Hissing and whistling in the STe audio output

[sandord]

Good thing you mention that because I was about to do it in my other STE.

I thought it'd be good to get measurements of the effects of the caps in comparison to a stock STE but things are usually not that simple, are they?
I'm putting the caps in my DAC-fixed STE now so don't worry

[exxos]

I'm putting the caps in my DAC-fixed STE now so don't worry

[sandord]

So, here's C567 and C520 replaced with ceramic caps:

IMG_20181112_221502.jpg (156.11 KiB) Viewed 7634 times

In the spectrum diagrams below, purple is before, green is after recap.
Note that no monitor is connected because the monitor cable being connected (even with monitor of) influences the result.
Of course the right channel doesn't show any significant change because I only recapped the left channel.

Disconnected (cable noise):

disconnected l+r cinch.png (27.55 KiB) Viewed 7634 times

Left channel only:

after-no-mon-L-only-merged.png (81.92 KiB) Viewed 7634 times

Right channel only:

after-no-mon-R-only-merged.png (32.82 KiB) Viewed 7634 times

Both channels:

after-no-mon-LR-merged.png (38.95 KiB) Viewed 7634 times

[exxos]

No real change then If anything looks worse ? Though that could be as the caps are "high up" on the board...

I think it would be interesting to cover the channel with foil ( insulated with paper first of course ) connected to gnd to screen the audio and see if its picking up interference via RF.. or if its just down to pcb noise...

Looks like big noise spike at 15.2Khz also...

[sandord]

No real change then If anything looks worse ? Though that could be as the caps are "high up" on the board...

I think it would be interesting to cover the channel with foil ( insulated with paper first of course ) connected to gnd to screen the audio and see if its picking up interference via RF.. or if its just down to pcb noise...

Looks like big noise spike at 15.2Khz also...

First, I see that there's something wrong in my recording path because when I connect the right channel plug to the left channel connector, I get kind of the same result as a disconnected cable.

So I have to measure the right channel again using the left channel cable.

I'm now trying to measure at the DAC, to see if there is any difference between L and R over there.

[sandord]

I decided to measure the audio path from begin to end like the last time.

Here are results, for the right channel only this time (as I recapped the right channel only, per the schematics in this thread).

= [R] DAC: U503 pin 4 ======================================================================

Before and after are the same because the caps sit later in the audio path.

dacr.png (33.13 KiB) Viewed 7619 times

= [R] U504 pin 14 ======================================================================

Before recap:

before-r-u504-p14.png (26.92 KiB) Viewed 7619 times

After recap:

after-r-u504-p14.png (27.46 KiB) Viewed 7619 times

= [R] U506 pin 5 ======================================================================

Before recap:

before-r-u506-p5.png (28.15 KiB) Viewed 7619 times

After recap:

after-r-u506-p5.png (25.22 KiB) Viewed 7619 times

= [R] U507 pin 7 ======================================================================

Before recap:

before-r-u507-p7.png (28.52 KiB) Viewed 7619 times

After recap:

after-r-u507-p7.png (28.32 KiB) Viewed 7619 times

[sandord]

Wow, if you look at the diagrams it almost looks like I swapped before and after... but that may be wishful thinking, right?

[exxos]

wow, the crappy electrolytics seems to be actually filtering out some of the noise...

What do you actually get on the DAC output ( pin 2 & 4 ) ?

[sandord]

wow, the crappy electrolytics seems to be actually filtering out some of the noise...

What do you actually get on the DAC output ( pin 2 & 4 ) ?

That's the first picture...

[exxos]

That's the first picture...

ah ok...

have you measured the power rails on each chip to see if anything spiky there ?

Also, hold down the reset button (it will stop the machine from running) so that would be interesting to test...