16mhz booster install in a mega 1

[RB80atari]

OK so now i know the dual tos works ok, i reconnected the decoder (booster disabled, pads bridged in 8mhz only mode) with the CE pin removed again from the dual tos board and ce wired up between them.

I forgot to plug in the gotek floppy drive and turned the mega on (with 1.04 selected) and it booted to desktop with no bombs, but with no desktop icons, the top drop down menus are empty, or rather very slow to show options and when they do it will start to corrupt and the screen will go white and it will reboot back to desktop.
Tos 2.06 just gives 4 bombs (no floppy)

With Gotek connected TOS 2.06 gives me the atari logo and 11 bombs and 1.04 gives me 20 bombs with some of the bombs having small colour corruption.

Well, unless anyone has any ideas of other things then i guess the booster isnt going to work for me.
BTW this mega does have all the mandatory fixes done to it.

Yep, tried that. Dual TOS board now boots into TOS 1.04 no problem.

Good news.

So some incompatibility with the motherboard or something wrong with the booster/decoder?

The decoder can only decode the address its given/sees. It's why the bus pullups needs to be changed as the bus is unstable, even more so at 16mhz. But I have never known (that I recall) problems after doing those mods with running in 8MHz. Generally it's a bad connection, or try swapping the IMP GLUE & MMU for something else and try again. But IMP I think was mostly having issues at 16MHz. But maybe your board is even worse than the ones I tested. Its difficult to know.

When the booster part didn't seem to be working yesterday, i did remove the 16 mhz wire between the booster and the shifter and also bridged the 8mhz pad on the booster (to disable it and just use it as a decoder) with CE still connected between them and 1.04/2.06 would not boot? both gave me bombs

I would suggest fluxing the CPU socket on the booster and reflowing it. Also wipe the CPU and booster pins with IPA. I can only assume you have a bad connection as in 8mhz nothing is boosted, it all runs at the stock 8mhz. There is no real difference from the GLUE doing decoding to the booster. The 16MHz wire isn't needed if running 8MHz.

[RB80atari]

So i thought id try removing the gotek and plugging in the old floppy drive, the old floppy drive is knackered and wont read disks, i think it needs a new drive belt.

With the decoder and dual tos still connected, tos 2.06 does now boot to the desktop, but it never does a ram test at boot.
Because the floppy drive isnt working i cant test any software.
It seems ok sat there idling on the desktop, can install desktop icons, go view desktop info and stuff like that. Then i noticed the blitter wasnt turned on, so i enable it and i get a couple of small bombs (2 i think) and it crashes.

Selecting Tos 1.04 also boots to the desktop, but if i try to view any drop down menu options, it starts to show them very slowly and then it crashes to a white screen.

So not much better, but its dfferent behaviour to when theres no drive connected and with the gotek which i thought was somewhat interesting.

Ive removed it all again, the dual tos and the decoder. Put the cpu back in and both tos 1.02 roms and it boots up fine, blitter enabled and no problems.

[exxos]

Did you check you don't have one of the "bad blitters" for starters as mentioned previously ? Did you wipe the pins etc with IPA ? All these things must be tried.

The only time I've seen a "slow blitter" was I think when DTACK was "missing" from the CPU. I think maybe the GLUE terminates the cycle eventually , so it still works but generally very slow.

I have have odd faults on the MEGA as well. One time /AS was missing from the MMU due to a bad via on the motherboard. It worked fine as a stock machine oddly that way.

Looks like @rubber_jonnie had via issues on his mega as well.
viewtopic.php?p=37205#p37205

[Badwolf]

The only time I've seen a "slow blitter" was I think when DTACK was "missing" from the CPU. I think maybe the GLUE terminates the cycle eventually , so it still works but generally very slow.

I have have odd faults on the MEGA as well. One time /AS was missing from the MMU due to a bad via on the motherboard. It worked fine as a stock machine oddly that way.

That's not a bad theory -- if, for example, not all the address pins are making their way. through to the Glue perhaps DTACK isn't be asserted somewhere (as the wrong address is detected) and bus cycles will terminate with BERR after 64 (I think) clock cycles.

Blitter is completely happy with this -- it doesn't care if it's bus error or dtack -- so you'd get about an 16x slow down.

The CPU *would* be unhappy with this of course, and things would go south very quickly if the appropriate failing range were accessed. But why only when the booster is installed? Perhaps there's a broken trace or dirty contact on one of the booster address line connections?

BW

[RB80atari]

I guess i could reflow the sockets, they look fine to me and i did check continuity from the bottom header pins to the top socket and all fine.
Not easy to bridge those solder pads to force 8mhz mode when its located under the IC socket :lol:
When i next have the board out i will test continuity from the cpu socket to nearby vias or components downstream justs to check any problems as i said the mega works fine without the booster/decoder installed, but no harm in checking. The CPU socket itself went in great, couldnt have gone any better.
Oh well, one of those things i guess... might try it in my mega 4 as well, see if it works in there, need to put the cpu in a socket first

boost 1.jpg

boost 2.jpg

boost 3.jpg

boost 4.jpg

[Badwolf]

i did check continuity from the bottom header pins to the top socket and all fine.

I suspect there are meant to be some pins that *don't* have connectivity. If they all do, that may be an issue in and of itself.

BW

[RB80atari]

Sorry yes your right, checked again and Pins 11 and 15 are not connected from the bottom decoder pin header to the corresponding pins for the CPU socket (on top) ...
Looking again the only pin that puzzled me was pin 49 from the (top) cpu socket which has a track connecting it to pin 29 (i think its pin 29) on the atmel chip, it also has continuity to the ajacent pin 49 on the decoder header despite having no visable tracks on either side of the board connecting them?
Header pin 49 on both sockets connect to pin 29 on the atmel chip. I dont know how many layers the board has so probably fine. I dont know.
Anyway, highlighted the pins in pic below, other than that it all checks out fine.

pin 49.jpg

i did check continuity from the bottom header pins to the top socket and all fine.

I suspect there are meant to be some pins that *don't* have connectivity. If they all do, that may be an issue in and of itself.

BW

[exxos]

PIn 49 is power and is connected on the bottom of the PCB.

Capture.PNG

If you want to pursue this. I suggest soldering a 68K socket ontop of the CPU, Plug that CPU into the MEGA. Then plug the booster into that new socket. This way you can try the raw ROM decoding logic without the CPU being a factor being plugged into the booster board.

[RB80atari]

OK cheers, yeah i might try that later.
I think ill try it in my Mega 4 rev 5 motherboard first, that one doesnt have any imp chips and from what ive read it seems to be the most stable of the mega motherboards? I need to install a cpu socket on it first
I would like to get it working if possible it just seems a waste otherwise.

PIn 49 is power and is connected on the bottom of the PCB.


Capture.PNG


If you want to pursue this. I suggest soldering a 68K socket ontop of the CPU, Plug that CPU into the MEGA. Then plug the booster into that new socket. This way you can try the raw ROM decoding logic without the CPU being a factor being plugged into the booster board.

[exxos]

Another thought, have you tried the original CPU in the booster running at 8mhz ?