Internal 16K RAM mod doesn’t quite work with Dr Beep hires games

[ILAH]

I dont know, i think LC. I will report later.

-> I can not identify, the surface is scratched. Will the cap depend on the version?

The RAM is from Hynix:
HY62CT081ED-70C = 70ns refresh

[Flatulentia]

I can not identify, the surface is scratched. Will the cap depend on the version?

I don't think so, but it may affect timings elsewhere that cause problems.

If memory serves, I think I used a 74HCT00, but I can't see mine as it's glued face-down to the top of the RAM. That's CMOS with TTL switching levels which seemed to make sense at the time.

Whether it matters or not, I don't know. It's just more information for others to be able to help you if they can.

[1024MAK]

The RAM is from Hynix:
HY62CT081ED-70C = 70ns refresh

The -70 actually refers to the access time in nanoseconds, that is, how quickly it can make data available (reading) or accept a write. The chip may actually be able to operate quicker, but this is the maximum amount of time that it would take.

Not to be confused with how long the microprocessor actually takes. The microprocessor can take as long as it likes.

I presume it works okay when the machine is running BASIC in SLOW mode?

Can you please point us to the exact details you followed to do the upgrade. Especially with regards to how pins 20 and 22 of the chip are connected.

And exactly where you put the capacitor. Some photos would help.

Mark

[ILAH]

Maybe this can help. The schematic of the mod and an older picture showing the wiring.
I tested putting the cap between the red marked pins.

Here is a detailed description of the mod.
http://www.8bit-homecomputermuseum.at/r ... ntern.html

Normal progams, for example 3DMaze, Fortyniner etc. are working good.

[1024MAK]

Okay, from the photo, if I had to guess, based on what I can see, the NAND gate logic chip is a 74LS00. That’s fine. And in any case, many other 74xx00 logic chips should work correctly (7400, 74S00, 74ALS00, 74HCT00, but note that I’ve not actually tried all these). And in any event, I would not expect this part of the circuit to be a problem with the reported symptoms.

I’ve had a look at the wiring for pin 22, I myself prefer to connect pin 22 to pin 20, especially since the board already has a PCB track linking these two pins together. But connecting pin 22 to 0V/GND (pin 14) works as well for most types of SRAM chip. And having quickly looked at the datasheet for your SRAM chip, I can’t see a problem with this arrangement.

Note that I have not checked all the other wiring.

Unfortunately for each type of SRAM chip, it may require experimenting with the value of capacitor used. And yes, the pins you marked is a suitable place for it (although I do prefer the position used by another member, that is across IC4B pins 8 and 9).

It’s a case of finding the “sweet spot”. Too small a value and it won’t fix the problem. Too large a value and you may mess up the timing in the other direction. Having said that, you can’t do any harm by fitting a larger value (well as long as you don’t get silly). So try various values up to 470pF.

Mark

[ILAH]

Am i right Pin 22 lifted up -> GND
Pin 20 in socket -> GND

doenst work an results in this screen.

[1024MAK]

Am i right Pin 22 lifted up -> GND
Pin 20 in socket -> GND

No.

Either both pins 20 and 22 in the socket and linked together (via the PCB track) and both controlled by the existing/RAMCS from the ULA. Either pin can have a capacitor connected. Do not connect either pin to 0V/GND.

OR

Pin 20 in the socket. Only a capacitor should connect to here. This pin must NOT be connected to 0V/GND.
Pin 22 lifted and wired to pin 14 (which is 0V/GND).

Your problem is not the wiring, but the SRAM chip having a faster response time. The SRAM chip is likely switching off it’s outputs so quickly that the video circuitry is not able to get the correct data. The point of the capacitor, is to delay the /RAMCS signal from the ULA going high (inactive), hence delaying the time before the SRAM chip turns off it’s outputs.

Mark

[ILAH]

OR

Pin 20 in the socket. Only a capacitor should connect to here. This pin must NOT be connected to 0V/GND.
Pin 22 lifted and wired to pin 14 (which is 0V/GND).

This is exactly my wiring.

I tried some caps. Up to 100 pF (nF) i get the first result, 200 / 300 pF (nF) make second pic. Higher nFs keep the ZX81 freezing with black screen or artefacts. Corrected to pF!

[1024MAK]

Okay, I’ve tried to show the effect adding a capacitor has.
Click on this link

The input shown as CLK is simulating the /RAMCS signal from the ULA. The 100kΩ resistor is not actually this value, but does exist on the ZX81 circuit board. It’s actual value is 680Ω. I’ve used a different value for demonstration purposes otherwise you would not be able to see a difference in the waveforms.

Similarly, I’ve used a simple SRAM chip.

But you can clearly see that by delaying the control signal to the SRAM chip by using a capacitor, the data from the SRAM chip gets delayed slightly.

Mark

[Flatulentia]

I tried some caps. Up to 100 nF i get the first result, 200 / 300 nF make second pic. Higher nFs keep the ZX81 freezing with black screen or artefacts.

That's way too high. Please be careful as there's only a 680R resistor between your capacitor to ground and the /RAMCS pin on the ULA.