Powering up the radio on the bench

[willnotwish]

I'm going to try powering up the board on the bench. I have a really old board purchased as a spare some years ago when I looked at this issue before. I have "modified" it extensively. It may not run. Let's see.

Originally posted by @willnotwish in https://github.com/willnotwish/colt-radio/issues/1#issuecomment-2308744325

[willnotwish]

When I say "modified", I mean butchered.

[willnotwish]

I should be able to power up any of the three boards.

[willnotwish]

I've made a test power cable which simulates the "accessories on" key position in the car. The radio ought to work with the 12V battery +ve permanently connected.

[willnotwish]

I have a pinout for the various connectors but it's not the correct model because there is a "remote control" connector included.

[willnotwish]

Looking at board C I can see pins 7-14 marked where the power supply 14-pin connector should be. (I removed it ages ago.)

Pins 10 ("Battery") and 11 ("Accessory") appear to match the diagram above.

[willnotwish]

The diagram also gives a clue as to the comms between the radio and the display, There are three lines in the diagram labelled M-SCK, M-DATA and M-BUSY. They are available at what is (on my units, not the diagram) the 14-pin display connector (what I have until now called red/white wires). They are also duplicated on the 13-pin DIN socket for an optional CD changer. It may be more convenient to probe the DIN socket, rather than the display connector.

[willnotwish]

On the bench I see no activity on any of the lines M-SCK, M-DATA or M-BUSY.

If I take unit A and plug it into the connectors in the car I see data. But on the bench -- nothing.

[willnotwish]

M-SCK sound like a serial clock.

[willnotwish]

I'm going to try mimicking my bench setup in the car by connecting pins 10 (battery 12V always on) and 11 (accessory 12V) ONLY.

[willnotwish]

With pin 10 connected to +12V in the car and a switched pin 11 unit A behaves as before. In other words, having the speakers connected makes no difference. This is not much of a surprise.

I noticed, however, that the display -- when disconnected from the radio -- appears to drive one of its lines (one of M-BUSY, M-SCK or M-DATA) high (I measured 4.6V) about once per second.

When I powered up the radio on the bench in isolation (no display - that's still in the car and I don't want to remove it), I noticed the same line pulsing low for about 6ms at about that rate (maybe a bit slower).

[willnotwish]

Maybe the radio needs to see some sort of wake-up signal from the display before it will start generating data. That would explain why I can't see any signals from the radio when on the bench.

[willnotwish]

Maybe it's not the radio that is at fault. Maybe it's the display.

[willnotwish]

When I connected the new radio (unit B) in the car the display worked for about 5 minutes before going blank. Perhaps there is nothing wrong with unit B.

[willnotwish]

A Google for Mitsubishi Colt clock yields a bunch of results to do with dry joints and dodgy soldering in what I am calling the display module. It seems to be a common problem. Before searching for a replacement used clock module (which may well have the same problems as the one I have now), I will try cleaning the PCB and reflowing some of the dry-looking joints. Several of the chips (microprocessors I think) are covered in a kind of rubbery glue. Maybe this is the problem. It is tedious to remove.

[willnotwish]

I reflowed every solder joint in display/clock module (there are two PCBs, which I separated to gain good access).

I reconnected the DCM (as I will now call it) ... and ... no change! Bah.

Unit A works with coaxing (but has no sound). Unit B shows no interest at all (but has sound).

When it is working, pressing the Display button on unit A alters the top line on the DCM. When comms are established, all functionality is available. The DCM will go blank on the bottom line after 5 seconds or so, though.

[willnotwish]

A further possibility is a defective connection between the two: a wiring problem.

A break in the wire would show up in a continuity test. But we know the problem is intermittent, and I do not seem to have to move the cables around to get unit A to display something.

[willnotwish]

Unit B definitely worked perfectly yesterday for about five minutes.

But it's strange the way its behaviour differs from that of A. A will connect when coaxed; B will not.

[willnotwish]

I think I need to eliminate a wiring issue as the root cause.

I'm going to do some continuity checks between the radio to DCM connectors.

[willnotwish]

Given that I now have the DCM out of its surround and just sitting on the dash, I could move it to the bench and try to get A or B plus the DCM working on the bench in order to analyse the protocol.

[willnotwish]

I can't get the DCM working on the bench. I would expect to be able to supply 12V and get it to show the time of day. But so far, nothing.

[willnotwish]

It's hard not to think there is a loose connection somewhere: a break in a wire or a poorly fitting plug and socket connection.

But the behaviour of units A & B is different.

It's a mystery indeed.

[willnotwish]

At one point I thought that the software running on the new radio (B) was incompatible with the DCM. But that theory seems to have been disproved by the two working perfectly for five minutes before radio silence returned.

[willnotwish]

I have both units talking to each other on the bench at last. Here are some typical waveforms:

and, zoomed in a bit

[willnotwish]

Top trace is (what I am calling) marked cs but in the connector pin out above it is referred to as M-BUSY.

Yellow trace marked data is M-DATA.

Bottom (blue) trace marked clk is M-CLK.

[willnotwish]

I am no expert, but it looks like an SPI connection, i.e., serial peripheral interface

Looks half-duplex, main being the radio and sub being the CDM.

M in M-DATA would seem to stand for main (formerly master); MOSI in SPI terms.

[willnotwish]

From inspection of the the lower trace, the data byte sent by the radio over MOSI is (assuming logic level 1 is HIGH 5V) in binary is

01010111

That's HEX 57.

[willnotwish]

I'm not sure whether my scope can decode the data. (Goes off to have a look.)

[willnotwish]

It can, but not very reliably.

Unless this is the reason why unit A "fades away".

[willnotwish]

I tried Unit B (traces above were from A).

I saw nothing, apart from a periodic blip on the main 12V input. A bit unexpected, to say the least! Some short-circuiting of the line perhaps. I need to investigate more.

In addition, now I have started to dismantle unit B. I see that the PCB in B is different (bigger, with more components on) than A and C.

[willnotwish]

In addition to CS, MOSI and CLK, there is one more direct connection (the brown wire) between the radio and CDM.

What is its function? I will try disconnecting it to see what difference it makes.

[willnotwish]

Turns out be an earth. Big deal!

[willnotwish]

I think the periodic blip was caused by the power amp chip overheating and recycling power. Maybe some sort of crowbar protection causing the power supply to shut down.

Why? I had removed the heatsink. I replaced it and it seems OK now.

The thermal compound paste had dried up. I have none to hand, so I must order some.

[willnotwish]

With unit B connected to the CDM, the CS line (red) is a continuous 1.3V.

If the output driver is faulty, I would still expect the microprocessor output to be working.

I need to probe the correct pad on unit B's PCB. The layout is different from A's.

[willnotwish]

The i/o pin of the MPU is steady at some intermediate voltage: 3V or so.

Is it being held there by some fault in the output driver? Or is the MPU faulty?

[willnotwish]

If I could disconnect the output driver from the MPU I could diagnose further.

[willnotwish]

I cut the trace and the voltage dropped to 0V. But there was still no activity on the CS line.

[willnotwish]

In preparation for swapping MPU chips (not an easy task) I stripped the glue from the MPU on unit A (the original radio). Then I checked the signals using unit A and the CDM. They were as expected, hence showing that I hadn't damaged anything. (The degreaser I use to remove the glue is pretty fierce.)

After watching the traces for a little while I realised that the setup was working perfectly with no dropouts or coaxing needed.

I left it running for a while and all was well.

So I reassembled the original radio (A) and the CDM and re-installed them temporarily in the car.

The audio worked perfectly, as did the display on the CDM!

[willnotwish]

My theory at present is that the glue that the radio manufacturer used to secure the MPU had, with age, become conductive.

Googling for this problem shows it to be a hot topic amongst audio equipment manufacturers.

[willnotwish]

I have now removed the glue on the CDM (as part of reflowing the solder joints when I thought that the CDM might be to blame), and that in unit A around the MPU.

I now need to remove the same from unit B (unit C has none because I removed it years ago so I could probe the pins - it's just a spare parts board now) and see if B works as well as the now glue-free A.

[willnotwish]

One small disappointment is that I can't play CDs in unit A. There seems to be something stuck in the playback unit which stops a CD being inserted.

I tried swapping the playback units from A & B but A didn't like it and wouldn't play. It was getting dark then though and it had been a long day so maybe I was just a bit confused.

I'm going to take both units apart again, give them a good clean and reassemble.

[willnotwish]

If the problem turns out to be the glue (and it certainly looks like it) then I will remove it wherever I see it.

It's likely that glue has been used in other modules in the car. That may include the module responsible for central locking! I'll have a look at that next.

[willnotwish]

I'm having trouble with the CD players. The player in the original radio (A) looks old and in poor condition, unlike the rest of the radio. Maybe I swapped the original CD player ages ago with the player from radio C. It certainly looks rough. When I opened it I found several parts missing, so it would never have worked.

I dismantled both players. They are very difficult to get back together, with hard to position parts and fiddly springs. I worked on them all day with no positive result. The CD unit in radio A would allow me to insert a CD, but refused to play it and ejected it with error code 1. According to the car instruction book (the "manual") that can be caused by dirt or moisture in the unit. I think it might be that the parts inside are dried up and stiff to move. I swapped what I think is the moisture sensor with the one in CD player B. So now I am totally confused.

[willnotwish]

The replacement radio (B) still does not talk to the CDM. The bottom half of the display is blank. I think there is another fault with that unit apart from the glue (which I removed).

[willnotwish]

I managed to reassemble the CD player from unit A (the original). When I fitted it back into A and put it in the car, it accepted a CD and started playing. But there was no audio.

I ejected the CD OK and witched to the radio, which refused to find any stations. There were a few loud clicks and pops and still no audio. I reinserted the CD. It started playing, but at a low volume, with distorted sound, and which could not be controlled by the volume control.

Communication with the CDM remains perfect. I am able to control the mpg/range indicator as well.

[willnotwish]

I must have done something to unit A, either when cleaning the glue from other chips (I'd already cleaned the MPU) or when connecting the CD player.

Unit B still plays FM & AM stations. Its audio is fine, even if it refuses to communicate with the CDM and its CD player is in bits on the bench, having been cannibalised. That's a bit annoying as it worked perfectly when I first had it.

[willnotwish]

Removing the glue seems to have solved the problems with the digital monitoring functionality, but has messed up the audio big time.

[willnotwish]

I think I will have to put both units A & B on the bench, connect the same aerial to both and compare the two by probing with the scope.

[willnotwish]

When fiddling with the Audio settings controls on the front panel I managed to select what I think may be an auxiliary analogue input channel. I am not sure if that is just the software, or an aux input actually exists. If so, I could use that as a further test.

[willnotwish]

I looked again at unit A. There was a spring (from the now removed CD drive) at the back, touching the pins of the power amp chip. :firecracker:

I removed it and tried unit A in the car, but with no luck. No audio, no tuning, but a perfect display.