phantom bus accelerator & clock patch

[exxos]

I looked into my clock patch again on my second Falcon and I think I know why the phantom patch needs the second buffer chip for the SDMA.

Adding loading resistors on the SDMA side only reduces the voltages and issue. A couple variants seem to divide the voltage in fact. IIRC the phantom used a really slow buffer to "cover up" the issues. Adding a loading capacitor seems to make things worse, when it should do the same as the resistor and smooth out the signal.. Basically the line is oscillating on some falcons it seems. heavy load resistors on the SDMA clock just halves the voltage and the issue. Slow buffer chips get around it as they are to slow to react. I don't like either solution. Both put a lot of mA into the long trace in the area of 100mA! I'm surprised nobody mentioned interference issues with so much RF going though a long 16-25MHz track.

So I've done a new clock buffer design, but I don't have any bus speeders to test it.. can anyone help try it out once I get a couple built ?

[Steve]

I would happily test it in my phantom Falcon just depending on the level of modification required. I currently have my 'home made' phantom clock patch, so it has the cut track at the sdma etc etc. If I had to repatch the trace and stuff I might not go down that route since it would require a bit too much messing, especially since my falcon is running like a dream already :)

Are you saying that your new clock patch will have added stability over the Phantom one?

Also have you ever had a conversation with the guy who made the Phantom patch? I think his name is ctirad I think, he's on atari forum. You guys would probably learn from each other.

[exxos]

I would happily test it in my phantom Falcon just depending on the level of modification required. I currently have my 'home made' phantom clock patch, so it has the cut track at the sdma etc etc. If I had to repatch the trace and stuff I might not go down that route since it would require a bit too much messing, especially since my falcon is running like a dream already :)

Hmm.. So was the track cut both ends and feed with a new wire ?

Also assume your patch had the "second" buffer chip ? Really if the track is replaced, there shouldn't be any need for the second buffer. Something like 68R in series with the line with the normal clock patch should be enough.

Are you saying that your new clock patch will have added stability over the Phantom one?

No way to know until its tested.

[Steve]

I'll post some pictures when I get home, just out to lunch at the mo :)

[ctirad]

Hi, Steve pointed me to this forum. I'm the Phantom accelerator author. I'm ready to answer any questions.

BTW, I already have written some historical and technical info here, but the forum kicked me to the login screen when I tried to send it and everything was lost :(

[exxos]

Hi, Steve pointed me to this forum. I'm the Phantom accelerator author. I'm ready to answer any questions.

Cool! Maybe you can clear up the slow buffer used for the sdma clock ?

BTW, I already have written some historical and technical info here, but the forum kicked me to the login screen when I tried to send it and everything was lost :(

oh no :(

I don't know why it would do that, its not normal for it to log anyone out like that.

[exxos]

I'll post some pictures when I get home, just out to lunch at the mo :)

Maybe you can solder the SDMA wire to a pin on the main buffer chip (bypass the second one) and use a 100R in series with the SDMA clock ?

[ctirad]

Cool! Maybe you can clear up the slow buffer used for the sdma clock ?

The original clock patch design (a single "slow" SN7404N with separated SDMA feed, terminated by 150pF capacitor) was not invented by me, actually. It was a 1:1 copy of the clockpatch of BSS Nemesis accelerator. I decided to use this circuit because it was the only one proven to solve all SDMA issues at any BUS speed. And it still is.

However, with later batches of the SN7404N i bought for the further production, some people reported a stability issues, especially in Cubase audio. In short, the new clockpatch chips (same type, same manufacturer) was the cause. I confirmed it by retrofitting one of the chips into my old Nemesised Falcon, that started to have the same stability issues immediately.

So i decided to design a new version of the clock patch, where the good old SN7404N (or equivalent) is used for the SDMA buffer only and the CPU/FPU/Expansion buffering is done by additional 74F04. This clock patch is recomended for any Falcon, CT6x(e) or even unaccelerated as the ultimate solution.

[exxos]

So i decided to design a new version of the clock patch, where the good old SN7404N (or equivalent) is used for the SDMA buffer only and the CPU/FPU/Expansion buffering is done by additional 74F04. This clock patch is recomended for any Falcon, CT6x(e) or even unaccelerated as the ultimate solution.

That is something I don't understand either, the 74XX series maxes out around 12MHz I think. So overclocking that chip to 25MHz would basically mean huge voltage drop in its output.. Which also seems to fit with why some clock patches use a voltage divider to drop the voltage..

Maybe you could try my idea as it is simple on your machine ?

[ctirad]

That is something I don't understand either, the 74XX series maxes out around 12MHz I think.

The SN7404N is rated up to 35MHz. The "F" is good maybe up to couple of hundreds of MHz.

Maybe you could try my idea as it is simple on your machine ?

I cannot. Mainly because I no more own a Falcon :(
Anyway, about 20 years ago I did a lot of tests that involved various 74xx types (LS/HC/HCT/F, even the socialistic "Tesla" MHB ones) and circuits including different kinds of termination. I didn't found anything, that would work at least the same as the discussed one.