Music 500 -> Music 5000 and Music 500 -> Music 3000

[maniacminer]

I am at a loss as to how to test a pair of M500 that I did the resistor mod to and another I did the resistor mod to, added the two capacitors and a wire mod to the other M500 to make it into a M3000. Of course, all I get is silence (apart from the thump when I switch it on) is there any way I can test what is working?

[DutchAcorn]

Are you looking for specific M5000/M3000 tests or a way to check if they are working at all? If it is the latter you can use an M500 disc image that I posted here.

[maniacminer]

Are you looking for specific M5000/M3000 tests or a way to check if they are working at all? If it is the latter you can use an M500 disc image that I posted here.

Thanks, I was looking for the former, I have the M500 disc and nothing happened. I just hooked the system up to the Beeb and the Beeb won't even boot up with it connected. That means one of the units is messing with the 1MHz bus in some way... Time to re-check those bodges that made a M500->M3000 and a M500->M5000...

[Coeus]

Obviously, you need to sort out the issue with the unit upsetting the 1Mhz bus first.

It must be possible to write a small test that accesses the hardware directly but I don't know of anyone who has done it. The way to test is to load one of the versions of AMPLE and trying playing a tune or even something basic like "SOUND ON GATE" (AMPLE BCE). It doesn't matter which version of AMPLE you use: the disc-based AMPLE BCE (supplied with the M500) or the ROM-based AMPLE Nucleus (supplied with the M5000) but any AMPLE program you try to load must be for the right version of AMPLE.

The Music 500/5000 is a write-only device so there is no way of reading an I/O area to determine if it is there.

[maniacminer]

Looking at this post viewtopic.php?p=183602#p183602

I found a bent pin on the 1MHz bus

Is this the problem?

nope...

Checked my bodge wiring

Bodge wires look ok

[hoglet]

It must be possible to write a small test that accesses the hardware directly but I don't know of anyone who has done it.

There is a small BASIC test case here:
https://github.com/hoglet67/Music5000/w ... modulation

For the Music 3000, I think the only change would be:

Code: Select all

?&FCFF=&5E

This does depend on running the AMPLE ROM/Disc before hand to load the default waveforms. But that would be easy to do from BASIC as well, then it would be standalone.

[maniacminer]

Hmmmm, I was thinking of something to exercise M5000/M3000 so I could look at the data path through the board with the scope. I found a modern schematic here https://github.com/jlprojects/music5xxx ... ematic.pdf and am thinking of going through that and seeing if I've got a blown IC somewhere that is stopping the Beeb from starting up. I am suspecting the Beeb looks at the 1MHz bus for part of its boot process and somehow the M500(0) is asserting something and that makes the Beeb freeze. It doesn't seem to harm the Beeb at all, no extra current over normal boot.

[maniacminer]

Going through it with the scope, I noticed that the 1MHz clock is almost completely gone when I measure it at pin #4 on the extension connector. Using the multimeter, I noticed there was zero ohms across the bodge resistor, so I cut one side and there is still continuity between pin #11 of IC34 and the 1MHzE input and still zero ohms. I cut the bodge wire leading from IC34 pin#11 to IC31 pin#7 and that made no difference. So that leaves me with a fault in the attached ribbon cable/connector, or there's something underneath the connector (one of the reasons I hate vias underneath connectors) it looks like I will have to desolder the 1MHz bus expansion connector first, then if the fault doesn't clear, remove the ribbon connector. I'd love a scan of the bare PCB to make it easier to trace faults like this.

[hoglet]

Have you fitted a 4K7 resistor in the mod?

I believe the Music 500->5000 mod had been incorrectly described....

Apparently it's actually a 470R resistor in series, and at 220pF down to ground, which make much more sense, because it acts as a simple RC filter to remove noise from the 1MHz clock.

Eelco discovered this:
viewtopic.php?p=282670#p282670

Dave

[maniacminer]

Have you fitted a 4K7 resistor in the mod?

I believe the Music 500->5000 mod had been incorrectly described....

Apparently it's actually a 470R resistor in series, and at 220pF down to ground, which make much more sense, because it acts as a simple RC filter to remove noise from the 1MHz clock.

Eelco discovered this:
viewtopic.php?p=282670#p282670

Dave

Thanks for the clarification. I had fitted the 4k7/220nF without thinking about the values beyond what was on the Internet (dangerous, I know) viewtopic.php?p=15326#p15326 this in particular puts in a resistor between two points already connected via a PCB track on the other side of the board! So what's the point of the mod? The resistor is bypassed by the PCB track... Am I supposed to cut the track going to pin #11 on IC34 on the top side of the board?

[hoglet]

I had fitted the 4k7/220nF without thinking about the values beyond what was on the Internet (dangerous, I know)

The values from eelco's post are 470R and 220pF (0.22nF) which make sense as a filter to eliminate noise on the clock.

The resistor is bypassed by the PCB track... Am I supposed to cut the track going to pin #11 on IC34 on the top side of the board?

If the resistor is bypassed by a track, that's an error. I'm not sure which track you are talking about here, but on Eelco's photo the track is clearly cut.

I think one end of the 470R resistor should go only to pin 4 of the 1MHz Bus connector (the 1MHz clock), the other end should feed all loads on the 1MHz clock network. The 220pF capacitor should, ideally, be positioned between the first load and ground.

What else does the 1MHz clock connect to, apart from IC34 pin 11?

Dave

[maniacminer]

I had fitted the 4k7/220nF without thinking about the values beyond what was on the Internet (dangerous, I know)

The values from eelco's post are 470R and 220pF (0.22nF) which make sense as a filter to eliminate noise on the clock.

The resistor is bypassed by the PCB track... Am I supposed to cut the track going to pin #11 on IC34 on the top side of the board?

If the resistor is bypassed by a track, that's an error. I'm not sure which track you are talking about here, but on Eelco's photo the track is clearly cut.

I think one end of the 470R resistor should go only to pin 4 of the 1MHz Bus connector (the 1MHz clock), the other end should feed all loads on the 1MHz clock network. The 220pF capacitor should, ideally, be positioned between the first load and ground.

What else does the 1MHz clock connect to, apart from IC34 pin 11?

Dave

Eelco's resistor mod is pointless unless the top track is also cut. I have added the top surface track in Eelco's photo.

There's a track on the top surface going to where the resistor is...

So my question is, am I missing part of the instructions where I have to cut the top track as well as the bottom track? The two locations the the 1MHzE signal goes to are pin #11 on IC34 and pin#7 on IC31. Removing the 220nF/4k7 filter gave back the 1MHzE but when the "3000" is added, the "5000" stops working.

[hoglet]

So my question is, am I missing part of the instructions where I have to cut the top track as well as the bottom track?

I would say yes, the there must be a cut on the top track that has been missed.

If someone has an original Music 5000, these mods should be apparant.

Do the photos on this page make any sense to you:
https://www.retro-kit.co.uk/page.cfm/co ... -Expander/

I think one of those track cuts (near C38) is cutting the 1MHz clock track, so is actually unrelated to the 5000->3000 conversion.

Dave

[hoglet]

The difference between a Music 5000 and a Music 3000 is the range of Jim pages they respond to. More specifically:
- an M5000 (and an M500) will respond when FCFF bits 7..4 are 0011
- an M3000 will respond when FCFF bits 7..4 are 0101

The M5000->M3000 conversion involves just this circuit change:

I think it's achieved by:
- cut existing track close to IC35 pin 10
- connect IC32 pin 7 (1MHz Bus D5) to IC35 pin 10
- cut existing track close to IC14 pin 1 (*)
- connect IC32 pin 13 (1MHz Bus D6) to IC14 pin 1

(*) I don't know exactly where this cut is made. Maybe with access to the PCBs you could confirm the details.

BTW, the full Music 500 schematics are here

Dave

[maniacminer]

So my question is, am I missing part of the instructions where I have to cut the top track as well as the bottom track?

I would say yes, the there must be a cut on the top track that has been missed.

If someone has an original Music 5000, these mods should be apparant.

Do the photos on this page make any sense to you:
https://www.retro-kit.co.uk/page.cfm/co ... -Expander/

I think one of those track cuts (near C38) is cutting the 1MHz clock track, so is actually unrelated to the 5000->3000 conversion.

Dave

Ha! yes, I cut the track, eventually, out of sheer frustration, but the M5000 still stops playing when the M3000 is connected. Then I went looking for duff ICs that would somehow upset the 1MHz bus and found these two. Replaced and still no sound from the M5000 when the M3000 is in-line.

[hoglet]

Did the M3000 start off as a working M500?

Can you detail the changes you made?

Dave

[hoglet]

In case it's any use, here's a BASIC test program for the M5000 and M3000.

m5000test.ssd

(1.5 KiB) Downloaded 3 times

It's standalone, and doesn't depend in any way on the AMPLE ROM.

Code: Select all

   10  REM &30 (M5000) or &50 (M3000)
   20  D=&30
   30  FOR B=0 TO 15 STEP 2
   40  ?&FCFF=B+D
   50  FOR J=&FD00 TO &FDFF
   60  ?J=0
   70  NEXT
   80  NEXT
   90  REM SETUP WAVEFORM 0
  100  ?&FCFF=D
  110  FOR J=&FD00 TO &FD3F
  120  ?J=127
  130  NEXT
  140  FOR J=&FD40 TO &FD7F
  150  ?J=255
  160  NEXT
  170  REM SETUP WAVEFORM 12
  180  ?&FCFF=D+12
  190  FOR J=0 TO 127
  200  J?&FD00=SIN(2*PI*J/128)*127
  210  NEXT
  220  ?&FCFF=D+14
  230  ?&FD00=&00
  240  ?&FD10=&05
  250  ?&FD20=&00
  260  ?&FD50=&00 :REM Waveform 0 is a square wave
  270  ?&FD60=&00 :REM Min amplitude
  280  ?&FD70=&2D :REM Set bit 5 to modulate next channel
  290  ?&FD01=&BE
  300  ?&FD11=&6D
  310  ?&FD21=&01
  320  ?&FD51=&C0 :REM Waveform 12 is a sin wave
  330  ?&FD61=&70 :REM Normal volume
  340  ?&FD71=&0D :REM Clear bit 5 not modulate next channel
  350  ?&FD81=&BE
  360  ?&FD91=&6D
  370  ?&FDA1=&02
  380  ?&FDD1=&C0 :REM Waveform 12 is a sin wave
  390  ?&FDE1=&70 :REM Normal volume
  400  ?&FDF1=&0D :REM Clear bit 5 not modulate next channel

It should produce a two tone siren sound.

To test the M3000, change line 20 to:

Code: Select all

20 D=&50

Dave

[maniacminer]

Did the M3000 start off as a working M500?

Can you detail the changes you made?

Dave

I thought I had tested it, but now I'm not sure. So I have to assume I had tried to modify a dead M500 into a dead M3000...

a) removed PCB track from via carrying the 1MHzE signal to pin #7 on IC31
b) soldered 4k7 resistor from the via to pin #11 on IC34
c) soldered 220nF capacitor between the pins #7 and #8 on IC31
d) soldered a wire from pin #11 on IC34 to pin #7 on IC31
e) soldered a wire from pin #10 on IC35 to pin #7 on IC32
f) cut track leading to pin#10 on IC35
g) soldered a wire from pin #13 on IC32 to pin #1 of IC14
h) soldered a 220nF capacitor between pins #7 and #8 of IC10

Later on...
i) cut track on top side of PCB to pin #11 on IC31
i) cut track on top side of PCB to pin #11 on IC34
j) replaced 4k7 resistor with 470R
k) replaced 220nF caps with 220pF caps

[maniacminer]

In case it's any use, here's a BASIC test program for the M5000 and M3000.

m5000test.ssd
It's standalone, and doesn't depend in any way on the AMPLE ROM.

Code: Select all

   10  REM &30 (M5000) or &50 (M3000)
   20  D=&30
   30  FOR B=0 TO 15 STEP 2
   40  ?&FCFF=B+D
   50  FOR J=&FD00 TO &FDFF
   60  ?J=0
   70  NEXT
   80  NEXT
   90  REM SETUP WAVEFORM 0
  100  ?&FCFF=D
  110  FOR J=&FD00 TO &FD3F
  120  ?J=127
  130  NEXT
  140  FOR J=&FD40 TO &FD7F
  150  ?J=255
  160  NEXT
  170  REM SETUP WAVEFORM 12
  180  ?&FCFF=D+12
  190  FOR J=0 TO 127
  200  J?&FD00=SIN(2*PI*J/128)*127
  210  NEXT
  220  ?&FCFF=D+14
  230  ?&FD00=&00
  240  ?&FD10=&05
  250  ?&FD20=&00
  260  ?&FD50=&00 :REM Waveform 0 is a square wave
  270  ?&FD60=&00 :REM Min amplitude
  280  ?&FD70=&2D :REM Set bit 5 to modulate next channel
  290  ?&FD01=&BE
  300  ?&FD11=&6D
  310  ?&FD21=&01
  320  ?&FD51=&C0 :REM Waveform 12 is a sin wave
  330  ?&FD61=&70 :REM Normal volume
  340  ?&FD71=&0D :REM Clear bit 5 not modulate next channel
  350  ?&FD81=&BE
  360  ?&FD91=&6D
  370  ?&FDA1=&02
  380  ?&FDD1=&C0 :REM Waveform 12 is a sin wave
  390  ?&FDE1=&70 :REM Normal volume
  400  ?&FDF1=&0D :REM Clear bit 5 not modulate next channel

It should produce a two tone siren sound.

To test the M3000, change line 20 to:

Code: Select all

20 D=&50

Dave

Thanks for the program, it works on the M5000 and with the M3000 attached, however, the M3000 makes no sound.

Using the M500 schematic (thanks), I am working from the 1MHz bus inwards towards the logic and found my third culprit. IC14 pin 3 has a snowstorm on its output despite pins #1 & #2 being around 1V. So it is next for pulling.

[hoglet]

g) soldered a wire from pin #13 on IC32 to pin #1 of IC14

I think the track that previously pin #7 of IC32 (D5) to pin #1 IC14 to also needs to be cut.

(otherwise D5 and D6 are now shorted together)

i) cut track on top side of PCB to pin #11 on IC31

Is that a typo? You mean IC34 right?

Dave

[maniacminer]

g) soldered a wire from pin #13 on IC32 to pin #1 of IC14

I think the track that previously pin #7 of IC32 (D5) to pin #1 IC14 to also needs to be cut.

(otherwise D5 and D6 are now shorted together)

i) cut track on top side of PCB to pin #11 on IC31

Is that a typo? You mean IC34 right?

Dave

Ah, yes, that's a typo. Although I'd better check my work. Just checked, yes, it's a typo. I think using the phone to type these out is creating autocorrect errors

I haven't made the additional track cut. I'll do that now, that'll probably explain why IC14 was misbehaving!

[maniacminer]

OK, cut the track and a weird behaviour I was seeing on the M5000 with M3000 attached has gone away (yay!) and now the recordings sound normal (I think) can someone have a listen and tell me if this is normal sound please?

I've spent a hour probing around and after reading the datasheet for the weird DAC, discovered that the DAC is being sent signals, but it isn't generating any analogue signal. Further probing around the analogue bits and discover that the 12V positive reference is missing. This is provided by ZD1 and it has +15V on one side and zip on the other. I suspect this is the why I can't hear anything despite the vast amount of digital jiggling and wriggling going on inside the guts of the M500. I need to get a 12V zener and see if that makes the AM6070 sing again.

P.S. - I just realised I have saved the files out as mono (despite being two tracks) I have replaced them.

[dp11]

If you have +15v instead of +12v I think all that will happen is the sound will be distorted. I have a failed DAC in mine. From what read it is a common problem.

[maniacminer]

If you have +15v instead of +12v I think all that will happen is the sound will be distorted. I have a failed DAC in mine. From what read it is a common problem.

FingersX that isn't the problem

Those AM6070 DACs are unobtainium, an old laser mirror positioning system used them and a programmable electrophoresis controller, probably because they were cheap at the time, but got LOL-expensive quickly and then became the proverbial albatross-around-the-neck in the design. I've never seen anything else, apart from the LinnDrum that had this DAC fitted as a retail product (the percussion on that sounds soooooo much better than the M500), I guess this DAC was a cheap way of getting a 12 bit sample from 8 bit data (when data rates were poor and DACs were expensive) I wonder what else they were used in, that way we can keep an eye out for junk that could be salvaged for them.

[danielj]

I had fitted the 4k7/220nF without thinking about the values beyond what was on the Internet (dangerous, I know)

The values from eelco's post are 470R and 220pF (0.22nF) which make sense as a filter to eliminate noise on the clock.

My word - never spotted this when it came up. I'd done the 4k7 w/out capacitor from *way* back when this was documented. I strongly suspect it's been the source of all ridiculous intermittent grumpiness of my Master when the M5k was plugged in!

[hoglet]

My word - never spotted this when it came up. I'd done the 4k7 w/out capacitor from *way* back when this was documented. I strongly suspect it's been the source of all ridiculous intermittent grumpiness of my Master when the M5k was plugged in!

The 4K7 resistor shouldn't have any effect on the operation of the Master itself, but it may well make the Music 5000 unreliable.

Dave

[danielj]

My word - never spotted this when it came up. I'd done the 4k7 w/out capacitor from *way* back when this was documented. I strongly suspect it's been the source of all ridiculous intermittent grumpiness of my Master when the M5k was plugged in!

The 4K7 resistor shouldn't have any effect on the operation of the Master itself, but it may well make the Music 5000 unreliable.

No, I can't really see why it would, but I do know that the Master's "All my cmos values are 255 and you can't change them!" trick invariably happens when the M5k is plugged in and doesn't seem to ever happen when it's not

d.

[maniacminer]

I thought I'd carry on testing the working M500(0) and compare it against BeebEm. I noticed a few files that were hugely different. The recommended test Ample program "Syncron" sounds entirely different on my M5000 I trimmed off around 20s from the hardware file (it has ascending tones and they are missing)

M5000 hardware.mp3

M5000 hardware rendition of "Syncron"

(2.98 MiB) Downloaded 7 times

BeebEm M5000.mp3

M5000 software rendition of "Syncron"

(3.49 MiB) Downloaded 6 times

I don't suppose there is some kind of service manual for the M5000? There's so much going on inside, I can't know what's supposed to be happening and what isn't. Nothing is getting super hot when looking with the thermal camera, I have checked the signals coming from the Beeb to the M5000 and they are all arriving unmolested. Do I have to remove every single IC to find the problem? That's going to take hours

[dp11]

I've got some more fixes to apply to the emulation, but with my M500 being broken I'm a little stuck. I've fixed some of the issues I've found to the Pi1MHz project, but need to port them to beebem.

[hoglet]

I thought I'd carry on testing the working M500(0) and compare it against BeebEm.

Please don't assume what you hear from BeebEm is correct! There's been lots of improvements to Music 5000 emulation in the last year, and I don't think BeebEm benefits from that.

Do I have to remove every single IC to find the problem? That's going to take hours

I would hold off with the soldering iron until you are really sure there is a problem!

Dave