Ok,

found a cure for the oscillations.

Solderd to each Trident a electrolytic capacitor to OUT & G

VDD: 47uF
VDD_XO: 2.2uF
VCC: 10uF

These are minimum values to stop the oscillation.

Question is, are there any negative side effects since the tridents do not have any output buffer C's and I think this is not because of reducing production costs.

Kind Regards
Peter

Hi,

oh, bad news to hear that the Trident 3.0 still have this problem!

I have ordered a complete set of Tridents Rev 3.0 because I thought that with the new, improved design the problems with oscillations are gone!

:-(

To repeat the question from my #2 post.

@Twisted Pear Audio: Are there any negative side effects adding buffer Cs to the Trident output ?

Any other prefered way to increase the stability?

Regards,
Peter

And just being curious.

You were seeing oscillation at what magnitude ?

edit: question aimed at igy

.

Edited by user Tuesday, August 18, 2015 6:46:59 PM(UTC)  | Reason: question aimed at igy

As promised, here're some pics from my Trident 3.0 regs, before adding extra caps at output, but after fixing placid hd ripple. This created some stuttering noise at the amplifier outputs, too.

Edited by user Wednesday, August 19, 2015 6:36:21 AM(UTC)  | Reason: Not specified

You may have some EMI or an HF ground loop?

Hi Russ,

thanks for jumping in :)
I'd be happy to take further measurements/scope captures if you'd like to look into this issue, just tell me what/where to check.

I started looking into this, because I got some stuttering noise at the amp outputs. However, it's sure was not caused by AVCC. Although it was also not clean, replacing the older AVCC (SRA2 1.0) with the most recent version (SR_1.0) did not solve this stuttering noise issue (even though the AVCC output is much better looking).

The stuttering noise was gone when I added the 22u cap to DVCC and VDD_XO. I guess it was DVCC causing that noise. Their output is clean now, and the noise is gone. (I attach a wav with the stutter recorded, just to hear what it was like).

Regarding EMI/HF ground loop. I'm not sure what/how to check for these. Everything was assembled simply on my desktop table, no case, no output connected, not connected to safety earth, stock trafo wired to placid hd (modified to remove the output ripple as described here in the forum, output clean), which is connected to buffalo iii dac with a ~10cm cable. The buffalo iii is v1.0.1, got the microcontroller removed, no other mods. I observed the same results at home and at my workplace, too. As input I use a dsp board (TI OMAP LCDK), but then to check I replaced it with a Papilio Pro FPGA board only outputting clock (no other lines connected), to make the buffalo locking. (I heard the noise at the dac output only when the dac locked). Nevertheless, the trident oscillaton was there without any input connected.

Summing up, let me know if you'd like me to make some further measurements.

Thanks,
igy

BTW DVCC and VDD-XO are not directly involved with the analog side at all. This is why I think what we are seeing is some ground bounce.

Also be sure VDD (1.2V) trident is putting out enough juice - it is actually the most taxed reg on the board. You can add say 27R at R4 to bolster it's output current. The demand of that supply goes up substantially at high sample rates.

I always measured the Tridents assembled in the BIII.
The oscillation/ripple on Tridents were there without anything connected to BIII input/output, just installed in BIII and powered as ususal.

On the output stuttering noise, there were two conditions needed to be met to reproduce it:
1. bit clock / lock (with/without actual i2s data lines connected noise was the same), until no lock, output was silent as expected
2. connected to amp, in my case Hypex NC400

I tried directly connecting the amp to BIII output, and also in BIII-IVYIII-AMP setup, the result is the same.
Wire to amp had length of 10-20cm, stock cabling supplied with amp (shielded Mogami mic cable).

I tried many things, isolating i2s input lines, output caps on IVY to break DC, etc.
Nothing helped. Then I started measuring trident outputs and found the ripples.

I just installed the 22u caps on VDD_XO and DVCC, and then the output noise was also gone, so was the ripple.

I can try to add/change (I guess rather replace? it's not in front of me) 27R to R4, though I'm not sure I've it lying around (especially in SMD), so it'll take some time to source it. BTW, I have had the noise issue with just the 2.8MHz bitclock fed in, so no higher bitrates at all...

Anyway, I'm not sure what are we looking with this now:
- the oscillation on the Tridents
- or the noise on the output when connected to amp

Do you think that if I increase VDD 1.2V current supply by changing R4, that would probably solve the output noise issue, which is currently "solved" with adding caps on the other two regs on VDD_XO and DVCC?

You actually should supply a true I2S signal - so that the data pin is not floating but you are truly sending silence.

in other words - sending only a bit clock is nothing close to sending PCM slience. :)

Edited by user Wednesday, August 26, 2015 3:36:16 PM(UTC)  | Reason: Not specified

igy - not arguing with you - just trying to find the cause. Ripple != oscillation. What you are describing is nothing close to normal operation - so I am just trying to help you find the cause. :)

If you run a trident on it's own - with a dummy load - you will see it is self-stable. What I think is happening is you are seeing massive ground bounce (which of course local decoupling at the measuring point will remove)

The why is a head scratcher - i can't duplicate it. Still thinking.