News:

Printed Amstrad Addict magazine announced, check it out here!

Main Menu
avatar_Duke

Amstrad CPC WiFi

Started by Duke, 07:36, 07 May 16

Previous topic - Next topic

0 Members and 10 Guests are viewing this topic.

00WReX

Quote from: CraigsBar on 23:58, 12 July 17
One question... What's your setup.... Ram expansion? C4cpc? M4 roms etc....

M4 board rev 2.1
The M4 is on FW 2.0.4beta5
All ROMs blank on the M4 (execpt for the M4 ROM itself).  No other expansions fitted.
QWERTY 464 Plus with an MC0122E revision mainboard.
AZERTY 6128 Plus with an MC0122E revision mainboard.
The standard BASIC/AMSDOS - Burnin' Rubber cartridge that comes with the Plus.

Intentionally had no expansions fitted and did not use the C4CPC during testing.

If you are having really bad results and have another ribbon/centronics connector to try, I would possibly give that a go a as a long shot.
I was starting to get some erratic behavior initially and swapped my cable. This solved most things.

I had previously not observed any issues with that cable/connector with an M4 prior to the 2.0.4 fw, but it must have a dodgy connection after being plugged and unplugged many times.

Cheers,
Shane
The CPC in Australia...
Awa - CPCWiki

RobertM

Hi Duke and others.

I have been watching this thread and thought I would throw my 2 cents worth in.

I have the earlier CPC6128 so I don't have specific points but I thought I would share some experience that may help with the "works sometimes - sometimes not" intermittent results.

I made a mass storage device consisting of a IDE adapter on the expansion port and used a Compact Flash to IDE converter to run the CF in IDE mode.

It didn't work for ages so I plugged in a real IDE Hard Disk and that would work sometimes. The problem with the hard drive was that it was sucking too much current and dropping the voltages. I have edge connectors that are bad for this problem at the best of times. I converted to pin headers and it would work more often but still failed at times. I soldered in a wire from the 5V to the main board to the 5V on the edge connector. It then worked all the time.

I then went back to the CF and it didn't work. The problem was that my choice of chips (LVTTL and the CF was CMOS) left only a 0.1V noise margin. Decoupling caps (0.1uF) don't really help with this because they don't store much energy so a sustained burst of activity will consume the stored energy and then the voltage will drop. Add a larger cap like 10uF - 100uF will help but only if there is enough voltage to start with.

The fix was to run the 3.3V chips at their limit of 3.6V

I also noticed that the CPC6128 I have has a jumper or marking for selecting CMOS or (I assume) NMOS. I suspect this is for the CPU.

I haven't really looked into the design of the M4 board so I hope this help with the intermittent results.

By the way I abandoned my IDE project as Dukes solution is far better :)

arnoldemu

Quote from: RobertM on 10:03, 13 July 17
I also noticed that the CPC6128 I have has a jumper or marking for selecting CMOS or (I assume) NMOS. I suspect this is for the CPU.
Oooh, pictures please. Or does it match one of the pictures from the 'mainboard revisions' page on the wiki?

My games. My Games
My website with coding examples: Unofficial Amstrad WWW Resource

Duke

@RobertM

Thanks for the input. I am fairly sure that is not the issue here, everything beside the CPC+ RMR2 rom emulation works for every CPC or the M4 would be an instant crash.
CPC+ RMR2 rom emulation, does not consume any more power than using ROMS in CPC mode.

A factor on CPC+ that could be the cause of the issues, that's the acid stream, that would mess up things if it glitches I think. Also I think there is different versions of the ASIC (something about the color shown when started without cartridge).
Hard for me to debug, when everything works on my CPC+, even with MotherX4 and XMEM connected at the same, all only powered from only the CPC+ monitor.

Anyway, those with problems could try to issue a |wifi,0 to power off the wifi module (biggest power draw on the board) before starting games with |ctr.  I doubt it changes anything, also results seem consistent for those with problems.

When I get some more time, I'll write a cart verification program, that say checksums the cart roms 1000 times or something to see if any read error occur.

chouimat

What is the shipping cost for shipping 1 board with edge connector to Canada?

RobertM

Quote from: arnoldemu on 14:00, 13 July 17
Oooh, pictures please. Or does it match one of the pictures from the 'mainboard revisions' page on the wiki?

In the pic here - http://www.cpcwiki.eu/imgs/f/f1/CPC6128_PCB_Top_%28Z70290_MC0020B%29.jpg

Look for Copyright and just above that is a resistor labeled  R154 and in a box beside it says "CMOS ONLY"

The schematic shows it as a 4k7 resistor as a pull up on the CPU /IORQ with the text "(HS-CMOS >USE)"

There are some chips that are specified as 74LSxx and some that are specified as 74HCxx. There are also some that are specified as 74LSxx OR 74HCxx.

The CPU is ONLY specified as a NMOS version and from memory the early CMOS version wasn't yet released in the early 80's

Duke

Quote from: chouimat on 16:48, 13 July 17
What is the shipping cost for shipping 1 board with edge connector to Canada?
3.5 euro (unregistered) shipping.

CraigsBar

Further testing. And the majority of the 128k cart images from the c4cpc sd card now work. 😉
IRC:  #Retro4All on Freenode

gerald

Quote from: Duke on 14:56, 13 July 17
When I get some more time, I'll write a cart verification program, that say checksums the cart roms 1000 times or something to see if any read error occur.
I've attached a 256k version of the C4CPC cartridge test cpr if you do not find the time  ;)
The 16 first rom slot except 0 are filled with pseudo random data and are checked :
- 1st quick pass, only the 4 1st byte of the page
- then all page in place
- then all page after copy with LDIR to RAM

The 2 last one are done endlessly

Duke

Quote from: gerald on 21:23, 13 July 17
I've attached a 256k version of the C4CPC cartridge test cpr if you do not find the time  ;)
The 16 first rom slot except 0 are filled with pseudo random data and are checked :
- 1st quick pass, only the 4 1st byte of the page
- then all page in place
- then all page after copy with LDIR to RAM

The 2 last one are done endlessly

Cool, thanks a lot gerald, very nice tester  8)

Here is my result, interesting to see what the others get.

Maniac

Quote from: CraigsBar on 21:00, 13 July 17
Further testing. And the majority of the 128k cart images from the c4cpc sd card now work. [emoji6]
May I ask what you changed please?

CraigsBar

Quote from: Maniac on 22:28, 13 July 17
May I ask what you changed please?
Only the cpr images I was using. I deleted the 512kb ones from gx4000.CO.UK and copied the full set of 128k cprs from my c4cpc sd card. These seems to work much better on my m4 (rev 2.4) board.

Sent from my ONEPLUS 3t using Tapatalk

IRC:  #Retro4All on Freenode

Maniac

Quote from: CraigsBar on 23:12, 13 July 17
Only the cpr images I was using. I deleted the 512kb ones from gx4000.CO.UK and copied the full set of 128k cprs from my c4cpc sd card. These seems to work much better on my m4 (rev 2.4) board.

Sent from my ONEPLUS 3t using Tapatalk
Ah, ok. Thanks for the info. I was wondering if you had cleared ROM slots from the M4. Unfortunately I've already done what you've just done and had the issues I've reported! I'll try to run the diag tool @Gerald has provided and see what it reports when I get a mo.

Duke

Quote from: Maniac on 23:22, 13 July 17
I was wondering if you had cleared ROM slots from the M4.
No need to do that, in latest beta I clear all ROM slots internally when switching to cart mode via |CTR.

Look forward to see your results from gerald's test.

Maniac

#1664

Quote from: Duke on 23:38, 13 July 17
No need to do that, in latest beta I clear all ROM slots internally when switching to cart mode via |CTR.

Look forward to see your results from gerald's test.


I get different results depending on if I have my C4CPC or original system cart inserted.


With the C4CPC after a fresh upload of the diag cart image:
[attach=2]
After a reboot and running of autoexec.bas selecting option 27 to start current image:
[attach=3]
After a few cycles of the test:
[attach=4]


With my original system cart inserted all seems good:
[attach=5]


With the above in mind I tried a couple of the cart images that hadn't been working. Pang now starts but I still can't play it as it crashes, Robocop 2 works again and Navy Seals now starts.


Things are sometimes better if I power-cycle the 6128 Plus (M4 is powered separately) and I'm guessing this is due to the way things are reset as the cart image seems to start straight-away!

gerald

It look like the ROMDIS from M4 is not long enough to fully mask the ROM access to the cartridge.
When the C4CPC is used, the faster logic compared to a regular EPROM have time to override the M4 data just before the Z80 fetch it.

Duke

Thanks @Maniac, very interesting
Quote from: gerald on 09:37, 14 July 17
It look like the ROMDIS from M4 is not long enough to fully mask the ROM access to the cartridge.
When the C4CPC is used, the faster logic compared to a regular EPROM have time to override the M4 data just before the Z80 fetch it.
Yes, or probably it is not early enough?  As it will keep ROMDIS until RD goes high. Will see what I can do to optimize the loop.
Thanks.

gerald

Quote from: Duke on 10:11, 14 July 17
Thanks @Maniac, very interestingYes, or probably it is not early enough?  As it will keep ROMDIS until RD goes high. Will see what I can do to optimize the loop.
Thanks.
I think you should extend it a bit after RDs goes high.
The OEn that goes to the cartridge is a mix of Z80 RDn, MREQn, memory configuration register and ROMDIS, so it is always late compared to the Z80 RDn.
According to the Z80 datasheet, the data is read on the clock edge that will make the RDn goes high. So if you're fast enough to release the ROMDIS there is a chance for the OEn to get active few ns.
The fact that failing pages change over heating time hints that we are on the edge (pun intended) for this timing. A nop in your handling function may do the trick.

00WReX

I also tried the ROMtest, with pretty much the same results as Maniac.
All good with the standard cart, but erratic with the C4CPC, and sometimes crashed spectacularly during testing.

Cheers,
Shane
The CPC in Australia...
Awa - CPCWiki

CraigsBar

#1669
A quick question.. When loading modern games from a directory on the m4 sd card (not tried with dsk files) does anyone else get screen corruption when loading title screens.

I should add that the games play perfectly so it seems to be only the graphic loading screens.
IRC:  #Retro4All on Freenode

CraigsBar

Quote from: CraigsBar on 21:00, 19 July 17
A quick question.. When loading modern games from a directory on the m4 sd card (not tried with dsk files) does anyone else get screen corruption when loading title screens.

Sent from my ONEPLUS 3t using Tapatalk
Examples attached from chibi akumas and vector vaults

Sent from my ONEPLUS 3t using Tapatalk

IRC:  #Retro4All on Freenode

00WReX

Hi Craig, sorry for crappy photo, but I had VectorVaults already on my M4 so just tried it again.

I do not see the loading corruption like you did.

[attachimg=1]

Cheers,
Shane
The CPC in Australia...
Awa - CPCWiki

CraigsBar

Was that loaded from a dsk file mounted or the files extracted to a folder?

Sent from my ONEPLUS 3t using Tapatalk

IRC:  #Retro4All on Freenode

00WReX

Quote from: CraigsBar on 12:31, 20 July 17
Was that loaded from a dsk file mounted or the files extracted to a folder?

Sent from my ONEPLUS 3t using Tapatalk

Oh sorry, yes it was loaded from a .dsk not extracted to a directory/folder on the M4.

Cheers,
Shane
The CPC in Australia...
Awa - CPCWiki

CraigsBar

Quote from: 00WReX on 12:55, 20 July 17
Oh sorry, yes it was loaded from a .dsk not extracted to a directory/folder on the M4.

Cheers,
Shane
Yeah, from dsk works fine here too. But chibi akumas is a RomDOS disc format so I don't think the m4 supports it mounted. So I extracted the files and like vector Vaults you can see the result. Both games play fine so it's not a biggy .

Sent from my ONEPLUS 3t using Tapatalk

IRC:  #Retro4All on Freenode

Powered by SMFPacks Menu Editor Mod