A1200 weak crackly sound - both channels went overnight!

[8bitbubsy]

Originally Posted by Loedown

The A1200 is indeed quite a different circuit to the A4000 audio [...snip]

Really? I thought A1200 and A4000 shared the exact same component design in the audio circuitry. I know that A500/A2000 (I think) and A600/A1000/A2000/A3000 differs, though.

[Allen1]

Originally Posted by Photon

I traced -12V to the via near RS232 yesterday, but I can't trace -12V to any pin on U28. If it's connected to it, the trace to the pin is under the IC itself. Is it supposed to/what did you mean?

Take a look at the decoupling schematic on the sheet I have uploaded, it shows U28 to be connected to -12v via pin 1 through diode D576. Another -12v connection can be found on the video connections before the actual connector on what looks like E255 (not shown in this sheet). Out of curiosity, you have downloaded the schematics for the 1200, I know that they are not very clear in places and the numbers are often hard to read plus there are variations of build in the motherboards but the audio design is basically the same throughout.


Basically, if you are getting -12v readings from these points then there is nothing wrong with the -12v tracks and nothing is short circuiting that supply rail or causing problems with the audio op-amp. The -12v for U28 would be found at D576, as you mentioned that you were getting a -12v reading around that area then the likelihood of something being wrong with that power supply is very slim especially with it being a low current supply. If you wanted to check the -12v line further then simply break the U28 circuit by removing D576 instead of touching U28 but I suspect that to make no difference in reality.

I still believe that you have a broken chip resistor or capacitor or maybe more track damage from the leaking capacitor than you found, failing that the next thing would be the actual op-amp. There isn't much more that could go wrong apart from ESD damage.


You will have to save the image and load it from your computer and enlarge it to see the info.

[Photon]

Yeah it was late last night, this morning I found it - I will remove the diode (detach U28 from -12V) and see if that changes anything. (After the diode it also goes to some extra solderpoint - is this just a test point, cos I can't trace where else it goes?)

I also traced the -12V to the video DAC days ago, but dismissed it, thinking "but the video output is fine" (the -12V was connected to the opamp at the time and measured -7V at the time). But perhaps -7V was no problem for it, so I'll have another look.

Did I reply about R324/334...? Well, they check out (even if I think it's much later in the game?)

[Allen1]

I'm slightly pleased that the resistors were ok as they are awkward to replace where they are (lots of nice plastic to avoid with the soldering iron).

[Loedown]

Originally Posted by 8bitbubsy

Really? I thought A1200 and A4000 shared the exact same component design in the audio circuitry. I know that A500/A2000 (I think) and A600/A1000/A2000/A3000 differs, though.

Go look through the schematics and you will see the differences, the A1200 has similar to the A500 where the A4000 is quite different. I just picked the A4000 audio to discuss because it's the one I see the most of and repair.

[8bitbubsy]

A1200 does not have the static lowpass filter, instead the Sallen-Key- and Butterworth-filter is slightly changed. A500 does have the static lowpass filter, independent on the "LED filter" status. I can't believe that they still had the static cut-off filter components in the A1200 circuit, isn't that a waste of money? I don't have the schematics for A4000, I'll fetch it soon and make a picture of that one too.

http://16-bits.org/amiga/a1200/a1200_audio_circuit.png

EDIT: Ah nevermind, the static lowpass filter is there, but the cut-off is 26kHz so it's disabled.

Fun facts from my ProTracker port project (this part by aciddose/aderj):

	/* Amiga filter emulation
	**
	** First comes a static lowpass 6dB formed by the supply current
	** from the Paula's mixture of channels A+B / C+D into the opamp with
	** 100nF capacitor and 360 Ω resistor feedback in inverting mode biased by 
	** dac vRef (used to center the output).
	**
	** Under spice simulation the circuit yields (F - 3dB) = 4400Hz.
	** In the A1200, the cutoff is 26000Hz, so the static filter is disabled.
	**
	** Next comes a bog-standard Sallen-Key filter with:
	** R = 10k Ω
	** C0 = 6.8nF
	** C1 = 3.9nF
	** Q ~= 1/sqrt(2)
	**
	** This filter is optionally bypassed by a MPF-102 JFET when
	** the LED is disabled.
	**
	** Under spice simulation the circuit yields (F - 3dB) = 2800Hz.
	** 90° phase = 3000Hz (so, should oscillate at 3kHz!)
	**
	** The buffered output of the Sallen-Key passes into an RC highpass with:
	** C = 22µF (also C = 330nF, for improved high-frequency)
	** R = 1.39k Ω
	**
	** F = 1 / 2πRC
	** High Hz = 5.2~ = 1 / (2π * 0.000022 * 1390)
	**
	** Under spice simulation the circuit yields (F - 3dB) = 5.2Hz.
	*/

[Photon]

Unsoldering D576, I get 2V on the -12V (.68V on the U28 power leg). I think I saw it coming up from -7V to +3V the first time I measured which points to a capacitor. I will give up the figuring out quest and go all zen and swap all silly little SMD electrolytics for brand new pin mount electrolytics, turned the correct way and inspect & clean the space under them, then replace the opamp if the voltages look ok.

It's just that I like to find out reasons.

[Loedown]

Originally Posted by Photon

Unsoldering D576, I get 2V on the -12V (.68V on the U28 power leg). I think I saw it coming up from -7V to +3V the first time I measured which points to a capacitor. I will give up the figuring out quest and go all zen and swap all silly little SMD electrolytics for brand new pin mount electrolytics, turned the correct way and inspect & clean the space under them, then replace the opamp if the voltages look ok.

It's just that I like to find out reasons.

C324 / C334 22uF 16V non polarised.

[8bitbubsy]

C324/C334 are AFAIK just crossover caps filtering inrush ripple for safety reasons, like unplugging/plugging in audio cables? They should not make any trouble else than low sound/cracking sound when they are bad. If you want, you could make a short instead of the caps just for testing reasons. Remember that they are there for a reason though.

[Loedown]

Originally Posted by 8bitbubsy

C324/C334 are AFAIK just crossover caps filtering inrush ripple for safety reasons, like unplugging/plugging in audio cables? They should not make any trouble else than low sound/cracking sound when they are bad. If you want, you could make a short instead of the caps just for testing reasons. Remember that they are there for a reason though.

They are there to stop DC bias from going into the audio connectors and they need to be non polarised as has been discussed ad infinitum because the voltage swings between positive and negative, even the schematic I have of the A1200 from the archive of Cosmos' A1200 1D PDF lists them as NP or non polarised.

[8bitbubsy]

Yes, they should be non-polarised, but polarised ones does the job fine as well. AFAIK they're more stressed this way, but it still works. If you don't have any non-polarised 22µF caps, then why not use polarised ones?

[Loedown]

Originally Posted by 8bitbubsy

Yes, they should be non-polarised, but polarised ones does the job fine as well. AFAIK they're more stressed this way, but it still works. If you don't have any non-polarised 22µF caps, then why not use polarised ones?

I've had this argument time and time again with people, ultimately it's your hardware and you do with it what you will.

1. Polarised capacitors in these locations will fail prematurely versus non polarised parts.

2. Commodore screwed up the original design and decided, probably as a cost-cutting measure to use polarised parts because they are easier to obtain.

[Photon]

All my 22uF are 25V polarised SMD electrolytics, but for C324/334 I guess 16V or even lower is ok since - well isn't it for regular line out levels, 8-ish Ohm?

Agree power line caps can be polarised and it seems strange putting this kind of caps on outputs that can be negative to common. Hence 'silly little caps' above, and a reason why my 'must understand this circuit' fetish leading to all kinds of questions might be a salmon-pink herring in unisex jumpsuit.

Loedown, are you saying the charge rise I saw points to those caps being bad also? I should update and say that now when I measure, I get 2.0V with no rise - every time - so it might have been a stored charge somewhere that is now (with the D576 unsoldered) never being re-stored.

[Loedown]

Originally Posted by Photon

All my 22uF are 25V polarised SMD electrolytics, but for C324/334 I guess 16V or even lower is ok since - well isn't it for regular line out levels, 8-ish Ohm?

Agree power line caps can be polarised and it seems strange putting this kind of caps on outputs that can be negative to common. Hence 'silly little caps' above, and a reason why my 'must understand this circuit' fetish leading to all kinds of questions might be a salmon-pink herring in unisex jumpsuit.

Loedown, are you saying the charge rise I saw points to those caps being bad also? I should update and say that now when I measure, I get 2.0V with no rise - every time - so it might have been a stored charge somewhere that is now (with the D576 unsoldered) never being re-stored.

If your capacitors are 25V then I recommend replacing them with 25V, on every other Amiga model they are 16V

A capacitor will not hold charge for an indefinite period, only supercaps were designed to hold charge for a considerable time. The problem you have with this -12V thing is quite bizarre and I am having some trouble trying to diagnose the unit without it sitting in front of me. By replacing the capacitors you are removing one very well known source of problems. If after that you're still having issues we can discuss it.

[Photon]

OK, decided yesterday to give this mobo one more chance before I swap it.

The -12V line is disconnected from the PSU and the serial port end of things.


I examined the electrolytic spill area where the original problem of "corroded via on the +12V first appeared.

With a patch wire the +12V is ok, so I moved on to the current problem, the -12V.

I unsoldered the 10 Ohm resistor and its cap, and that removed the "rising voltage" from 2V and slowly upward from the -12V line. Hm.

I traced the +12V backward and got a strange voltage past the resistor nearest the keyboard connector, 9V compared to 11.85 on my other A1200.

I cut the trace from the resistor to the nearest via, and that fixed it right up.

I concluded that a layer has been damaged inside the PCB near that via. That might also be the cause for the strange 2V, when it should be 0V since the -12V is disconnected.

I cut traces to vias near the spill and sure enough, voltages totally different and bad at the vias, while the unconnected part of the trace were not.

I unsoldered the op-amp and cut traces to via underneath it, too.

I cleaned up the PCB thorougly.

Before this fixup I could actually measure 6-9V by holding the pin to the laquer between copper on the PCB!

Now it's down to 0.5 V, dunno if it's a problem still.


The one issue I have left is to check pin 9 (well, input 1 - pole) on the op-amp. On my other A1200 this is 2.5V, whereas on mine it's 1.18V.

Also, the completely disconnected -12V on the opamp leg pad is 1.88V. I don't know if it's because I measure 0.9V to the substrate (laquer) near it, but I don't care because I will replicate the -12V R/C circuit and connect it directly to the chip leg, not to the leg pad.

I will check the Vref, hopefully it's 2.5V.

But since the Input 1 - pole pin is 1.18 and goes to a via under the op-amp, I'd like to cut that as well and maybe replicate the R/C.

I'd like to know

1. What capacitance is the small darker SMD caps, is it documented? The ones that go with 470 Ohm resistors.
2. Where does that via go to/come from, once it's gone under the opamp and into the board? Logically, from the VREF generator left of the right CIA chip, but I haven't found it by measuring or looking at the underside, when I get home I'll have time.

[Loedown]

Photos old boy photos.

Arrows are good too, we like arrows.

[Photon]

Holding my hand over the flash to get a less shiny shot, it seems that red eye reduction or whatever hilights unevennesses in the surface in red (disregard hilights on traces and pins, it's on the right side:




With the flash, the clean shot of the leak area looks like this (VERY revealing, you see the tiniest thing, even dragging the multimeter over the top opamp solderpads once, doesn't look near this ugly even with a lupe):


I've isolated the damaged internal layer by cutting off traces coming from vias (marked red):

Big handdrawn contour in red is my estimate of the spread inside the PCB, on the topside it's only between C303/C304 and up to the unsoldered SMD components top right. Underside looks brand new.

There is 1 soldered via, marked blue (which I assume go to an internal layer unlike the non soldered vias) that I'd like to cut off.

The problem is, on the underside there's just a solder point and no trace.

And for the life of me I can't find any voltage or point on the mobo that it goes to.

It's connected to the lower (in the picture) end of C711. If someone can help me out that'd be appreciated

Also, a reliable point to measure VREF before it comes to R/C at opamp pin 9, because I forgot

What I'm trying to do is cut out the sickness and reconstruct everything identically to my spare mobo, to save this mobo. For example: R301 and C301 will not be resoldered to those ugly solderpads but reattached to the right points with non SMD parts, C303 and C304 likewise (since +/-12V pads are cut off).

[8bitbubsy]

So I spent some hours in my lab desoldering components on my broken A1200 mobo, then I scraped off the protection layer on the traces. This is an extremely good way to see if you did some traces wrong. I had to draw some of the lines myself in photoshop, but I can assure you that it's 100% exact!
http://16-bits.org/amiga/a1200/a1200...uit_traces.jpg

[Loedown]

What did you use to get rid of the lacquer off the tracks, it's a clean job?

[8bitbubsy]

First I removed all the solder and components, then I used ... *drumroll*... Some kind of liquid you use to remove things that get stuck on your oven. It's a liquid that is kind of rough, so it scratches off the outer layer.
I just thought I'd go wild and try this since I saw the bottle, and it worked very well. You need to scrub like a maniac, but it works. It's not something I'd do on a PCB that works, but this mobo was toast anyways.

It was not a too clean job, but it wasn't dirty or messy either.