Micro music box project. - Page 2 - UK Vintage Radio Repair and Restoration Discussion Forum

Slothie · 9th May 2023, 2:29 pm

When I started looking at this back in ... 2018 was it? I typed in the code from the article and got something that assembled, I noted all the errors in the listing and the inconsistencies with the hex dump and I think I resolved many of them. Unfortunately I am not sure what happened to my Source code... Hopefully its on one of the computers I still have access to!
Since then it has somewhat dropped down my to-do list since I was mostly curious as to how the auto-composed music would sound. It would appear that if there are "bugs" in the software we've been handed down we may never know for certain!

SiriusHardware · 9th May 2023, 3:39 pm

I did see you say (in #11) that you had ASMed the code but I couldn't remember you ever having posted it. It's a shame you don't still have it so we could compare them to see what you noticed, and I missed.

Slothie · 9th May 2023, 4:26 pm

Quote:

Originally Posted by SiriusHardware

I did see you say (in #11) that you had ASMed the code but I couldn't remember you ever having posted it. It's a shame you don't still have it so we could compare them to see what you noticed, and I missed.

I'll have a poke about on some of my USB drives because I do have loads of backup files on some of them.

SiriusHardware · 9th May 2023, 5:17 pm

I still haven't really worked out what the signal is which the external circuit feeds to the SC/MP but I would assume a long string of pseudo-random 1-bit noise. I don't have such a circuit handy but you might, if you ever built up that additional part of your OrtonView. I never did, unfortunately.

It's a shame nobody ever actually made the MAYBE gate which was featured in an April issue of Elektor a long, long time ago. That was the one where, when the input changed state, the output might or might not change state. It would have been perfect for this. It was 'expected to find applications in games and random number generators'.

SiriusHardware · 9th May 2023, 6:23 pm

After looking a bit more at the external PRBS circuit I see that it uses NADs as a starting point so that the random digital bits sampled by the software are always in an unambiguous state.

The hardware write up states that the output from the PRBS goes into Sense-B, but the circuit diagram shows the output of the PRBS generator going to two unidentified SC/MP pins, one of which must be the Sense-B pin, what's the other likely to be? SIN?

Phil__G · 9th May 2023, 6:29 pm

Oh dear.

As we said earlier, this is a mare!
You may well be right about the ST/LD thing G it just looked wrong from a glance but when I get a mo I will try to study it properly. I wish I understood the gobbledygook explanation, that would be a start! One-note tunes?

Re the two PRSG connections, in the code I see only sense-B tests, returning just a 1 or a 0 at random.
For note selection and then for rhythm it seems to only test the one last output bit of the shift-register, to make a yay/nay decision, theres no random number involved that I can see other than 1 or 0

Should be easy to recreate in software

(said he...)

Slothie · 9th May 2023, 6:34 pm

Quote:

Originally Posted by SiriusHardware

I still haven't really worked out what the signal is which the external circuit feeds to the SC/MP but I would assume a long string of pseudo-random 1-bit noise. I don't have such a circuit handy but you might, if you ever built up that additional part of your OrtonView. I never did, unfortunately.

It's a shame nobody ever actually made the MAYBE gate which was featured in an April issue of Elektor a long, long time ago. That was the one where, when the input changed state, the output might or might not change state. It would have been perfect for this. It was 'expected to find applications in games and random number generators'.

You could also use a reverse-biassed transistor junction feeding into a schmitt trigger gate to make random noise. I never got around to buying one of the CMOS gates I needed for the random no generator so its not been tested.

That MAYBE gate sounds interesting though

April edition, you say?

Phil__G · 9th May 2023, 6:42 pm

For testing, a 555 running at an unrelated freq would work, it seems to be only looking for a high or low, it doesnt (that I can see) 'shift in' a longer random number
If I can find a 2k x 8 nvram I could replace the 2716 in the Aitken SC/MP, with the code preloaded into the first 256 bytes...
its only a 24 pin socket though so it needs to be a 2k nvram, smallest I have is 8k 48Z08 so I'd have to buy one unless we faff about tying the top address lines... & the Aitken sc/mp is a bit fragile with its piggy-backed chips

oooh...

I could make one from a 6116, a DS1210 & a coincell....

SiriusHardware · 9th May 2023, 6:52 pm

I've just tried and failed again to understand the overall principle but as far as that LD instruction at OUTLOOP is concerned - and remember both the listing code and the hex dump code agree that the code there is C6 01, which is LD @1(P2), and not LD 1(P2) as per the article mnemonic.

If we look at it again

Code:

OUTLOOP	LD	@1,P2	; Increment the cycle counter
	LDI	80H
	XPAL 	P1 ;   Corrected from P2 (typo)
	LDI	-128

At first glance the LD seems wrong because whatever it reads into Acc is being overwritten immediately by the following LDI 80H, but I suggest that the actual purpose of the instruction at OUTLOOP is only to increment P2, as implied by the comment, without writing anything to anywhere (hence LD rather than ST).

Mark1960 · 9th May 2023, 6:58 pm

It also seems to be pointless to increment P2 as the low byte is exchanged with 80h and then acc is overwritten with -128.

SiriusHardware · 9th May 2023, 6:58 pm

Quote:

For testing, a 555 running at an unrelated freq would work, it seems to be only looking for a high or low,

On which TWO inputs of the SC/MP though? The output of the PRBS generator clearly goes to TWO pins on the SC/MP, one of which has to be Sense-B since it says so in the write up. There is no mention in the write-up of the OTHER pin which the PRBS gen output is shown going to.

SiriusHardware · 9th May 2023, 7:02 pm

Quote:

It also seems to be pointless to increment P2 as the low byte is exchanged with 80h and then acc is overwritten with -128.

Good spot, but that was actually just a typo on my part. The real code is

Code:

OUTLOOP	LD	@1,P2	; Increment the cycle counter
	LDI	80H
	XPAL 	P1
	LDI	-128

It's the low byte of P1, not P2, which is being loaded. So P2 remains useful for a bit longer.

Slothie · 9th May 2023, 7:32 pm

Quote:

Originally Posted by SiriusHardware

Quote:

For testing, a 555 running at an unrelated freq would work, it seems to be only looking for a high or low,

On which TWO inputs of the SC/MP though? The output of the PRBS generator clearly goes to TWO pins on the SC/MP, one of which has to be Sense-B since it says so in the write up. There is no mention in the write-up of the OTHER pin which the PRBS gen output is shown going to.

Its probably SIN, and they intended to use it to clock in longer streams of bits but didn't implement that. If I was making a random no generator that's what I'd do, but maybe when they came to write the software they realised they only had room in ROM to make single bit tests. Or the dude drawing the diagram just made a mistake. I don't recall seeing any tests of any other sense pin that would make sense (!) in the context of random numbers.

SiriusHardware · 9th May 2023, 7:36 pm

At xxB5, ( Label: SIO1: ) there is actually an SIO instruction. With absolutely no explanation in the comments. This is another difficulty, the sometimes quite verbose comments often don't start on the line of code which is doing what the comment says is being done.

Phil__G · 9th May 2023, 7:51 pm

I found a 48Z02 but its legs had been eaten by the black foam.
I plumbed it to a 24-pin header and it tests ok in the TL866pro but doesnt retain anything.
Poo.
But then this chip must have been in my stash for 35 years or so!
I've some 330µF smd tantalums, maybe a 6116 with a big tant across its power rails? It only has to retain for a few seconds whilst I transfer it...

Phil__G · 9th May 2023, 8:48 pm

Quote:

Originally Posted by SiriusHardware

At xxB5, ( Label: SIO1: ) there is actually an SIO instruction. With absolutely no explanation in the comments. This is another difficulty, the sometimes quite verbose comments often don't start on the line of code which is doing what the comment says is being done.

Your right again G, it shifts in the note length which is in EX when the SIO loop exits, then stores it one location above the pitch via P3
So the prsg feeds senseB and SIO. No twonkying on the picl then!

Slothie · 9th May 2023, 8:56 pm

Quote:

Originally Posted by SiriusHardware

At xxB5, ( Label: SIO1: ) there is actually an SIO instruction. With absolutely no explanation in the comments. This is another difficulty, the sometimes quite verbose comments often don't start on the line of code which is doing what the comment says is being done.

The article is one of the worst I have seen in ETI who normally did a pretty good job of it. I guess back then the editorial team didn't know enough about microprocessors to know if the author had done a good job.

Way back then the printing technology would probably have meant some typist was given a listing and told to typeset it so it could be pasted into the page, which is where all the errors crept in. Its disappointing there are errors in the hex listing because that is actually fairly easy to verify if you get a couple of people to type in the same thing and compare them.

SiriusHardware · 9th May 2023, 8:58 pm

Quote:

it shifts in the note length which is in EX when the SIO loop exits, then stores it one location above the pitch via P3

It sounds as though you already have a much better grasp of how this works than I do.

Quote:

So the prsg feeds senseB and SIO.

Strictly speaking, Sense-B and SIN (I often say 'SIO' when I really mean SIN or SOUT).

Did anyone manage to find an 'Errata' page in ANY of the issues from 1979? Hard to believe they weren't pressed for a correction.

SiriusHardware · 9th May 2023, 9:19 pm

I have to say I'm quite enjoying this almost archaeological process of code restoration. What we really need is a Rosetta Stone, an ancient, working version of the same code written in Z80 or 6502.

I can't shake the feeling that we will put a huge amount of effort into this, and it will (of course) eventually work, and we'll listen to it working for about 30 seconds and say:

"Right! What shall we do next?"

Phil__G · 9th May 2023, 9:37 pm

SIN yes, I've got SIO on the brain this week after all the picling

G, at our age, its all about "What shall we do next?" !!!

and Ian I totally agree this has to be the worst ETI project ever for mistakes! you wonder if any were successfully made...

I think we have to assume that Geoff P's MK14 variations are complete & correct, and that with these changes it does actually work on a MK14...

9th May 2023, 2:29 pm	#21
Slothie Octode Join Date: Apr 2018 Location: Newbury, Berkshire, UK. Posts: 1,287	Re: Micro music box project. When I started looking at this back in ... 2018 was it? I typed in the code from the article and got something that assembled, I noted all the errors in the listing and the inconsistencies with the hex dump and I think I resolved many of them. Unfortunately I am not sure what happened to my Source code... Hopefully its on one of the computers I still have access to! Since then it has somewhat dropped down my to-do list since I was mostly curious as to how the auto-composed music would sound. It would appear that if there are "bugs" in the software we've been handed down we may never know for certain!

9th May 2023, 3:39 pm	#22
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. I did see you say (in #11) that you had ASMed the code but I couldn't remember you ever having posted it. It's a shame you don't still have it so we could compare them to see what you noticed, and I missed.

9th May 2023, 5:17 pm	#24
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. I still haven't really worked out what the signal is which the external circuit feeds to the SC/MP but I would assume a long string of pseudo-random 1-bit noise. I don't have such a circuit handy but you might, if you ever built up that additional part of your OrtonView. I never did, unfortunately. It's a shame nobody ever actually made the MAYBE gate which was featured in an April issue of Elektor a long, long time ago. That was the one where, when the input changed state, the output might or might not change state. It would have been perfect for this. It was 'expected to find applications in games and random number generators'.

9th May 2023, 6:23 pm	#25
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. After looking a bit more at the external PRBS circuit I see that it uses NADs as a starting point so that the random digital bits sampled by the software are always in an unambiguous state. The hardware write up states that the output from the PRBS goes into Sense-B, but the circuit diagram shows the output of the PRBS generator going to two unidentified SC/MP pins, one of which must be the Sense-B pin, what's the other likely to be? SIN?

9th May 2023, 6:29 pm	#26
Phil__G Octode Join Date: Mar 2011 Location: North Yorkshire, UK. Posts: 1,115	Re: Micro music box project. Oh dear. As we said earlier, this is a mare! You may well be right about the ST/LD thing G it just looked wrong from a glance but when I get a mo I will try to study it properly. I wish I understood the gobbledygook explanation, that would be a start! One-note tunes? Re the two PRSG connections, in the code I see only sense-B tests, returning just a 1 or a 0 at random. For note selection and then for rhythm it seems to only test the one last output bit of the shift-register, to make a yay/nay decision, theres no random number involved that I can see other than 1 or 0 Should be easy to recreate in software (said he...) Last edited by Phil__G; 9th May 2023 at 6:38 pm.

9th May 2023, 6:42 pm	#28
Phil__G Octode Join Date: Mar 2011 Location: North Yorkshire, UK. Posts: 1,115	Re: Micro music box project. For testing, a 555 running at an unrelated freq would work, it seems to be only looking for a high or low, it doesnt (that I can see) 'shift in' a longer random number If I can find a 2k x 8 nvram I could replace the 2716 in the Aitken SC/MP, with the code preloaded into the first 256 bytes... its only a 24 pin socket though so it needs to be a 2k nvram, smallest I have is 8k 48Z08 so I'd have to buy one unless we faff about tying the top address lines... & the Aitken sc/mp is a bit fragile with its piggy-backed chips oooh... I could make one from a 6116, a DS1210 & a coincell.... Last edited by Phil__G; 9th May 2023 at 6:52 pm.

9th May 2023, 6:52 pm	#29
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. I've just tried and failed again to understand the overall principle but as far as that LD instruction at OUTLOOP is concerned - and remember both the listing code and the hex dump code agree that the code there is C6 01, which is LD @1(P2), and not LD 1(P2) as per the article mnemonic. If we look at it again Code: OUTLOOP LD @1,P2 ; Increment the cycle counter LDI 80H XPAL P1 ; Corrected from P2 (typo) LDI -128 At first glance the LD seems wrong because whatever it reads into Acc is being overwritten immediately by the following LDI 80H, but I suggest that the actual purpose of the instruction at OUTLOOP is only to increment P2, as implied by the comment, without writing anything to anywhere (hence LD rather than ST). Last edited by SiriusHardware; 9th May 2023 at 7:04 pm.

9th May 2023, 6:58 pm	#30
Mark1960 Octode Join Date: Mar 2020 Location: Kitchener, Ontario, Canada Posts: 1,294	Re: Micro music box project. It also seems to be pointless to increment P2 as the low byte is exchanged with 80h and then acc is overwritten with -128.

9th May 2023, 7:36 pm	#34
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. At xxB5, ( Label: SIO1: ) there is actually an SIO instruction. With absolutely no explanation in the comments. This is another difficulty, the sometimes quite verbose comments often don't start on the line of code which is doing what the comment says is being done.

9th May 2023, 7:51 pm	#35
Phil__G Octode Join Date: Mar 2011 Location: North Yorkshire, UK. Posts: 1,115	Re: Micro music box project. I found a 48Z02 but its legs had been eaten by the black foam. I plumbed it to a 24-pin header and it tests ok in the TL866pro but doesnt retain anything. Poo. But then this chip must have been in my stash for 35 years or so! I've some 330µF smd tantalums, maybe a 6116 with a big tant across its power rails? It only has to retain for a few seconds whilst I transfer it... Attached Thumbnails

9th May 2023, 9:19 pm	#39
SiriusHardware Dekatron Join Date: Aug 2011 Location: Newcastle, Tyne and Wear, UK. Posts: 11,558	Re: Micro music box project. I have to say I'm quite enjoying this almost archaeological process of code restoration. What we really need is a Rosetta Stone, an ancient, working version of the same code written in Z80 or 6502. I can't shake the feeling that we will put a huge amount of effort into this, and it will (of course) eventually work, and we'll listen to it working for about 30 seconds and say: "Right! What shall we do next?"

9th May 2023, 9:37 pm	#40
Phil__G Octode Join Date: Mar 2011 Location: North Yorkshire, UK. Posts: 1,115	Re: Micro music box project. SIN yes, I've got SIO on the brain this week after all the picling G, at our age, its all about "What shall we do next?" !!! and Ian I totally agree this has to be the worst ETI project ever for mistakes! you wonder if any were successfully made... I think we have to assume that Geoff P's MK14 variations are complete & correct, and that with these changes it does actually work on a MK14... Last edited by Phil__G; 9th May 2023 at 9:43 pm.