Twin Shifter mod Prototype build results

[sporniket]

Well, basically any electronic book or reference will advice against mixing CMOS and TTL parts. The voltage levels are not compatible. More specifically, it is usually ok to connect a CMOS output to a TTL input, but you should not connect a TTL output to a CMOS input. You can check the datasheets. That's precisely why there are HCT parts, that are CMOS chips but with TTL compatible voltage levels.

Ha, I see, I am aware of that. For this case I believed that it was not a problem, because the inputs are pulled toward VCC, so it reach the CMOS threshold for logic input high.

HCT parts are also designed to sink enough current to survive with TTL chips. HC chips will be being strained, so even if they work for a bit they may die.

Thank you both for the tip about HCT, i will keep it in mind for the next revision : https://github.com/sporniket/atari-st-m ... t/issues/1

[ijor]

Ha, I see, I am aware of that. For this case I believed that it was not a problem, because the inputs are pulled toward VCC, so it reach the CMOS threshold for logic input high.

How are they pulled toward VCC? I don't see any pullups on the left side of your datasheet, and there are no pullups on the motherboard neither. May be you are thinking on the pullups on the main CPU data bus, but you don't connect to the main bus. And besides the data bus, there are two control signals from the 74LS257 to the 74HC245's

[ijor]

HCT parts are also designed to sink enough current to survive with TTL chips. HC chips will be being strained, so even if they work for a bit they may die.

Are you talking about the output? Aren't you mixing up with AC/ACT families?

[sporniket]

@mrbombermillzy, here is a test picture that explains what happens, namely, when displaying the picture, the bit manipulation of the data bus is activated for the second shifter.

TESTPIX2.zip

On a STe, the picture looks like this :

testpix2-on-ste.png

You can see a dark stripe under the light grey pixel, and nothing under the blueish pixels. On the twin shifter, the dark stripe will also appear under the blueish pixel (if I counted the pixels correctly), following the mapping here (e.g. bit 14 of the databus is fed to bit 14 AND bit 9 on the data input of the Shifter) :

fantom-pixel-source.png

[mrbombermillzy]

@sporniket: Just had a look at the post above (also looked at the included image, but I'm not familiar with the PI1 format header info, so may be easier to ask the info here, as I don't know which colour/position bits you are setting to what).

Forgive my mental exhaustion this weekend, but I'm still none the wiser from the explanation above. Sorry. :?

So does the board take 16 bit data from the bus and sort of DEMUX it between both shifters' inputs?

[sporniket]

ha, this has nothing to do with technical details of the PI1 format. The picture just shows a palette, some bright pixels as markers, and a test zone. The last line of pixels is in fact noise, I should have remove it.

This is related with the fact that pixels are rendered by group of 16, because data is received through a 16bits databus.

The number displayed tells which bit is tested. E.g. bit 15 is the left most pixel. But, the second shifter receives this bit at 2 places : bit 15 and bit 6.

So, in the test image that use colors only rendered by the second shifter, there are real pixels (under the light grey ones), and there are no pixels under the blueish ones, BUT the second shifter will render pixels under both, hence the pixels under the blueish markers are fantom pixels.

Now, what it means in a real picture using the extended color range provided by the twin shifter is a light color bleeding. The main shifter will render the MSB of the colors normally, but the LSB rendered by the second shifters are false on specific columns, as they are a copy from another specific column.


I admit that it is difficult to explain.

[mrbombermillzy]

I admit that it is difficult to explain.

Ok, I *sort of* get it. ;)

Could you maybe construct some sort of bitwise table/chart of which shifter positions are falsely repeated/missed by each shifter?

[sporniket]

ok, I'll work on it.

[mrbombermillzy]

No rush. But thank you. :D

[sporniket]

Could you maybe construct some sort of bitwise table/chart of which shifter positions are falsely repeated/missed by each shifter?

Here it is. I hope it is clear enough.

TruthMap-exp1.pdf