Terrible Fire Accelerators

[elenachen]

Originally Posted by plasmab

I’m betting I’ve not decoupled the ram cpld enough. However the issue seems to get worse the more caps I put on. Very strange

[ Show youtube player ]

Dave made an interesting video regarding just that. The decoupling is mostly dependent on frequency. He also shows that in one of his test setups, the second added capacitor makes the bandwidth even worse.

[plasmab]

Originally Posted by elenachen

[ Show youtube player ]

Dave made an interesting video regarding just that. The decoupling is mostly dependent on frequency. He also shows that in one of his test setups, the second added capacitor makes the bandwidth even worse.

Seen. Taught at uni even. In theory I shouldn’t have an issue but I need to measure what’s happening.

Plus I always use X7R multilayer caps which do this pretty much for free. I think I’m just low on caps.

[keith]

@Plasmab just wondering with this @ different lower frequencies under 50Mhz does the behaviour change.

[8 Bit Dreams]

Originally Posted by plasmab

Seen. Taught at uni even. In theory I shouldn’t have an issue but I need to measure what’s happening.

Plus I always use X7R multilayer caps which do this pretty much for free. I think I’m just low on caps.

@ Stephen,
reWatched last stream and noticed that You have done Your research with FPU istalled..i'm suspecting the FPU can cause some errors, maybe You should try to remove it and look how TF534 works without it?

Another question, why on all retro accelerators the frequency generated from the crystal oscillator is divided to 2? Does it improoves stability in some way?

[plasmab]

Originally Posted by 8 Bit Dreams

@ Stephen,

reWatched last stream and noticed that You have done Your research with FPU istalled..i'm suspecting the FPU can cause some errors, maybe You should try to remove it and look how TF534 works without it?



Another question, why on all retro accelerators the frequency generated from the crystal oscillator is divided to 2? Does it improoves stability in some way?

I have the issue with or without the FPU. I have the issue anytime the cpu is running at any frequency above 12mhz. The ram cpld will not flash but will run.

The 040 needs a doubled Pclk frequency from the Bclk.... but it’s usually so you can time on half clock steps in the logic.

[8 Bit Dreams]

Originally Posted by plasmab

I have the issue with or without the FPU. I have the issue anytime the cpu is running at any frequency above 12mhz. .

maybe we need a ferrit on the clock line to keep clock signal clear?

[plasmab]

Originally Posted by 8 Bit Dreams

maybe we need a ferrit on the clock line to keep clock signal clear?

It could be many things.

I have not been putting JTAG terminator pull ups on. Going to try that. It’s not the clock itself. It’s when the cpu is running at 12mhz or above.. if it’s in a double bus error state it programs without issue. I suspect the bus buffer is causing the issue somehow but I am struggling to imagine why.

[keith]

Hi Just wonder is it not something strange making the other triggers not to be left floating...

[plasmab]

Originally Posted by keith

Hi Just wonder is it not something strange making the other triggers not to be left floating...

I don’t understand what this means.

[keith]

Hi Plasmab, with the above^ was just wondering on the the buffer ic that some thing might be affecting the latching or clock, but I might be way off here so please forgive me.

[plasmab]

Originally Posted by keith

Hi Plasmab, with the above^ was just wondering on the the buffer ic that some thing might be affecting the latching or clock, but I might be way off here so please forgive me.

Well yes. It’s the only major change between this board and the last revision so the buffer IC is going to be the obvious cause.

It doesn’t latch anything though. It’s just a buffer.

[elenachen]

Maybe the rise/fall time of the buffer ic is the culprit? To fast a edge, to slow a edge for the rest of the circuitry?

[plasmab]

Originally Posted by elenachen

Maybe the rise/fall time of the buffer ic is the culprit? To fast a edge, to slow a edge for the rest of the circuitry?

I think the issue is more along the lines of crosstalk between the JTAG lines and data bus.

[plasmab]

Ok i think i'm fighting about 3-4 issues at the same time.

1. DiagROM has an autoconfig bug in it with ZIII cards in the ZII space.
2. I have a jtag crosstalk issue.. not a problem now i know but it means i cannot reflash while the cpu is running.
3. I have a bus overlap issue occurring between fast ram and the buffer.

Basically can all be sorted in firmware i think.

[alenppc]

Originally Posted by 8 Bit Dreams

Since i started to use 1k resistor array's - i have noticed that almost all cards have problems at boot. LED light flashes for 2-3 seconds, then reboot follows, LED flashes, Amiga reboots (again) till finally Software Guru message appears, after pressing left mouse button - board loads fine and works without any issues. I can load WHDLoad games and software without any issues..till i do soft reboot. then that all repeats... Boards with 2k2 or 4k7 resistor array's are not affected with this issue.. i have to admit 1k resistence is a bit too low..
Did anyone had same issues here?

I haven't experienced any of that.

I have however just built 4 TF530s running at different speeds and using ISSI ram to experiment with. RC1 fixed the ISSI ram support, however I am noticing that these boards start to crash on Frontier after 15 mins running, which is odd because they don't crash on anything else so far. I am going to try to go back to 10k resistor networks to see what happens.

[plasmab]

Originally Posted by alenppc

I haven't experienced any of that.

I have however just built 4 TF530s running at different speeds and using ISSI ram to experiment with. RC1 fixed the ISSI ram support, however I am noticing that these boards start to crash on Frontier after 15 mins running, which is odd because they don't crash on anything else so far. I am going to try to go back to 10k resistor networks to see what happens.

A crash after ten mins might be CPU temp. Or at least that’s worth keeping an eye on.

EDIT.. another random thought if you have a reproducible crash is to try a pull-up on DTACK ... 2.2K should be perfect.

[8 Bit Dreams]

Originally Posted by plasmab

A crash after ten mins might be CPU temp. Or at least that’s worth keeping an eye on

Donno how much sun do You have in Canada currently, Alen, maybe it's indeed overheating?
We have almost summer here: 23° today with lot of sun..ideal weather for retrobright, lol
Have prepared black one on the trip to Italy:

[EzdineG]

Originally Posted by alenppc

I haven't experienced any of that.

I have however just built 4 TF530s running at different speeds and using ISSI ram to experiment with. RC1 fixed the ISSI ram support, however I am noticing that these boards start to crash on Frontier after 15 mins running, which is odd because they don't crash on anything else so far. I am going to try to go back to 10k resistor networks to see what happens.

Alen, I’ve seen the same behavior regardless of RAM. I’ve glued heat sinks to a few of them to see if it was indeed heat related. Sometimes the demo can loop as many as 6 times before locking up. I can’t otherwise crash these repeatably.

Does the release you’re using begin with a hopping bunny cracktro? Perhaps I need to find another version. I haven’t tried 10k resistors.

[alenppc]

Donno how much sun do You have in Canada currently, Alen, maybe it's indeed overheating?
We have almost summer here: 23° today with lot of sun

We had -21°C the other morning, now a bit warmer but still in the negative territory... 2 more weeks of winter and it should be finally done...

Originally Posted by EzdineG

Alen, I’ve seen the same behavior regardless of RAM. I’ve glued heat sinks to a few of them to see if it was indeed heat related. Sometimes the demo can loop as many as 6 times before locking up. I can’t otherwise crash these repeatably.

Does the release you’re using begin with a hopping bunny cracktro? Perhaps I need to find another version. I haven’t tried 10k resistors.

I am using an original disk, not a cracked version. The WHDLoad version locks up faster than the floppy version though. The CPU stays lukewarm so it's not overheating, it's something else, but being a parent severely restricts my free time for testing, so I haven't had a chance to investigate further yet.

[plasmab]

I’ve made some progress last night. Have family staying so haven’t streamed.

I got a spi rom chip, soldered it down and flashed it with a pi....



I have started compiling up flashrom for the tf53x series.