LarryL - another Raven060 Adventure

[LarryL]

Hi,

#3 is also working :D

IMG_4716.jpeg

After returning from my business trip I found the CPU in the post box.
Fired it up today and all is working fine
Only one of my remaining 68150FN40 is stable, the other two let the system freeze after a few minutes.
Anyway, one is enough :D

Tested with 32 and 48Mhz and x1 and x2

IMG_4718.jpeg

@agranlund I assume, this is S/N #16 now - right?

[agranlund]

@agranlund I assume, this is S/N #16 now - right?

That's incredible!
Yep, that should be #16 :)

[LarryL]

Hi,

Although suffering a bit from a flu (my wife caught Covid again, so maybe I have it, too…) I managed to get Raven #17 up and running :mrgreen:

I did not plan to build this one myself, but one of my friends „gave up“ after soldering the first few components, so I completed this board, too.

IMG_4742.jpeg

IMG_4741.jpeg

IMG_4740.jpeg

This one is having just a rev5 CPU and a 68150FN33 - so, I have to experiment a bit, finding the best clock setup…

But for this one we will try with a kind of special VGA card, having two ET4000/w32 Chips…
Lets see - doing my first tests with a standard AX, though.

Cheers
Michael

[agranlund]

I managed to get Raven #17 up and running :mrgreen:

Wow, Germany is now well ahead in number of Ravens :)

This one is having just a rev5 CPU and a 68150FN33 - so, I have to experiment a bit, finding the best clock setup…

Oh, I've implemented a rev1/5 cpu bug workaround in the lastest rom code (but I haven't yet rebuilt a new release package).

I'm not sure if Atari TOS or programs are affected much or at all, but according to Motorolas Errata document some certain sequence of instructions can produce the wrong result on Rev1 and 5 cpus. One recommended workaround for that bug was to disable load-store bypass so I'm doing that when detecting those revs.

(The way team Amiga handles this issue is more elegant and at some point I should copy that approach.
Instead of deciding based on cpu revision they apparently run a piece of code known to provoke the issue and checks if the result is correct or not)

[LarryL]

I managed to get Raven #17 up and running :mrgreen:

Wow, Germany is now well ahead in number of Ravens :)

This one is having just a rev5 CPU and a 68150FN33 - so, I have to experiment a bit, finding the best clock setup…

Oh, I've implemented a cpu bug workaround in the lastest rom code (but I haven't yet rebuilt a new release package).

I'm not sure if Atari TOS or programs are affected much or at all, but according to Motorolas Errata document some certain sequence of instructions can produce the wrong result on Rev1 and 5 cpus. One recommended workaround for that bug was to disable load-store bypass so I'm doing that when detecting those revs.

(The way team Amiga handles this issue is more elegant and at some point I should copy that approach.
Instead of deciding based on cpu revision they apparently run a piece of code known to provoke the issue and checks if the result is correct or not)

Oh, didn’t know that this is a competition - thought it’s pure fun :mrgreen:

Cool idea, to implement a workaround for such bugs!

Few observations during my first tests with this board:
Although there is a 68150FN33 fitted, it only works with 40MHz OSC, not with 32MHz… strange…
With 32MHz I have always freezes during the first three Gembench tests
Regardless, if I use x1 or x2 clock
With 40MHz the system is stable running Gembench for a while
Interestingly, the results are pretty much the same in x1 and x2 mode… also strange…
…and with 25MHz the system is not starting at all - same like with the other boards
I haven’t tried 48MHz yet…

[LarryL]

Hi,

Time for a pre-Christmas update :D

After identifying my stupid mistake, populating the wrong version of the CY2308 (the one NOT doubling the frequency), I have now successfully updated 3 of the 4 Raven Boards with the right Chip (-2 version). The 4th board is still traveling with a courier towards my place for some time now - have my fingers crossed that it did not got lost…)

Much better now :mrgreen:
The x2 jumpers now really lead to a double base frequency - as expected and desired.

The three boards are working. Had some stability issues while testing different frequencies with x1 and x2 settings.
Using the 251029 Nessie Firmware.
Some combinations work better, others worse - 32MHz base clock is definitely not working at all. I don’t mind :D

Had to test with different 68150 (FN33 and FN40), but found enough chips working.
It was also important to have a good cleaning of the pins of these old chips - some were really looking a bid corroded…

During my testing I struggled a lot with intermediate freezes and locks of the system.
Mainly during VDI tests in Gembench. CPU tests in Gembench and running Coremark68k always went well.
Also sometimes the system would not start (black screen), although the serial output went through until „starting TOS“
And sometimes the serial log stopped at the line showing the VGA BIOS message, and even sometimes the log only showed „8 bit VGA memory“ instead of 16Bit (next boot it was OK again)

So I had a closer look on the ET4000AX used for testing (which seemed working OK for me)
This specific card with its SC11486CN-66 RAMDAC (checking the Nova website from Idek - https://silicon-heaven.org/atari/nova/ ... gaST/STGA/) would usually require the T6 driver package
But the Raven only offers T0, T3, T4 and T8.
I used T4, since it seemed to work
But maybe this combination did cause the instability? Not sure at all

In the end, I changed to a different ET4000AX which requires the T0 driver and now my system is much more stable.
It now works pretty flawlessly with 25, 40, 48 AND 50MHz

@agranlund is there a special reason, why the T6 driver is missing? Not sure if the root cause of my problems with the other card is based on this - maybe the card simply has a defect - but, I am curious :)

Now it is time to assemble everything back into the case and get network and sound card running - and MINT installed

Cheers
Michael

[agranlund]

@agranlund is there a special reason, why the T6 driver is missing? Not sure if the root cause of my problems with the other card is based on this - maybe the card simply has a defect - but, I am curious :)

Nope, not any conscious decision on my part.
But I see there was no T6 in the list here so that would be why :)
https://silicon-heaven.org/atari/nova/TT030/ET4000/

It looks like there is a T6 for ST but I don't have access to my code or github so I'm not sure if I made the necessary stuff for the non-TT drivers to work.

(bootrom pmmu-maps the range where those et4k/mach drivers expect to find the card on a standard Atari with isa adapter -- which can differ depending on card/driver and for sure different between ST and TT versions of them)

[LarryL]

Nope, not any conscious decision on my part.
But I see there was no T6 in the list here so that would be why :)
https://silicon-heaven.org/atari/nova/TT030/ET4000/

Ah, sure - the TT drivers are the basis, of course…
I just looked for the RAMDAC numbers
I have no idea, if the drivers provide any other specifics for the different RaMDACs that can cause instabilities if the wrong driver is used…

Anyhow, seems I have a card+driver that works :D

[agranlund]

AND 50MHz

Mine did too, up until a couple of days ago. Before i decided to swap around between oscillators for testing, and now i need a new 50.

Pro tip: don’t attach the oscillator in the wrong orientation — that’ll murder it instantly :)

(Too bad i slipped up with the 50 and not one of the others)

[LarryL]

Some recent pictures of the Raven sitting in its new home…

IMG_4751.jpeg

IMG_4750.jpeg

IMG_4749.jpeg