High Frequency tone in the noise floor on SB3 R channel

[*Isaac*]

measurements are:

-measurement system noise floor

-brightness 4

-brightness 3

-brightness 2

-brightness 1

-brightness 0 display off

in that order.
Sorry about the formatting, haven't posted much before.

anyone else want to do some measurements on theirs?

[Eric Carroll]

This was an old thread, with other threads tracking the issue more closely. This is a known issue, seen by Stereophile's review and acknowledged by Slim. The issue is that the gain is too high between your SB source and power amp. The resolution is to add attenuation between your SB3 and the power amp - either turn down your preamp or add passive attenuators.

See the ConnectToPowerAmp wiki page http://wiki.slimdevices.com/index.cgi?ConnectToPowerAmp in the "Other Power Amp direct connection problems" section.

[*Isaac*]

If this doesn't bother you, or you have an external DAC, great.

I like to see a circuit perform as well as it can be made to; for me the problem is that this noise limits the s/n ratio of the whole system, at it's worst it is 27dB above the thermal noise floor.
The fact that it only affects one channel led me to believe that the issue was one of board layout, and close inspection seems to confirm this:

As you can see, the inverting input node, which is usually the node most sensitive to interference (dotted yellow) of the left channel is much smaller than the Rch one. Further, the right channel trace is directly behind the end of the neon display, where much energy is radiated. It also changes layers in this area (between the pink dots), possibly increasing susceptibility.
It is genrally good design practice to minimise the length of the inverting input node to minimise stray capacitances, and to reduce the scope for parasitic oscillation and instability.
In this case the noise is probably being coupled through the RCH inverting node, and to a lesser extent through the trace from the DAC.

To try and reduce this I made a quick and dirty shield from aluminium foil, insulated with mylar tape. I fitted this between the back of the neon display and the PCB, then soldered a wire to the incoming power ground.
Here is a noise floor measurement done after the mod, of the worst case situation (brighteness max -1) encountered before the mod:
as you you can see, the noise is virtually gone. This mod may cause problems with the wifi reception though (because there is now a shield over most of the front). Also don't do this unless you're happy voiding the warranty, maybe destroying your SB, and also probably violating some FCC rules

Anyway, something can and should be done to fix this problem in the next hardware revision.

[Phil Leigh]

Very good - well done.
So basically you've physically inserted a grounded screen between the VFD and the main circuit board...and the noise is gone or greatly reduced. I agree that SD/Logitech should improve the screening of the VFD in future hardware.
Hopefully they will notice this - they may already be aware of it.

[tyler_durden]

The next version of the SB should be a box with no display and a remote control with a big display. Of course, the big display will be a power hog, so the remote will need to sit in a nice display stand/charging dock.

You could do this yourself now by modifying the heck out of the SB - get rid of the display and put in an analog supply for the audio stages that are normally powered by dc-dc converter in the power supply. Use N800 type device as the remote control/display.

TD

[NewBuyer]

Does anybody know, if the Transporter has any similar issue?

[Eric Carroll]

Does anybody know, if the Transporter has any similar issue?

It does not.

[Eric Carroll]

To try and reduce this I made a quick and dirty shield from aluminium foil, insulated with mylar tape. I fitted this between the back of the neon display and the PCB, then soldered a wire to the incoming power ground.

Nice work Isaac. I will add a link to your article to the wiki page.