Limited TF536 Run - Sort of Sales (All gone)

[PaulJ_2.0]

A temporary fix.
Seems to be stable enough cooled.

IMG_20200117_223237.jpg

[alenppc]

A temporary fix.
Seems to be stable enough cooled.
IMG_20200117_223237.jpg

I was thinking of the same fix for me but it seems a bit ridiculous

[PaulJ_2.0]

It's still running Frontier Demo solid as a rock for about 45 mins now.

[terriblefire]

And yet the CPLD is cold?

Ok let me propose something....

The TF330, 536, 360 and 1260 all sample a slower clock and then recreate this clock in the CPLD to switch between fast speed and slow speed.

It seems to work perfectly on the AGA machines but is a little variable on the A500s. The clock rise time on the A500 is much worse... and much more variable between machines. This is why i put a schmitt trigger on the 520 rev 2.

Anyways one of 2 things we can try...

1. Patch in the schmitt trigger onto the CLK7M line. This would involve cutting the CLK7M line between the CPU socket and CPLD and putting one of the clock buffer chips from the 530/534. Not hard if you have a microscope.

and/or

2. On the TF360 we have a speed control CLI app that we can set the CPU speed but also the CPU clock phase relationship with the onboard 14Mhz clock. I cant make the speed change but i can port the clock phase control to the 536 quite easily. We could then adjust the phase in software to look for the phase that is stable. (if this is the issue).

[PaulJ_2.0]

Anyways one of 2 things we can try...

1. Patch in the schmitt trigger onto the CLK7M line. This would involve cutting the CLK7M line between the CPU socket and CPLD and putting one of the clock buffer chips from the 530/534. Not hard if you have a microscope.

and/or

2. On the TF360 we have a speed control CLI app that we can set the CPU speed but also the CPU clock phase relationship with the onboard 14Mhz clock. I cant make the speed change but i can port the clock phase control to the 536 quite easily. We could then adjust the phase in software to look for the phase that is stable. (if this is the issue).

Would you like me to send you my board?
I'm fairly certain it will play up for you, I've tried 5 different 500's and 6 different PSU's now.

[terriblefire]

Would you like me to send you my board?
I'm fairly certain it will play up for you, I've tried 5 different 500's and 6 different PSU's now.

Let me fire a few firmwares over to you first... if no progress by monday then yeah.. i may also try blowing hot air over mine... but i have a 1200 and a cd32 setup on the bench just now and moving back to the 030 would be painful.

[PaulJ_2.0]

Let me fire a few firmwares over to you first... if no progress by monday then yeah.. i may also try blowing hot air over mine... but i have a 1200 and a cd32 setup on the bench just now and moving back to the 030 would be painful.

Yeah sure.
Just to note, it crashes on anything but cool. If I remove the heatsink and put my finger on the top it crashes before I can detect it getting warm.

[terriblefire]

Ok i have no idea if this worked properly or not but here is a set of autocompiled firmwares that each have a different phase setup...

tf536_phases.zip

each file is named with its phase...

Maybe one of them is more stable? its the simplest thing to try for now.

[PaulJ_2.0]

Test results:

All behaved well in diagrom uncooled and detected and tested ram ok.
Original firmware needed cooling in diagrom for good results.

Best performing was 6 where it would crash uncooled using KS2.04 and stable cooled, but fail to boot to disk screen on KS3.1 and KS3.1.4

4 and 5 wouldnt boot on any KS

7-12 got progressively worse errors. mostly no disk screens.

[terriblefire]

Ok this makes sense.. By default we had 7.

The results are telling me that the CPU is happy running at slow speed but doesnt like switching.

Could you try a few different CPUs if you have them? Alen and I didnt send CPUs over the pond i dont think.