Reducing filament hum in a Tubelab SE

[boywonder]

I posted this question in an existing/old thread over in the tubes forum, but it appears that I'll get more action over here.............

When playing the Tubelab SE on my Half-Changs (95db), I have a little hum that I would like to reduce. It's not bad at all, but if I could knock it down a bit that would be great.

So.......I'm running 300B's, and an Edcor XPWR-131 power transformer.

C1, C3 and the other stuff around the filament regulator is per the schematic. I'm measuring 1.25V PP at C1, and about 28mv PP at C3, and 5.93V dc at C1 Are these reasonable values?

Per Georges calcs from another related thread, it appears that I have less than .5V of headroom for the regulator. I'm measuring 4.9V at the filament tube socket.

5.93-1.25/2=5.3v.

I've paralleled 34000 uf with C1, lowering the Vpp to about 540mv. This, at least on paper, is enough headroom for the regulator, but the ripple at C3 is still about 30mv PP with 15000+34000 uf capacitance at C1.

I've also experimented with a separate (6A current rating) filament transformer that increases the dc volts at C1 to about 6.3V, which should be plenty of headroom for the regulator. This had no effect on the 28-30mv ripple at C3. Adding another 10000 uf to C3 also had very little effect on the ripple at C3...

Any ideas?

 

[Evenharmonics]

Have you measured AC at the speaker terminal? My TSE is 1 mV and no audible hum from my 97db speaker. I use choke in place of R4.

 

[boywonder]

Have you measured AC at the speaker terminal? My TSE is 1 mV and no audible hum from my 97db speaker. I use choke in place of R4.

I'm using a 5H choke in place of R4 and I have 15mv PP at the speakers.

 

[gootee]

Do you have some photos that show your layout and wiring? Maybe you left some "loop area" in some wire pairs (instead of twisting or shielding them), which might be radiating and/or picking up AC like antennas. Or maybe an input ground reference point shares some portion of a ground return conductor with some AC.

 

[boywonder]

Do you have some photos that show your layout and wiring? Maybe you left some "loop area" in some wire pairs (instead of twisting or shielding them), which might be radiating and/or picking up AC like antennas. Or maybe an input ground reference point shares some portion of a ground return conductor with some AC.

Here it is. Power transformer is on a separate chassis with an umbilical. RCA signal cables are small coax........I suppose that I could twist the heater wires where they enter the chassis over to the PCB.

 

[Evenharmonics]

I have 15mv PP at the speakers.

That's very high. Something doesn't seem right.

1. Are you using volume pot at the input and how far is it turned up (or down)?
2. What value are you using for R31 and 32? This changed the hum level for mine. Tubelab said something about oscillation when lower value was used.

 

[boywonder]

That's very high. Something doesn't seem right.

1. Are you using volume pot at the input and how far is it turned up (or down)?
2. What value are you using for R31 and 32? This changed the hum level for mine. Tubelab said something about oscillation when lower value was used.

No volume pot, power amp only.

I'm using 36K R's for R14 & R25 (Mosfet sources) since my B+ is 350-400 per George's notes.

 

[rknize]

Have you tried shorting the inputs to ground?

 

[boywonder]

Have you tried shorting the inputs to ground?

The 15ma is with the inputs shorted to ground. Doesn't change much with my preamp plugged in either.

15ma PP/(2xsqrt2)=5.3 ma RMS

 

[rknize]

Do you have 100x probes so that you can look at C5 and C7?

 

[boywonder]

I have a couple of 600V 10X probes.

With inputs shorted:


At C5 ripple is 250mv PP (around 88V RMS) @ 120Hz. C5 is a 180u Pana TS-UP bypassed with a 100u/370VAC motor run. Disconnecting the 100uf motor run yields 350mv PP at C5.

At C7 ripple is 60mv PP (around 21V RMS)....freq reading on my Dig scope was all over the place.

 

[boywonder]

It wasn't filament hum at all......

I finally figured out the problem and it's fixed. I'm running a 9 conductor umbilical between the power transformer chassis and the amp chassis. The safety ground from the power entry module in the power transformer chassis is tied to a ground stud in that chassis, then on to the amp chassis to another ground stud which serves as the star ground for pretty much everything in the amp chassis.

Initially, to save on umbilical conductors, the power transformer CT was tied to the ground stud in the transformer chassis, resulting in the safety ground and signal ground sharing the same conductor in the umbilical (apparently not a good idea).

I ended up with one extra umbilical conductor (not used) in the initial build, so I used this conductor as a dedicated line to bring the power transformer CT up to the amp chassis ground stud.

Now with the inputs shorted I'm measuring about 1.5mv rms at the speaker terminals and the hum is barely audible on my 95db sens speakers with my ear right next to them.

And finally..one correction to my prior post: The C5 and C7 ripple rms values should be 88mv and 21mv, not 88V and 21V.

 

[Ian444]

Ha! Well done! Good find!

 

[rknize]

Glad you figured it out. Ground loops are a pain. Just to make life easier, I've been using Pete Millett's "ground loop breaker" trick in my new amps. Makes my setup much less finicky about what is plugged into what and which outlet that are plugged into, etc.