I'm very happy with the STE DAC upgrade

[sandord]

Well, I swapped bits 0 and 2 for each primary colour but the result is still wrong

I decided to write down the mapping of the bits, so I took a screenshot from the schematics and drew three rows of bits on it:

• Top row: the names of the DAC pins each resistor is connected to via traces on the DAC PCB. Since I don't have the schematics of the board, I only wrote down those pins that I could trace visually. The board is currently soldered into the ST so I can't look at the bottom at the moment.
• Middle row: the bits are numbered in such way that they can be fed to the DAC which means they are in normal order (0=least-significant, 3=most-significant).
• Bottom row: the bits as numbered by Atari, where bit 3 is actually the least-significant bit, while the remaining bits are order properly. This matches the resistor values: B3 has the highest resistance and thus is the least-significant (Atari did this to maintain compatibility with ST software that was not aware of the extra bit)

@exxos: could you please complete the DAC pins on the drawing, so that we can see how everything is currently mapped and what exactly needs to be rewired?

bit-mappin.png (193.5 KiB) Viewed 270 times

[exxos]

I'm not at home weekends so can't check.. But if what you posted is correct then that would be the DAC order I used.

The ST as I posted here viewtopic.php?p=138659#p138659
Bit0 is 12k.. So higher bits is less resistance..

The STE values you posted show bit0 as 12k.. So that's correct.. But then looks like bits 1&3 are the wrong way around. But I'd check the order yourself anyway to verify..

[sandord]

But if what you posted is correct then that would be the DAC order I used.

The ST as I posted here viewtopic.php?p=138659#p138659
Bit0 is 12k.. So higher bits is less resistance..

The STE values you posted show bit0 as 12k.. So that's correct.. But then looks like bits 1&3 are the wrong way around. But I'd check the order yourself anyway to verify..

Can you post the schematics of the DAC pcb?

[exxos]

Can you post the schematics of the DAC pcb?

Its the same as on the H5 schematic..

[sandord]

I went a bit out of my way to finally understand how the wiring is wrong and how it should be, resulting in the drawing below.

I think this perfectly describes my symptoms.

Because of the incorrect mapping, the bits are effectively rotated right by one position. In GEM or GEM programs that don't change the palette, this is mostly not noticeable because of how the default palette is populated (e.g. white and pure green have all the bits set). But in other programs, this effectively halves the brightness.

In programs that are aware of the extra STE bit, gradients show alternating contrasts because the LSB is mapped to the MSB on the DAC.

By the way, I verified all routing mentioned in the diagram on the DAC board using my multi meter.

dac-wiring.png (179.93 KiB) Viewed 221 times

[exxos]

Wasn't the extra bit on the STE pallette the higher order bit? I think that was done to keep ST compatibility.. So the higest colour bit is actually the lowest brightness which makes it out of order as you say..

[sandord]

Wasn't the extra bit on the STE pallette the higher order bit? I think that was done to keep ST compatibility.. So the higest colour bit is actually the lowest brightness which makes it out of order as you say..

That's exactly what it is!

[sandord]

OK, this time I'm for real. I fixed all issues with the bits and I now have a rock solid, distortion-free, correctly coloured and properly lit picture!

proof.jpeg (440.89 KiB) Viewed 178 times

I rerouted all pins by using wires from the mainboard to the DAC pcb instead of the header pins.

rewiring.jpeg (359.45 KiB) Viewed 178 times

Here's the final result on the main board:

FINAL.jpeg (506.31 KiB) Viewed 178 times

Sadly, I ripped a few pads and traces during the process. Since I had to install the PCB 3 times and the pins are seated really tight, I had to use some force to remove them.

surprise.jpeg (237.25 KiB) Viewed 178 times

So, in the end I had to add 5 bodge wires for all the ripped pads and traces

The worst one was a pad that connected to a pin on the shifter, which I luckily could find using the schematics.

bodge.png (849.98 KiB) Viewed 178 times

bottom-bodges.jpeg (472.38 KiB) Viewed 178 times

[exxos]

OK, this time I'm for real. I fixed all issues with the bits and I now have a rock solid, distortion-free, correctly coloured and properly lit picture!

I'd suggest you flux and reflow some of those joints and on the DAC as looks like you have a few cold solder joints.. They will likely cause connection issues later..

[sandord]

I'd suggest you flux and reflow some of those joints and on the DAC as looks like you have a few cold solder joints.. They will likely cause connection issues later..

You're completely right. Thanks!