31 Jan
2012
31 Jan
'12
10:59 a.m.
The Color SRAM is a 2114. Very ordinary and, in my experience, prone
to failure.
If you have another C-64, you could try swapping the PLA. You will
need an 18-pin socket for the SRAM. The chips are available from a
number of surplus places like Unicorn Electronics, BG Micro, and more.
Speed ranges for the 2114 are, IIRC, 250ms-450ms. I don't remember
what speed the C-64 requires, but you can typically sub a faster chip
than the circuit requires.
-ethan
I have already purchased a replacement sram for the 2114 (acutally 2 of the units, one
P2114 and one MM2114N-2) and intend to replace the IC sometime today. I do have another
C64, but no chips are socketed, and I don't want to ruin it, thus I'll just try
replacing it in the bad unit. My understanding is that I need a chip that has a speed
faster than half a machine cycle for the system to work properly.
The spec sheet on the P2114 states that it has a 450ns access time. For the MM2114N-2 it
looks like its 200ns.
I understand that the access speed requirement is .5*(cpu freq/1)/machine-cycle. I
don't know the machine cycle for this operation, but assuming the worst with an IC
having an access time of 450ns, the machine-cycle couldn't be over 2.25 for things to
be ok. If I use the 200ns IC, I should be good with a machine-cycle of just over 5. I
could probably solve this by opening up the unit and just pulling the IC scribe marks and
looking up the data sheet, but the location with the ICs is 50 miles from the house, and I
happened to be nearby yesterday...so I just winged it. Can anyone checksum my thoughts
here? Is my algorithm and its implementation ok?
I also bought a replacement for the PLA, however, I had misunderstood the requirement. I
thought it was just a plain ole N82S100, but now I understand that it is a specially
programmed N82S100. Although I have found a ROM file I could perhaps burn on the chip, I
simply don't have the gear. I'm hoping this isn't the problem, as I have to
mail order the part, but I understand that this is the most common IC to fail on the C64
board. Since I'm getting only character color palette cycling, and have a border and
don't have a problem with a blank screen, I am encouraged that this IC is probably
ok.
Just to be safe, I bought a replacement for U29, as the description in the c64 chip and
common problems guide has content that really looks similar to my problem, moreso than
even the description for problems with U6 (the 2114).
Thanks for your commentary though. I'm likely to buy sockets for these chips as well,
as they aren't that expensive, and it may save me issues should the selection I made
be suboptimal.
Kevin
31 Jan
31 Jan
2:49 p.m.
On Tue, Jan 31, 2012 at 1:59 PM, Kevin Reynolds <tpresence at hotmail.com> wrote:
I have already purchased a replacement sram for the
2114 (acutally 2 of the units, one P2114 and one MM2114N-2) and intend to replace the IC
sometime today. ?I do have another C64, but no chips are socketed...
Bummer. ISTR my C-64s have socketed PLAs, but perhaps they are older units.
My understanding is that I need a chip that has a
speed faster than half a machine cycle for the system to work properly.
That is more-or-less true, but the devil is in the details.
In the case of the color SRAM, both the 6510 and the VIC chip access
it, on alternate phases of the clock, so with a clock speed of approx
1MHz, the SRAM can be accessed twice in 1ms. That suggests that a
450ns chip is fast enough, though I might want something faster myself
so that I knew it wouldn't be a problem. A 450ns chip is plenty fast
enough for most 1970s and 1980s microcomputers with a single-access
1Mhz design (some of the early SRAMs like the 5101/2101/1822 256x4
SRAM had access times of 650ms... those can have problems in some
places, but you won't find those in common usage after 1978 or so.
They were displaced by various 4Kbit parts).
I also bought a replacement for the PLA, however, I
had misunderstood the requirement. ?I thought it was just a plain ole N82S100, but now I
understand that it is a specially programmed N82S100.
Yes. It's an 82S100 that is programmed to "do the right thing" in a
C-64. Yes, it fails often, but fortunately, it's a well-understood
chip and can be reproduced.
In the medium-term, you might be able to find someone who can burn
your blank for you.
Thanks for your commentary though. ?I'm likely to
buy sockets for these chips as well, as they aren't that expensive, and it may save me
issues should the selection I made be suboptimal.
Let me echo Tony Duell's recommendation of turned-pin (machined-pin)
sockets. I use them exclusively, even in kits I buy (I throw the
supplied cheap sockets in a box and use my own).
-ethan
1 Feb
1 Feb
1:16 p.m.
Let me echo Tony Duell's recommendation of
turned-pin (machined-pin)
sockets. I use them exclusively, even in kits I buy (I throw the
supplied cheap sockets in a box and use my own).
Actually I built a couple of kits recently and did use the suppleid
formed-pin sockets. But theyr'e simple enough (a telephone ringing
voltage detector [1] and a monitoring audio amplifier based roudn the
LM386() that if the socket fails it'll be very easy to trace and and that
point I fit a turned-pin one. Not like a RAM IC where a bad conenctions
will cause faults that would be hell to trace.
[1] This seems to have been designed by 'Mad Man Muntz'. It worked a lot
better after I added a resistor and capacitor (to totally different parts
of the circuit.
-tony
4660
days inactive
4661
days old
2 comments
3 participants
participants (3)
-
ard＠p850ug1.demon.co.uk -
ethan.dicks＠gmail.com -
tpresence＠hotmail.com