Salas DCG3 preamp (line & headphone)

Founder of XSA-Labs
Joined 2012
Paid Member
I have a problem

I just measured the spectrum of my DCG3 and it looks like there are some oscillation issues. I can't really hear it all that well but it shows up like a xmas tree on the FFT. My simple 2 transistor amp measures clean so I know the rig and software are not at fault. Using UCA202 and RightMark software.

What do you all think could be causing this?

592345d1484503892-salas-dcg3-preamp-line-headphone-dcg3-fft-1k.png


As a comparison of a clean amp, the same rig measures this:

592184d1484436971-sound-two-transistors-clapping-pocket-class-good-fft.png
 

Attachments

  • DCG3-FFT-1k.png
    DCG3-FFT-1k.png
    17.3 KB · Views: 1,568
diyAudio Chief Moderator
Joined 2002
Paid Member
Could even be radio interference or rails oscillation. It shows lower noise floor even to the clean one in areas its not excited and that's weird. If you will ground the Jfet cans with crock clips to the signal ground, does it change something?
 
Regarding the signal wiring, if running twinax with a sheild should I connect the shield at both ends on all connections inside the amp or just connect it at the source end? Is it better to "complete" the shield throughout the input signal wiring or leave each individual sections with the shield only connected at the source end? Iam thinking it's better to connect all the shields except the one to the amp board input so it all drains to the source. Any opinions?
 
Founder of XSA-Labs
Joined 2012
Paid Member
I set it up in another room with a known good PSU (7815/7915 based with CRCLC filtering good for solid 150mA bias). Unfortunately got the exact same HD xmas tree bloom. So I added a 50k pot and below a certain level (-2.7dB on VU meter - it says -1dB should be good), the HD goes nuts - just explodes. But a tad below and it vanishes. There is a minimum input load that is required to keep the amp stable. I should measure what that setting on the pot is and just add a hardwire resistor to ground to keep it from happening.

592374d1484517433-salas-dcg3-preamp-line-headphone-dcg3-setup-new-psu-pot.jpg


Here is the new FFT with the pot set so that the distortion doesn't blow up. Much better - this amp is predominantly H3 and H7 - I suppose having a differential input pair does that?

592373d1484517324-salas-dcg3-preamp-line-headphone-dcg3-fft-1k-pot.png
 

Attachments

  • DCG3-FFT-1k-with-pot.png
    DCG3-FFT-1k-with-pot.png
    15.5 KB · Views: 1,561
  • DCG3-Setup-with-new-PSU-and-pot.jpg
    DCG3-Setup-with-new-PSU-and-pot.jpg
    114.2 KB · Views: 1,600
Don't you have single coax for each channel signals? Its better.. Connect all shields to the PCB common. I would connect both ways the amp's wire shield too. I would try differently only in case of hum loop.

Seperate twinax each channel. 2 wires plus shield each channel.

In this diagram should I connect the shield at the vol pot and amp or leave it disconnected at the amp to "drain" like some interconnects do?

For the psu I will probably just connect the shield at all connections.
 

Attachments

  • IMG_0656.jpg
    IMG_0656.jpg
    317.5 KB · Views: 342