BBC ports to the Atom

discussion of games, software, hardware & emulators relating to the Acorn Atom
User avatar
hoglet
Posts: 6605
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC ports to the Atom

Postby hoglet » Fri Jun 02, 2017 9:32 am

Hi Roland,

My Pi Zero get to about 39°C after 10 minutes of idling in the fast 6502 Co Processor. This is just sitting in the open on the workbench, and the ambient temperature is about 21°C.

This is the Cobra release, and the GPU is not very active.

If you are running the older atom_vdu branch, and making use of the GPU then an extra 10°C is probably to be expected.

Dave

User avatar
roland
Posts: 2808
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: BBC ports to the Atom

Postby roland » Sat Jun 10, 2017 1:31 pm

Question: I connect the Matchbox Copro to the tube port of this board. I can start the TUBE program and I get a message like '6809 TUBE' or 'ACORN TUBE 04MHZ 654C102 CO-PRO' but when I load HIBASIC or BASIC2 it either hangs or returns with a * prompt.

Where should I look to troubleshoot this?
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 6605
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC ports to the Atom

Postby hoglet » Sat Jun 10, 2017 1:42 pm

roland wrote:Question: I connect the Matchbox Copro to the tube port of this board. I can start the TUBE program and I get a message like '6809 TUBE' or 'ACORN TUBE 04MHZ 654C102 CO-PRO' but when I load HIBASIC or BASIC2 it either hangs or returns with a * prompt.

Where should I look to troubleshoot this?

Hmm, this can be tricky.

Did you include the clock selection jumper that was present in Atom Tube? It might be worth quickly trying the other setting.

Failing that, try to narrow down the problem by trying simple things first.

1. Does the boot message ever get corrupted?

2. Do simple * COMMANDS work reliably (like *CAT several times)

3. Does pressing Escape work reliably (this tests the interrupt)

4. Does saving / loading small files works?

*SAVE TEST F800 F810

The contents should be:
A2 00 BD 00 FF 9D 00 FF

The try reloading it to RAM, and then saving it again

*LOAD TEST 2000
*SAVE TEST2 2000 2010

Dave

Edit: Another useful debugging technique is to use the TUBED3 host, which should give lots of additional logging during a data transfer.

User avatar
hoglet
Posts: 6605
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC ports to the Atom

Postby hoglet » Sat Jun 10, 2017 2:43 pm

Roland,

One more slightly random thought....

We've just had a rather nasty problem with the Matchbox Co Pro on the new AP5 that's in development. The main symptom was that if Data Centre was connected to the 1MHz bus, the Matchbox Co Pro suddenly became very unreliable, just like you are seeing.

It turned out this was the problem:
IMG_0952.JPG

The top trace is nRST on the matchbox, the bottom trace is the Phi2 clock.

The nRST glitch drops down to about 1.6v. It's caused by crosstalk on the tube cable from the data lines switching at the same time. But Data Centre made it much worse, because it was dragging the nRST line down from 4.6v to about 3.6V. (It must be using 3.3V parts with clamping diodes, and driving these directly with nRST).

Anyway, I'm mentioning this because it surprised me how sensitive the Matchbox was to glitches on the nRST line.

It's just possible you are having a similar issue on the Atom, especially a New Atom, because nRST is driven from a 3.3V CPLD.

What's the "high" voltage on nRST?

If it is this, two things you could try to help:
- plug the Matchbox directly in to the Tube connector - this should eliminate most of the crosstalk
- solder a 100pF capacitor between pin 10 (nRST) and pin 9 (GND) on the Matchbox itself

Dave

User avatar
roland
Posts: 2808
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: BBC ports to the Atom

Postby roland » Sat Jun 10, 2017 8:05 pm

hoglet wrote:Did you include the clock selection jumper that was present in Atom Tube? It might be worth quickly trying the other setting.


No, I didn't include this pin because I never noticed any differences on the Atom Tube when I changed the jumper.

hoglet wrote:1. Does the boot message ever get corrupted?
2. Do simple * COMMANDS work reliably (like *CAT several times)


The boot message is always correct en *CAT works always perfectly.

hoglet wrote:3. Does pressing Escape work reliably (this tests the interrupt)
4. Does saving / loading small files works?
What's the "high" voltage on nRST?


I'll check these in the next week.

hoglet wrote:If it is this, two things you could try to help:
- plug the Matchbox directly in to the Tube connector - this should eliminate most of the crosstalk
- solder a 100pF capacitor between pin 10 (nRST) and pin 9 (GND) on the Matchbox itself

The Matchbox is plugged directly in to the Tube connector and I can try the capacitor.

If these options don't work I can:
  1. Find a scope and check the signals
  2. Place buffers between the board and the Matchbox
  3. Re-install the Matchbox back into the Atom :)
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
roland
Posts: 2808
Joined: Thu Aug 29, 2013 8:29 pm
Location: Born (NL)
Contact:

Re: BBC ports to the Atom

Postby roland » Fri Jun 16, 2017 6:48 pm

hoglet wrote:3. Does pressing Escape work reliably (this tests the interrupt)
4. Does saving / loading small files works?

*SAVE TEST F800 F810

The contents should be:
A2 00 BD 00 FF 9D 00 FF

The try reloading it to RAM, and then saving it again

*LOAD TEST 2000
*SAVE TEST2 2000 2010


Pressing Escape works reliably. Every time I press the Esc key the message ESCAPE appears on the screen.

Loading and save gives no error but it saves the Atom memory, not the co-pro's memory. The contents don't match with your data.

The voltage on the nRST line is about 3.77V. I don't think this signal is very noisy because I the I/O board leads the nRST from the CPLD into a buffer so the nRST to the TUBE is buffered on the board. It's just a thought because I don't have an oscilloscope to check this.

Soldering a 100nF capacitor makes things worse, this appears after loading BASIC2:

IMG_2127.JPG


So I removed the capacitor but the garbage remained. However, after a few breaks it disappeared. It shows that BASIC2 is loaded into the Atoms memory instead of the TUBE. I can reproduce this image by loading BASIC2 at #8000 in normal Atom mode.
256K + 6502 Inside
MAN WOMAN :shock:

User avatar
hoglet
Posts: 6605
Joined: Sat Oct 13, 2012 6:21 pm
Location: Bristol

Re: BBC ports to the Atom

Postby hoglet » Fri Jun 16, 2017 7:31 pm

roland wrote:Loading and save gives no error but it saves the Atom memory, not the co-pro's memory. The contents don't match with your data.

Sounds like you mighty have the wrong AtoMMC ROM.

What version do you have?

You need version 3.00.
roland wrote:Soldering a 100nF capacitor makes things worse, this appears after loading BASIC2.

100nF ???? I hope I said 100pF :shock:

Dave


Return to “acorn atom”

Who is online

Users browsing this forum: Google [Bot] and 2 guests