LarryL - another Raven060 Adventure

[LarryL]

Just found out, that the Nessi Firmware allowing 1x clock and different clocks is not part of the release but was shared in the main thread
Will check that out first…

[PhilC]

Just found out, that the Nessi Firmware allowing 1x clock and different clocks is not part of the release but was shared in the main thread
Will check that out first…

Yes give that a go and post your results. I thoink you've probably got a level shifter or 68150 issue too though.

[Mikerochip]

I have taken Friday off, so, hopefully I'll catch up!

My biggest problem, atm, is RAM!
I've picked out some PSRAM that I'm *hopeful* will work. (Just not wanting to use the expensive current IC!!)

I've no real experience routing BGA stuff at all, and there's traces and vias going everywhere.
I'm going to start again (again) this weekend.

[LarryL]

Just found out, that the Nessi Firmware allowing 1x clock and different clocks is not part of the release but was shared in the main thread
Will check that out first…

Yes give that a go and post your results. I thoink you've probably got a level shifter or 68150 issue too though.

The new WIP firmware made the x1 clock setup work
Booted into a clean desktop

IMG_4634.jpeg (741.04 KiB) Viewed 167 times

So, the plan for the weekend is:
Carefully checking soldering, and
Clean the baby!

Enough for today…

[PhilC]

Oh that's good @LarryL, nice to see another Raven booting to desktop.

[agranlund]

Awesome! Congratulations!!
That must mean your machine is number 14 which has shown signs of life

Does it still say 8bit vgamem during boot, right before Starting Tos?
That may indicate some kind of issue as I would expect all ET4000 cards to be 16bit capable.

Perhaps, as @PhilC mentioned, the transcievers next to 68150, or some other kind of issue anywhere else on the 16/8 bit bus.
Though only running in 1x mode but failing in 2x is a bit of a new one to me

If you enter the monitor by pressing the NMI button, and then just send over "roms/srec/selftest.19" through the terminal. Will it run through ok? (might be worth a try but the test is really simple and underdeveloped so probably wont give much)


And also, similarly, press the NMI button to enter the monitor.
then runing:

Code: Select all

vga init
pw $820a0000 $aa55
pw $820a0000
pw $820a0000 $55aa
pw $820a0000

Code: Select all

vga init
pw $800a0000 $aa55
pw $800a0000
pw $800a0000 $55aa
pw $800a0000

Code: Select all

vga init
pb $800a0000 $55
pb $800a0000
pb $800a0001 $aa
pb $800a0001

Would be interesting to see the values that are read back and how they differ from the written ones.
Those commands are read/write to address. pw is word and pb is byte. the ones with three arguments are to write something and the ones without a last value argument are reads.

Screenshot 2025-11-06 at 20.33.00.png (133.88 KiB) Viewed 137 times

[agranlund]

actually, depending on how that vga write/read test goes it could be worth checking long access too, exactly as how it's being tested during startup

First through the 16bit aperture:
(32bit between cpu<->68150, 2x16bit between 68150<->videocard)

Code: Select all

vga init
pl $820a0000 $deadbeef
pl $820a0000

And if that didn't return $deadbeef then check if it does through an 8bit aperture:
(32bit between cpu<->68150, 4x8bit between 68150<->videocard)

Code: Select all

vga init
pl $800a0000 $deadbeef
pl $800a0000

(fishing for clues on which side of the 68150 the error is, on the 32bit bus or the 16/8 one)

[LarryL]

Thanks @agranlund
I am more than happy to be at that stage so quickly
I will do the proposed tests over the weekend - pretty busy at work tomorrow

[Atarian Computing]

@LarryL are you trying with the CPU set to 1x or 2x? I think for the firmware you might have it will need to be 2x and try a 40mhz oscillator at first is a good idea. What CPU are you using?

I am trying with 1x clock - I have read lately, that the actual release should support this.
But yes, this is a similar result like @Atarian Computing had
Issue is, that with 2x the Raven will not boot at all.

So I need some checks of soldering around the clock area - and maybe a good cleaning also

But anyhow, I am quite pleased that I got that far already

BTW: the CPU is supposed to be a rev6 71E41J (and the serial output confirms a rev6)

Wow!!! Great job!

Isn't this very similar to what @Atarian Computing was seeing in his build? Unfortunately if I understood correctly it just started working after a while, not sure if there was a definitive root cause found. Did you already clean everything carefully after the soldering?

Good job! Indeed, I had weird issues running at 2x mode at first. At one point it worked great, then just apparently bootloopping. Then everything just worked since then without me seemingly doing anything. It still kinda bugs me. But who knows, could have been a stray tiny solder ball or something.

[LarryL]

actually, depending on how that vga write/read test goes it could be worth checking long access too, exactly as how it's being tested during startup

First through the 16bit aperture:
(32bit between cpu<->68150, 2x16bit between 68150<->videocard)

Code: Select all

vga init
pl $820a0000 $deadbeef
pl $820a0000

And if that didn't return $deadbeef then check if it does through an 8bit aperture:
(32bit between cpu<->68150, 4x8bit between 68150<->videocard)

Code: Select all

vga init
pl $800a0000 $deadbeef
pl $800a0000

(fishing for clues on which side of the 68150 the error is, on the 32bit bus or the 16/8 one)

Ok, seems like a success now
16bit vgamem and system is booting with x1 and x2 (tried with 32 and 40MHz OSC)
Also the mentioned tests with the monitor were all successful now

IMG_4638.jpeg (732.59 KiB) Viewed 60 times

Gave the whole board (THT components only) a good reflow and cleaned the pins of the 68150 - some of this made the trick

Great adventure so far