STE SOUND DAC CRACKLE & POP FIX - Discussion thread

[complicated]

Is that the correct one?

Looks correct. Don't forget you need 2 of them.

EDIT:

You could try adding something like 40R in the 5V pin on the chip first. At least this will be easier than changing the chip to start with.

40R, is that a 40ohm resistor? You got any pictures of such a mod so i know how to apply it to test?

[exxos]

40R, is that a 40ohm resistor? You got any pictures of such a mod so i know how to apply it to test?

There is a image in the first post..

You basically cut the power pin close to the MB. Bend the pin up, and connect a 40R ( 40 ohm ) resistor there. I just took it to a cap as it had 5v on it and it was just quicker to hook up.

Other values may work but its impossible to tell without trying some out. Typically values are 10, 22,33,47,68 ohms. So 47 might be best to start with as it's close to what I used. This will reduce the gnd current on the chip. It works on the STE, but anyone's guess on the MSTE until experiments are tried and documented,

[complicated]

40R, is that a 40ohm resistor? You got any pictures of such a mod so i know how to apply it to test?

There is a image in the first post..

You basically cut the power pin close to the MB. Bend the pin up, and connect a 40R ( 40 ohm ) resistor there. I just took it to a cap as it had 5v on it and it was just quicker to hook up.

Other values may work but its impossible to tell without trying some out. Typically values are 10, 22,33,47,68 ohms. So 47 might be best to start with as it's close to what I used.

Thanks. I know i have some DIP-20 IC sockets laying around and some 42 ohm resistors. Also ordered a few 74F374 and 74LS374 chips cause they were quite cheap and its not worth destroying the motherboard trying to desolder av 74F374 chip.

Will try your resistor fix first when i got time, of course i will report back if it works or not. If it doesnt work i cut off the F chips and put sockets in to try LS chips.

[rubber_jonnie]

Ok, so I have tested with my MSTE and I have the fault.

Right now I don't really want to disassemble it, as it's a bit awkwardly placed, so I will see if either of my STEs has the problem and circle back to the MSTE at a later point.

It has also just occurred to me that I should check the MSTE at 8Mhz as well as 16Mhz, and I can check at 8Mhz and 32Mhz on my STEs which are both running boosters.

I'll report back as and when I get the chance to look.

[ijor]

That is what I did at first with the caps. But I didn't like the idea in the end.

I can understand. There is probably no clean solution here. Other than using a different Shifter version, all fixes are really workaround. And probably all STE Shifter versions have the issue to some degree, just that in some versions it is worse.

Well, sure , it could have been for a reason. But unless a lot of people try the fix, we likely won't know anything else. Either that or people put up with the issues. Currently there are no other options for the mentioned MSTE machine. Unless someone else wants to spend the time looking into it all that is.

That's a very good and valid point. Unfortunately sound issues are very difficult to diagnose. Not only the original issue, but the result of the fix is sometimes also difficult to judge and sometimes even subjective.

Regarding the reason that Atari used "F" parts instead of LS chips. I suspect it was for the output stage and not for the input. Not because a faster output it is required, not in the strict sense. But because of the skew between both channels. Slower devices, as an indirect consequence, have a larger difference between the maximum and minimum possible delay. And then multiple slower devices can produce more skew than faster ones.

I didn't fully analyze the DAC logic, and honestly it is not my area of expertise. So I can't really tell if the skew issue is really significant or not. But I do suspect that was the idea behind using "F" parts here. To reduce the skew between the output of both channels as much as possible.

[exxos]

.
But I do suspect that was the idea behind using "F" parts here. To reduce the skew between the output of both channels as much as possible.

Possible I guess. But what speed are the DACs running at ? 50khz tops ? I didn't look at what speed it was all running at. Even so, I'm not sure a few ns would have any effect on audio signals. There would probabley be more channel skew across the amplifier sections than the digital side.

They could be latching the data very fast though. Where FAST series might be needed in that case. But in terms of the STE using the slower LS doesn't have any immediate side effects. If it did, I assume the audio would be worse.

I did test multiple shifters and the 3 main versions of the STE motherboard. The gnd wire fixed the audio on , i think, 4 motherboards in total I have tried so far.

I have a new STE board which seems a bad one. Its the one I posted about fixing in my blog. I am going to try the fixes on that when I get chance. I may try higher value resistors to see how far I can push it before the chips crap out.

Indeed like you basically say, there isn't likely going to be a "one size fits all" solution to fix problems on these machines.It just needs trials doing by the community. Even if it only fixes 50% of machines out there. Its still a lot less malfunctioning machines out there.

[complicated]

40R, is that a 40ohm resistor? You got any pictures of such a mod so i know how to apply it to test?

There is a image in the first post..

You basically cut the power pin close to the MB. Bend the pin up, and connect a 40R ( 40 ohm ) resistor there. I just took it to a cap as it had 5v on it and it was just quicker to hook up.

Other values may work but its impossible to tell without trying some out. Typically values are 10, 22,33,47,68 ohms. So 47 might be best to start with as it's close to what I used. This will reduce the gnd current on the chip. It works on the STE, but anyone's guess on the MSTE until experiments are tried and documented,

First test completed, i cut the 5v pins and soldered 2x 42.2ohm resistors on the chips. Unfortunately this did not solve the crackling sound issue on my Mega STE. Will try a higher value later, how high can we go, max 100ohm or something?

[exxos]

First test completed, i cut the 5v pins and soldered 2x 42.2ohm resistors on the chips. Unfortunately this did not solve the crackling sound issue on my Mega STE. Will try a higher value later, how high can we go, max 100ohm or something?

Yeah I think 100ohm is probably about as high as you could go..

[sporniket]

A question about the fix involving a "quieted" VCC through the 40R resistor : could it be that a ferrite bead would also be appropriate, or that would be for filtering other type of noise ? I ask because it reminds me some rules of thumb I see here and there about supplying voltage to "Analog VCC" pin ? They usually use a ferrite bead. I can also see that on the ST/STe. On the ST, I also see a filter using inductor

edit : typos

[exxos]

A question about the fix involving a "quieted" VCC through the 40R resistor : could it be that a ferrite bead would also be appropriate, or that would be for filtering other type of noise ? I ask because it reminds me some rules of thumb I see here and there about supplying voltage to "Analog VCC" pin ? They usually use a ferrite bead. I can also see that on the ST/STe. On the ST, I also see a filter using inductor

Its surge current on the power rails when its switching between states which causes the ground spike. Of course this is the idea of having a decoupling capacitor.. but things are never that simple. Like I posted earlier, some output stages in chips can momentarily create a short circuit across the power rails which compounds biking problem regardless of the "load" put on the outputs of the chip. The idea is to limit the current through the chip so just limits the current through the ground, and thus limits the ground spikes.

On the STE, the shifter ( and basically everything for that matter) causes ground spikes across the motherboard. So anything you can do to reduce the current flow through the ground will inherently reduced spikes. Hence the idea of limiting the current through the latches.

However, while this works on the STE ( and indeed reinforcing the ground wires) the MSTE is a completely different motherboard of course. The spikes generated by the latches will be reduced, but by how much is the real question and if it makes any effect or not.. So seems so far 42R is not having any effect, so presumably 100R will not either.

I guess the simple way to look at it is that adding the resistor is like turning the FAST chip into a LS chip.
I did briefly look into this here in fact viewtopic.php?p=79808#p79808

LS chips are slower than FAST. I think the best chips were about 5ns, and the LS ones about 30ns. so with them inherently being slower they should not respond" to such spikes whether they are on the ground or on the inputs. This is why they work a lot better in noisy environments.

The alternative is placing capacitors on the data line inputs but I would prefer the LS be tried to see if it solves the problem or creates any additional problems. This way at least I can make a note for people with a MSTE that such fixes are simply not suitable.

The last resort is placing the capacitors on the inputs of like 33pF.. But again if this the noise is that severe then it may not solve the problem either. It is possible that a combination of the LS chips and 33pF on its input may be the best overall solution.. We just do not know until things are tried out..