Re: I'm very happy with the STE DAC upgrade
Posted: 19 Jan 2026 02:49
And if possible, summarise please what have you done and why for future generations. :-) This thread was too confusing to follow. :-P
Alright, here's a short description of what happened :Pmikro wrote: 19 Jan 2026 02:49 And if possible, summarise please what have you done and why for future generations. :-) This thread was too confusing to follow. :-P
Post Mortem
I ordered the STE DAC upgrade from the Exxos store because I wanted the picture one of my STEs was producing to improve. It had quite a lot of distortion in it: vertical bars and noise.
After I soldered the DAC chip onto the the DAC board and installed the board into my STE (following the instructions of course),I thought at first that the upgrade had gone fine. I drew this conclusion after looking at the desktop only.
But further testing at a later moment showed that the video output was actually too dark. I didn't realise it at first but it was actually non-desktop programs (non-STE aware programs that change the video palette, to be precise) that were too dark.
As a side note: one problem with this kind of things is that when you take a picture with your mobile phone, it automatically corrects brightness. This means that a dim picture may look pretty bright after all, which makes it really difficult to convey differences in brightness.
Anyway, I suspected that the resistors in the video circuit *after* the DAC board must be of a too high value so I tried shortcutting those. But still, the output level was way too low.
By the way, I tested this using an all-white picture (including the borders), to be sure that the average level of the colour lines was maximum (even though the signal is blanked during the video blanking periods, which roughly correspond to the invisible area outside the shadow mask).
This is what my scope showed:
I then looked at the video circuit diagrams and saw that there were some extra resistors connected to each primary colour signal that came off the DAC board. Now, I don't have much knowledge of electronics but they looked to me like being part of some kind of a voltage divider.
After changing some thoughts with Exxos, he concluded that (some of) these resistors had to be removed as well because they drew a little power from the video signal, explaining why the output level was too low.
Removing the resistors did not fix the video level, however, it was still way too low.
I then decided to do some more in-depth testing by starting Deluxe Paint and play around with the palette. That's when I noticed some very strange things regarding the colours.
When I created a gradient from $000 to $FFF, I noticed that it wasn't smooth at all but had alternating dark/bright entries. This immediately raised the suspicion that something must go wrong regarding the bit order of the primary colors (R/G/B) in the video signal.
I first posted some pictures of the effect, without explaining my exact suspicions (I didn't want to come to a conclusion too early as I didn't want to suggest that something was wrong with the DAC board) in this thread and what followed were a few iterations where Exxos suggested to swap some of the bits. But it still didn't work as it should...
(this shows the alternating colours/shades where a smooth gradient should be)
Now I knew already that the extra colours in the STE palette were added in such a way that programs written for the ST (which weren't aware of the existence of an STE) would still work properly. Since the STE does 4096 instead of 512 colours, it has 4 instead of 3 bits per primary colour (2^(3*3)=512 and 2^(3*4)=4096). The extra bit is however placed before the existing bits, instead of after them. This means that the bits are not placed in order.
Fun fact: regular (non-STE) programs would leave this extra bit at 0, which means that strictly, regular programs that change the palette are a bit less bright (1/16th) than STE programs could get.
The second thing I noticed was that the brighter colours in Deluxe Paint (the alternating ones) were actually MUCH brighter than the dark ones.
As my STE still didn't display the colours properly, I decided to go all the way to the bottom of it and draw a diagram of everything involved.
This is when I concluded that indeed, the bit order that the DAC board assumed, didn't take the bit order of the STE into account.
In the diagram, the red text indicates where the lines should have gone instead.
I then desoldered the DAC board, connected wires to all header holes and route them to the STEs main board in such a way that the bits were now in the correct order so they end up at the DAC chip where they should.
(the extra gray wires you can see on the DAC board here are bodge wires because sadly I ripped some solder pads off the board during the process of removing the board and re-soldering it again).
And yes, the picture now was perfect! Gradients looked how they should and brightness was perfect.
I do have to mention that the output levels are a bit high now, like 3.42V average instead of the 3.3V as rated by the DAC chip. But it didn't prove to be a problem for my monitor but strictly speaking it's probably better to place a resistor on the main board to reduce the voltage a bit. When I measured my other STE, it read an average of 2.88V so that should be the target if you ask me.
What made this whole process a bit more complicated that in should have was that there were two problems in play that both influenced brightness (the additional resistors that had to be removed and the incorrect bit order). There was also quite some confusion, mostly created by myself for concluding too soon that everything was fixed at several stages :oops:
But in the end, I'm really glad it now works and I learned quite a few things in the process.
In the meantime, Exxos has updated the installation instructions to include my findings: https://www.exxosforum.co.uk/atari/last ... /index.htm.
So, FINALLY the title of this thread becomes the truth: I'm very happy with the STE DAC upgrade :yay: