for capacitor reformers.

There is nothing wrong with using ceramic capacitors, the PowerPC Graphics system I developed @ work uses around 1000 ceramic capacitors, values from 10pF to 47uF and works well, but I did check all component values over the temperature and voltage ranges Just be aware of the parts you choose and their limitations. I could easily write an essay on capacitors, types and operation in circuits, because I spend too much time dealing with circuits.

As with all things electronic, your mileage may vary, I hope you have no issues with your new set up, I am not trying to alarm, just to educate.

Ian

Excellent, I was waiting for you to show up Ian still waiting for Zetro

I knew about the capacitance changes in using parts that are greatly over rated, like using a 10uF 100V instead of a 10uF 16V but to speak about technique related to hand soldering of the ceramics. What are your thoughts on placing the ceramic in position and holding it there with tweezers before heating the far end of the pad and letting the solder wick up and automatically jump onto the end of the ceramic as a way of reducing thermal shock because few people are going to have access to pre heaters? Then by taking your tweezers off the component and applying the same technique to the other end do you think this will provide the best solution for the average hobbyist?

As I previously mentioned in this thread Murata and AVX are starting to make parts with higher capacitance based on multiple parts in a stack arrangement. It looks like the capacitors are spot welded much in the same way as tagged batteries are.

http://www.powersystemsdesign.com/li...t/figure3b.jpg

If a viable solution can be found then a very long term solution can be found and make more of our precious hardware nearly storable indefinitely or close to it without fear of degradation. The damage that aluminium electrolytics does is horrendous.

[Stedy]

Hi,

@amigasith

These Murata parts from Farnell would be good enough:
Farnell Stock code 1797008, manufacturer part number GRM31CR61A476ME15L.
For the 100uF parts, these TDK parts would be good, when back in stock:
Farnell stock code 2211193, manufacturer part number C3216X5R1A107M160AC

Your original 47uF parts are 16V, they will be fine, as will be the 10 & 22uF, it's only the 100uF capacitors that could change.

I am currently preparing orders with Farnell and Rapid Electronics, so if anyone wnats some parts, PM me. I don't worry about minimum orders

@rpupka

Tantalums are suitable replacements, just ensure two things, you match or improve the voltage rating of the original capacitor and for larger value (100uF and above), you pick Low Equivalent Series Resistance (ESR) parts. Electrolytics have ESR of 2 ohms to 50 m ohms, standard Tantalums can be 12 ohms to 50 m ohms, depending on type.

If you follow those rules and fit them correctly (important for the CD32) they should be fine.

I have re-capped my A600/A1200 and CD32 with Aluminium electrolytic capacitors. I think 3 had gone on the A600 (now 20 years old), mainly due to lack of use. The A1200 seemed fine but this was done before I bought my ESR meter

My A500, which needs repairs, the through hole electrolytics, were fine. Capacitance and ESR were as expected.

@Loedown

My engineering workaround to the capacitor issue invovles a heat gun and thermocouple.
I gently heat the PCB and part to about 120C, the thermocouple records the temperature, then solder to the board in the normal method, this reduces the thermal shock and helps with 18 layer PCBs.

The Murata/AVX parts you linked to, I would steer clear of. They are the type that have caught fire, when soldered incorrectly. The sheer amount of heat you need to solder that much metal, using s soldering iron, fractures the part and this leads to short circuits and sparks.

Apart from my A600, I think that two other devices that used electrolytic capacitors, one of which was used outside (with conformal coating) over the last 15 years, have had failures.

Until the next time!

Ian

[cosmicfrog]

wow