Page 5 of 17

Re: ULA revistited.

Posted: Fri Apr 15, 2011 12:30 am
by Prime
Humm just done a quick experement, built an oscilator with a 6.5536MHz crystal which is somewhat easier to find than 6.5MHz, the Sinclair ULA works fine with this, however the Xilinx replacement does not, so I somewhat suspect that clock doubling may need a re-think of the code....

Cheers.

Phill.

Re: ULA revistited.

Posted: Fri Apr 15, 2011 8:53 am
by superfo
Here is what I layout, fit xc9572 into 600 mil space.

Re: ULA revistited.

Posted: Fri Apr 15, 2011 9:16 am
by Prime
superfo wrote:Here is what I layout, fit xc9572 into 600 mil space.
That should work if you can source some 64 pin 9572s, I've never been able to find any :( Better still wonder if you can get 95108s in that format as that would allow for an enhanced ULA.

Cheers.

Phill.

Re: ULA revistited.

Posted: Fri Apr 15, 2011 9:36 am
by sirmorris
@superfo: Nice work!

Re: ULA revistited.

Posted: Fri Apr 15, 2011 10:10 am
by superfo
@Phill
I see Digi-Key has 64 pin 9572 in stock, 95108 is a little bit bigger, not sure it can be fit.

@ Sirmoris
Thanks!!

Re: ULA revistited.

Posted: Fri Apr 15, 2011 11:22 am
by Prime
superfo wrote:@Phill
I see Digi-Key has 64 pin 9572 in stock, 95108 is a little bit bigger, not sure it can be fit.
Ahh right, I only tend to order from digi-key if I really can't get it elsewhere as I find their £22 'surcharge' a litle bit LOLPRICE :) Have ordered from them in the past tend to batch up and make a big order, perhaps it's time to again.

95108 are prolly the 100 pin package, they deffo won't fit, I have a board designed usint the 100 pinwith the chip mounted just below the ULA footprint.

Cheers.

Phill.

Re: ULA revistited.

Posted: Fri Apr 15, 2011 5:58 pm
by Andy Rea
Prime wrote:Humm just done a quick experement, built an oscilator with a 6.5536MHz crystal which is somewhat easier to find than 6.5MHz, the Sinclair ULA works fine with this, however the Xilinx replacement does not, so I somewhat suspect that clock doubling may need a re-think of the code....

Cheers.

Phill.
What problems are you having with the clock doubling ? doe it work at all or just crash as soon as it's enabled. i have noticed a peculiarity and that is if clock doubling is enabled on power-up (or reset) it starts up in fast mode (i think this is because when it does the check to see if an NMI generator is present the code no longer waits long enough, becuase it running twice as fast), i also have a funny display now, but have not changed anything else some display lines (8 scanlines) are shifted by 2 pixels... i suspect this may be down to stray capacitance as placing my hand over the entire circuit alters it's behaviour.. but besides these anoyances it works without crashing and the boosts in speed is impressive at first :D

Re: ULA revistited.

Posted: Fri Apr 15, 2011 6:02 pm
by Andy Rea
a 100 pin qfp could be mounted square on you would lose the middle 5 pins on each side of the dip socket but that still allows 30 of them to be thought hole adding to the mechanical strenght.

SuperFlo, am i missing something.... have you no oscilator circuit on the board ?

Andy

Re: ULA revistited.

Posted: Sat Apr 16, 2011 2:07 am
by Prime
Andy Rea wrote:
Prime wrote:Humm just done a quick experement, built an oscilator with a 6.5536MHz crystal which is somewhat easier to find than 6.5MHz, the Sinclair ULA works fine with this, however the Xilinx replacement does not, so I somewhat suspect that clock doubling may need a re-think of the code....
Well The above was just trying to run the Xilinx chip with a slightly different crystal 6.5536MHz as opposed to 6.5MHz, turned out that I had to bypass the two transistors, as I am using an LS00+crystal+descretes and so the clock input was correct without the transistors. This is now working in normal ULA emulation mode, and works also with the boarder / paper inversion.
What problems are you having with the clock doubling ? doe it work at all or just crash as soon as it's enabled. i have noticed a peculiarity and that is if clock doubling is enabled on power-up (or reset) it starts up in fast mode (i think this is because when it does the check to see if an NMI generator is present the code no longer waits long enough, becuase it running twice as fast), i also have a funny display now, but have not changed anything else some display lines (8 scanlines) are shifted by 2 pixels... i suspect this may be down to stray capacitance as placing my hand over the entire circuit alters it's behaviour.. but besides these anoyances it works without crashing and the boosts in speed is impressive at first :D
Just seems to die as soon as enabled, I wired up a switch to the CPLD so that I can switch it in at any point, as soon as I enable it things go haywire, I added echoes of the enable disable signals (out FD/FE/FF) onto spare pins of the CPLD, and monitored them with my PC logic probe, they toggle reliably until I actually enable the lock output and then they stop after a few cycles, so I suspect things are gradually getting out of sync. Current code is now posedge triggered at your sugestion, and I don't think the clock is glitching from my probe traces.

I'm gonna build the external clock doubler with some LS chips as I can validate that with the Sinclair ULA and then try it with the replacement to see if that works or if it's triggering a bug in the main code, as I'll need to get that fixed first before even trying the doubling logic in the CPLD....I'll get there eventually :) :) :)

Cheers,

Phill.

Re: ULA revistited.

Posted: Sat Apr 16, 2011 7:07 am
by superfo
@ Andy, On board I have 3.3V regulator and ltc1799 chip for generate 6.5MHz.
@ Phill, I will order 9572 64pin soon , I can ship 1 to you.

Here is layout for 95108, have not include oscillator circuit and pin for programming yet. And need to add copper pour fro ground plan.
XC95108_100PQFP.jpg
(353.06 KiB) Downloaded 3806 times