TF1260 New Firmware Alpha Released!

[terriblefire]

Did you try the fw i sent in PM? it has different startup timings.

I did, same behaviour unfortunately. (couldn't PM back)

@voscoboss If you are able to would you mind probing the reset line during boot and the 5V line. I have a hunch its stuck low and your 5V line is about 4.5v.

[markus0321]

Before getting too much deeper into things try the things Gadget suggested with PSU connectors.

I'm also seeing that on the friendly Polish website there is one guy having issues with CPUs at 100Mhz. Claims his CPUs will do 100Mhz because they work on Warp. Well you need to boost your CPU core voltage to get a stable overclock on some CPUs. I am told warp does this. There is plenty of things out there about how to calculate your CPU boost voltage for a stable overclock. Its not something TFs do. They all use some form of 3.3v regulator that is not a TF designed part. So you are on your own if you CPU is not stable at 100mhz by itself.

That said i think the BUSCPLD is getting too hot and this may be contributing. I would invite people to freeze spray this and see if it helps at all. It might help me with deveopment.

Unfortunately the individual having issues is just being macho about his CPU stock and is demonstrating that he really doesn't know very much about either electronics, amigas, 060 cpus or just how to accept help.

A. PCB No. 1 - BUSCPLD-10C and RAMCPLD -10C - Alliance memories

1. BUSCPLD freeze and 3.33V power supply - board unstable at 100MHz CPU clock
2.RAMCPLD freeze and 3.33V power supply - 100MHz stable board passes povray tests until I stop cooling RAMCPLD.
3.No chilled, but the power supply is reduced to 3.25V - the board is stable all the time, the povray comes to an end without errors.

B. PCB No. 2 - BUSCPLD-10C and RAMCPLD -6C - Samsung memories

1.No cool, but 3.33V power supply - the board is stable all the time povray comes to an end without errors.

C. PCB No. 3 - BUSCPLD-10C and RAMCPLD -6C - Alliance memories

1.No cool, but 3.33V power supply - the board is stable all the time povray comes to an end without errors.

It seems to me that RAMCPLD is a problem if it heats up and it is the -10C version even though it is generally cooler than BUSCPLD.

[terriblefire]

Thanks thats useful info. RAMCPLD is working very hard so i think its reasonable to expect to need 6ns to get to 100Mhz without cooling.

I might be able to get this more stable but it might also mean a 10-15% reduction in fastram bandwidth... discuss.

[markus0321]

Thanks thats useful info. RAMCPLD is working very hard so i think its reasonable to expect to need 6ns to get to 100Mhz without cooling.

I might be able to get this more stable but it might also mean a 10-15% reduction in fastram bandwidth... discuss.

The chips on all three boards had no heat sinks, and the boards were blown by one larger fan placed on the bare CPU
I raised the supply voltage to 3.5V on this third board and it still works stably. On the first one, the Amiga starts at 3.5V, but the system fails to start.

I think I have a good processor because at 100MHz it works on a supply voltage reduced to 3.25V. ;-)

Sorry. The third PCB at 3.5V also became unstable, but when I cooled the RAMCPLD with liquid butane, it works again even at 3.5v

[terriblefire]

If anyone gets a dead machine after startup check the 5V line and RESET lines. I have a suspicion that the low voltage detectors may be holding the system in reset. I think this is happening intermittently with @GadgetUK164 and i've asked him to confirm.

The basis for this hypothesis is that i am seeing this exact behaviour on the CD32 as I tried to port over the code to TF360. The machine is stuck in reset with a 5V line around 4.6V. I plan to replace the LVD and see what happens.

[hardgb]

First I want to thank you for the opportunity to own a great new 68060 accelerator. And I'm grateful to the @supaduper for sending this gem to me.
So I have a 060 rev5 and I haven't upgraded yet.
Does anyone here have a rev5 and have the new update done? If yes, what speed did you get and is it working steadily?

[supaduper]

First I want to thank you for the opportunity to own a great new 68060 accelerator. And I'm grateful to the @supaduper for sending this gem to me.
So I have a 060 rev5 and I haven't upgraded yet.
Does anyone here have a rev5 and have the new update done? If yes, what speed did you get and is it working steadily?

Hi hardgb, Glad you got your TF1260 safely over to Brazil :)

I have tried my rev5 and the best I get is 66mhz but it is rock steady at, who knows you may be able to squeeze a little more out of yours,
good luck my friend

[voscoboss]

I did, same behaviour unfortunately. (couldn't PM back)

@voscoboss If you are able to would you mind probing the reset line during boot and the 5V line. I have a hunch its stuck low and your 5V line is about 4.5v.

I only have a basic multimeter but I measured the lines at U49. +5 volt is +5.04, reset (assuming I'm correct that this is measured at U49) is 4.99V. I can see the voltage "building up" very shortly after boot, but wether that's a limitation of the multimeter or if it's actually "building up" I'm not sure.

PSU is the icomp CA-PSU. Connector on that should be fine. Power connector on the board I believe was NOS, but not entirely sure. I'll give the connector a clean as soon as I have contact cleaner. I'll also check if I have another PSU to try with, not sure if I have one though...

[voscoboss]

I did, same behaviour unfortunately. (couldn't PM back)

@voscoboss If you are able to would you mind probing the reset line during boot and the 5V line. I have a hunch its stuck low and your 5V line is about 4.5v.

Measured reset line at the MPU as well, it has 5V so it's not being held low for sure. Resetting the Amiga (using reset jumper), reset drops to 0 and goes back up to 5V, as it should. Also checked the previously outlined pins ont he CPLD's, they are definitely connected to each other and not shorted. I don't have another PSU to test with. Let me know if any further testing is desired.

[Higgy]

@terriblefire you said discuss.
I think we have to be careful of chasing figures. Like with engine tuning and HP. You might be able to get max HP but the car is not very drivable and the engine might not last as long.
I don't think we would moan in a small drop here and there for stability and longevity improvements.
One thing I am mindful of is I expect you and Chucky and maybe Supdadupa are testing with a motherboard on the bench.
Most TF1260 users will have them in a standard case and then even fewer will have fans installed. Winter ambient temps current are helping us, especially us who don't have house heating cranked right up :lol:
Are too many compromises being made to reach 100Mhz and beyond? For example can you still have increased Chipram and overclock friendly EHIDE with the cooldown trick of dropping to 14MHz? That seemed to top out at 75-80Mhz, but that is still very very fast compared to 50Mhz 030, which most of us will be using as the comparison.

Interesting on the LVD. When I went from TF328 to T330 and had some issues with non boot, I then just added a CD32 plug to a standard 4.5amp Amiga desktop PSU and that has been fine even with the TF360 running for 4hrs straight. I wondered when you would start tinkering and port the changes over to the TF360. This currently has GFX errors on just the TF360:
https://www.pouet.net/prod.php?which=1344

Thanks for your continued super work/support