Phase Inverter hum

[Giaime]

Hello all,

I have a problem with a new build, a 2 channel SLO-ish amp.

The problem is: annoying hum, that isn't affected by the preamplifier, and it does not go away when grounding the input (before the capacitor, of course) of the LTP. The only way to get rid of it is to short together the two LTP grids.

The LTP is fairly standard, 12AX7 with 100k/82k anode resistors, 470ohm common cathode resistor, then 22k resistor and a 8k2/1k divider for negative feedback. 100nF capacitor between the NFB

Heaters are AC and properly referenced to a voltage very near the actual cathode voltage, this heater chain also serves the 2x6L6GC output tubes (pretty standard, 5k6 load, 490V on plates). Swapping LTP tube doesn't help.

Suggestions?

Here is the schematic for the LPT and a screenshot of the hum measured at 8ohm speaker terminal (with speaker connected).

[Alex R]

does it go away if you pull out the PI tube?

[Giaime]

Yes, it goes away completely!

Edit: I should add that:
- the output transformer secondary ground tap is properly grounded: I connected a shielded cable between output jack and LTP circuit (I know shielded cable grounded at both ends is a no-no, but I also tried disconnecting the shield on one side and running a separate ground wire, no change);
- the HV reference for the filaments (1meg - 330k) is properly bypassed to ground with a 100nF cap;
- tried a 8n2 2kV capacitor connected across HV winding, trying to reduce diode noise (cap-coupled to the heaters via stray capacitance in PT?), no change, I have fast diodes there (UF5408 Graetz bridge).

[Alex R]

I'd say, assume you miswired something and check it over with that in mind.

[Chuck H]

Have you tried pulling the last preamp tube to see if the hum and distortions on your scope may be created BEHIND the PI??? That scope trace looks anawful lot like a floating ground to me.

[Giaime]

Hello all,

thanks for your replies.

Tried today:
- moving the B+ point (and its associated ground return) for the HV heater reference from small signal stages to power stage, to be more specific, to the screen grid supply, and the hum isn't changed
- removing NFB, either by shorting the 1k resistor or disconnecting the 8k2 resistor increases hum
- it shouldn't be a residual ripple problem, because if I have too much ripple on the PI supply the hum should be worse with the PI tube disconnected (direct unattenuated path between B+ line and output tubes grids), instead the hum disappears
- changed PI tube with a brand new, selected for balanced triodes, Tung-Sol 12AX7 and the hum is the same

@ Chuck H: no need to pull the last tube before the PI, since before the PI is the master volume and I can keep it down. However to be sure, as I said in my first post, I tried grounding directly the PI input and still got the same hum.

Could be that the 1meg + 330k//100n resistor divider, to obtain the reference voltage for the heater winding, has too much impedance to sustain the cathode-heater leakage current that occurs with 2 6L6GC and 3 preamp tubes? I have to try that.

[Giaime]

Found it. It is the +12V supply for the first tube heaters and relays. Disconnected the AC coming from the PT to the rectifier board, no hum.

But it's very strange. I already did an amp like this and got zero hum. The +12V supply has less than 500mA load and the relative AC wires are tightly twisted and run very near the chassis, far from the phase inverter tubes or any associated wires. Moving around those AC wires does not influence hum, but placing a 2R resistor in series with the AC wires, to limit peak current, does reduce hum, but this produces too low DC voltage.

[Chuck H]

Maybe examine grounding scheme and decoupling again?

[Giaime]

Very strange. One could expect that the pulsated rectifier current would create stray fields, so higher current = more hum.

Instead, increasing input capacitance, which should increase peak current, actually REDUCED hum. There is enough headroom to put a voltage regulator, so I did and the hum is now zero.

That's very strange because the PI tube isn't DC heated, and there is only a single ground connection to the star point from the +12Vdc supply.

[Euthymia]

If you're still curious about this, I would suspect an issue with the center tap or virtual center tap on that heater supply. I know you said that the heater supply was "properly referenced," but what happens if you just put that center tap to chassis ground?

Is it an actual center tap or a virtual (100 ohm resistors) center tap?

[Giaime]

Hello!

The 6.3V winding (that powers up the power tubes and the PI tube) has a physical center tap, and that's connected to the HV reference point (about 80V).

If I connect that center tap to ground, the hum is exactly the same. Of course leaving it floating increases hum.

I went from 1000uF to 4700uF on the capacitor input filter for the +12Vdc supply, and hum is drastically reduced, as I said adding a regulator has brought the hum to zero. But I wonder what is the connection between +12Vdc ripple and the AC heaters...

[Alex R]

Maybe it's not so much EM interference as to do with the way your rectifier draws AC current from the 6.3vAC supply. If it was trying to unbalance that supply, or feeding ripple back into one side of it somehow, that might upset the way the PI tube's heaters were trying to 'humbuck'. Can you recall if pulling the DC-heated tubes made any difference?

[Giaime]

Maybe I wasn't clear enough, but the rectifier has its own AC winding in the PT and DOES NOT share ANYTHING with the AC winding of the PI and power tubes heaters (aside the common ground reference)!

[Alex R]

Oh, right, I see. DC heaters are funny things!

[Giaime]

Thank you for your help, much appreciated. Now the amp is quiet, I still can't understand why...