Uzebox & SRAM

Discuss anything not related to the current Uzebox design like successors and other open source gaming hardware
User avatar
Roukan
Posts: 113
Joined: Sat Oct 27, 2012 7:50 pm
Location: Lancashire / England

Uzebox & SRAM

Post by Roukan » Wed Jan 09, 2013 8:47 pm

Hi All

One for the experts out there:

Would it possible to make an Uzebox with more SRAM to give more flexibilty where video modes are concerned?

IE: A different Atmel microcontroller (yet compatible) with more on-board SRAM from what I could see 16/32 might be the maximum.

Also what about external RAM - too slow?

An Uzebox computer springs to mind with Flashed Kernel / Interpreter with source stored in external RAM (can be saved/loaded to/from SD card) Internal SRAM used for stack/variables/VRAM.

Kind of take 1 x Uzebox + 1 http://en.wikipedia.org/wiki/Tatung_Einstein (my favourite computer I ever owned) + a box of Viagra pills and see what offspring they produce :o - lol.


Cheers


Roukan / Jim

User avatar
uze6666
Site Admin
Posts: 4488
Joined: Tue Aug 12, 2008 9:13 pm
Location: Montreal, Canada
Contact:

Re: Uzebox & SRAM

Post by uze6666 » Wed Jan 09, 2013 11:47 pm

There's currently no easy way to upgrade ram that is 100% backward compatible. The atmega1284 has 16k ram but alas, it can't be overclocked reliably. External ram would be too slow for gaming needs. Using an xmega would be the closest thing but would not be binary compatible. It would also entail a complete redesign beign a 3.3v part. And it's smt so not much hobyist friendly as our current good ol' 40 dip.

AtomicZombie
Posts: 23
Joined: Wed Jan 09, 2013 5:16 pm

Re: Uzebox & SRAM

Post by AtomicZombie » Thu Jan 10, 2013 12:12 am

uze6666 wrote:There's currently no easy way to upgrade ram that is 100% backward compatible. The atmega1284 has 16k ram but alas, it can't be overclocked reliably. External ram would be too slow for gaming needs. Using an xmega would be the closest thing but would not be binary compatible. It would also entail a complete redesign beign a 3.3v part. And it's smt so not much hobyist friendly as our current good ol' 40 dip.
I'm curious what kind of results you had with the 1284 running at 28.626MHz. I had samples sent from Atmel when the chips was fresh out of the oven and found it a bit quirky at that speed. Later orders from Digikey ran fine for me, even up to 32Mhz. There seemed to be some real quirkiness when setting aref for some reason that would cause random resets on the first 1284 I tried, even when not using the ADC.

The Xmega is an overclocker's dream though. 75MHz is fine, and I have not tried further yet.

Cheers,
Brad

User avatar
uze6666
Site Admin
Posts: 4488
Joined: Tue Aug 12, 2008 9:13 pm
Location: Montreal, Canada
Contact:

Re: Uzebox & SRAM

Post by uze6666 » Thu Jan 10, 2013 2:18 am

Well, results were not that great. That was back in 2010 so perhaps it was something with the early batches? I tested about 10 chips from 2 orders and almost all them were unstable. I tested both 1284P and the more rare 1284. About 60% were running code ok but all exhibited various glitches in the video output. Raising the voltage at 6V somewhat lowered the occurrences of the problem but I could not see myself distributing a design that stresses the chip near its breaking point.

Can you let me know the exact digikey P/N you ordered that was working? I'll give it another try...

AtomicZombie
Posts: 23
Joined: Wed Jan 09, 2013 5:16 pm

Re: Uzebox & SRAM

Post by AtomicZombie » Thu Jan 10, 2013 2:57 am

uze6666 wrote:Well, results were not that great. That was back in 2010 so perhaps it was something with the early batches? I tested about 10 chips from 2 orders and almost all them were unstable. I tested both 1284P and the more rare 1284. About 60% were running code ok but all exhibited various glitches in the video output. Raising the voltage at 6V somewhat lowered the occurrences of the problem but I could not see myself distributing a design that stresses the chip near its breaking point.

Can you let me know the exact digikey P/N you ordered that was working? I'll give it another try...
Aha, I see the problem... ceramic crystal! I could never get the 1284 stable with anything but a true clock oscillator module. You have the 1284-PU, and that is the one I found worked well.
Even on the XMega, I find ceramic introduces a bit of error as compared to either the internal osc or a real external osc.
After seeing this happen, I decided to only run clock modules from now on, and have not had any issues.

It would be great to see your UZEBox get another life cycle with the 1284. I wonder how it would run if you just fed the output from a 57.272MHz clock module into the xtal1 pin?
Digikey has that great service where you just key in the frequency at order time and they will program the clock to whatever you like. I have NTSC clocks in just about every conceivable division now.

I have a real suspicion that that's all it will take to get it running, and would explain why at 6v you seen more success. That probably squared off the signal from the RC a bit more.

Brad

User avatar
uze6666
Site Admin
Posts: 4488
Joined: Tue Aug 12, 2008 9:13 pm
Location: Montreal, Canada
Contact:

Re: Uzebox & SRAM

Post by uze6666 » Thu Jan 10, 2013 3:27 am

Oh, the Uzebox is using a real crystal oscillator, not a ceramic resonator. But you say you had better results with an oscillator module eh? Worth trying for sure! The thing that sucks a simple pop-in replacement chip would have been awesome. No new circuit or hack soldering on existing boards out there. But I'll give it a spin for sure. Worst case one could always do a dip-40 footprint module with a VFQFN 1284 and the oscillator module.

I'll order parts and report on it. :)

AtomicZombie
Posts: 23
Joined: Wed Jan 09, 2013 5:16 pm

Re: Uzebox & SRAM

Post by AtomicZombie » Thu Jan 10, 2013 4:11 am

Yes, I meant to lump resonator and crystal into the same boat. Neither worked well for my overclocking experiments.
if you can't find a 28.626 clock module let me know and I will ship one off to you.
Since we are both in Ontario, it shouldn't cost more than a few thousand dollars to send an envelope across the province... ack!
Look forward to hearing the results.

Brad

User avatar
DaveyPocket
Posts: 378
Joined: Sun Sep 14, 2008 8:33 pm
Contact:

Re: Uzebox & SRAM

Post by DaveyPocket » Thu Jan 10, 2013 6:02 am

I had done several experiments with the 1284 as well, I did notice differences if using different types of oscillating components. One was stable at 5.7 volts and the other 6 volts. I figured the only solution was to increase voltage and find an oscillator type and fuse setting combination that wouldn't screw everything else up!

AtomicZombie
Posts: 23
Joined: Wed Jan 09, 2013 5:16 pm

Re: Uzebox & SRAM

Post by AtomicZombie » Thu Jan 10, 2013 6:14 am

When I was running the 644PU at 28.626MHz using the clock module, I ran it only at 5 volts. I drove almost every IO line, had the program memory full, and used most of the internal SRAM. The only thing I did not test is the EEPROM.

Brad

User avatar
DaveyPocket
Posts: 378
Joined: Sun Sep 14, 2008 8:33 pm
Contact:

Re: Uzebox & SRAM

Post by DaveyPocket » Thu Jan 10, 2013 7:03 am

AtomicZombie wrote:When I was running the 644PU at 28.626MHz using the clock module, I ran it only at 5 volts. I drove almost every IO line, had the program memory full, and used most of the internal SRAM. The only thing I did not test is the EEPROM.

Brad
Well if we can confirm this works, I can use some of those spare 1284s I have :D

Is this clock module common or does it need to be "programmed"?

Post Reply

Who is online

Users browsing this forum: No registered users and 2 guests