[dsavitsk]

OK, you sure about that with the relay? I got the impression from the Morgan Jones book that a B+ delay was a good idea with a silicon PSU due to cathode stripping,

Only an issue if you have rare expensive tubes, and a very high voltage supply.

I thought single-ended had poor PSRR, so I've got a lot of capacitance in there, I wish I'd known, these caps are huge and it's crowded already.

Terrible. The best things for keeping a SE tube amp quiet are space between magnetic parts, and a regulator. Try using a MOSFET as a pass transistor, and put an RC circuit on the gate to provide a slow start if you are concerned about the above.

 

[elliottstudio]

Waki,
I hope you feel better soon.

 

[wakibaki]

Thanks, I'm already pretty much recovered.
 
A question for all you tube enthusiasts. I see quite a few designs out there like the Millet variations, most of which are inverting, e.g. a common cathode stage (inverting), buffered by a mosfet source follower (non-inverting), which adds up to an inverting amplifier. Nobody seems to care. I searched one of the Millet threads for the word 'inverting' and it wasn't mentioned once. Phase inversion is a big issue for many audiophiles though, and it is increasingly considered a no-no. It's easy to avoid in solid state amps, particularly opamp-based ones and discrete amps mostly have the option of applying positive feedback to the inverting input of a LTP.
 
I have the option of building this amp as inverting or non-inverting depending on how I connect the transformer, it's just a question of which wire to connect to B+ and which to connect to the OP tube plate, but it makes a big difference when considering any global NFB. I can apply NFB from an overall non-inverting amplifier to the cathode in the input stage, but I think this is a bit ugly, but it means that the amplifier overall  can be built non-inverting.
 
What do you think? Obviously I can switch it around later, but I'm close to finishing the first iteration and it affects the neatness of the wiring. Once cut, I'll have to patch in a bit of wire if I want to change it later.
 
w

 

[elliottstudio]

You have two stages - that's non-inverting unless I'm missing something...

I say build it as you have it drawn, as non-inverting and deal with the NFB, if you apply it, as you must.

 

[qusp]

I think its a red herring, ask the engineer if he used an inverting or non-inverting signal path when he produced the album …

 

[wakibaki]

Here's the sim:-
 
 

 
Green is Vout, blue is Vin and red is the grid of U3, the output stage. Now look at the phase dots on the OPT and how I've got it connected. I've had to take B+ to the bottom of the transformer to get it to invert, it's naturally a non-inverting stage, which with the first stage being inverting would make the whole amplifier invert.
 
I first noticed this when I connected up the feedback and the gain went UP.
 
Or perhaps I've got something completely bassackwards?
 
w

 

[wakibaki]

Quote:

I think its a red herring

 
I'm inclined to agree qusp, but I note from Doug Self's Small Signal design book that it's the practise in professional equipment such as mixing desks to preserve polarity. I always make the effort to do so now in my solid state designs, it rarely costs anything other than paying attention, but now that I have started paying attention to it, it feels a bit odd to deliberately go against the grain, which is what I'm tempted to do here, because I'd rather feed back an inverted output into the high impedance non-inverting input instead of taking a non-inverted output back into the comparatively low impedance at the cathode, to say nothing of the potential complications of providing a second DC path (and magnetising current) to ground via the transformer unless it's completely AC coupled.
 
w

 

[elliottstudio]

It is common practice in power amps to take the feedback from the output winding to the cathode of the input stage (or appropriate stage) through a divider, assuming you intend to take the feedback from the secondary of the OPT. Your headphone amp is essentially a power amplifier, though very low power. As for voltage at the cathode across the secondary causing a magnetizing current; the secondary winding is so small it's DC resistance will effectively shunt it to ground. The feedback divider at the cathode will isolate it from this shunting resistance and greatly limit the current that could actually shunt through the secondary so it should have little to no effect on the bias setting.

 

[wakibaki]

So do I take the B+ to the blue as in the simulation, or to the brown as in the schematic to yield an inverting stage, and hence a non-inverting amplifier overall?
 
I've searched all over the place, but I can't find a definitive answer as to whether a transformer-coupled common cathode amplifier is normally inverting, although I suppose a sensible answer to the question would be, 'it depends which way you connect it up'. I would expect that if the phase dots are at the same ends of the transformer (brown connected to B+), that as the voltage at the grid rises, the current in the primary would rise and this would result in a rising current in the transformer secondary (non-inverting stage with black connected to ground) but as per the simulation, which I'm inclined to trust, the phase dots must be at different ends of the transformer (blue connected to B+) to get an inverting stage.  It's when the stage is resistively loaded that the rising current results in a greater voltage drop across the load (falling output voltage at the anode-resistor junction) and the stage is inverting.
 

 
w

 

[elliottstudio]

There is no phase reversal across a transformer if hooked up in phase. Your amp is two stage so it is non inverting and will still be so on the secondary. If you want to apply NFB you can go to the cathode of the first tube from the secondary winding, as you know. Since you have to have one side of the secondary grounded for the feedback to work you can't reverse the phase on the primary or you will be in 'buck' mode and lose signal (this being an SE amp). If the output was not providing NFB you could float the output and hook either the primary or the secondary with a phase reversal.
 
At the power level you're going for global NFB is probably not necessary, and taking it from the transformer secondary introduces phase shifts and other nasties. You may consider looping from the plate of the second stage to the cathode of the first stage. Yes, you would need AC coupling but the series resistor at the cathode of the first tube would necessarily be large enough that the coupling cap could be a small value film cap. Or you could run with no loop feedback and adjust your gain structure accordingly; you may want to reduce the drive voltage from the first stage plate to second stage grid with a divider for best linearity.
 
The simulation is confusing. Leave the primary leads long enough so you can build it either way and prove it out with the secondary floating. I hope I'm answering your questions.. I'm sure others will chime in.

 

[wakibaki]

Quote:

I've got a lot of capacitance in there, I wish I'd known, these caps are huge and it's crowded already.
 

Quote:

Terrible. The best things for keeping a SE tube amp quiet are space between magnetic parts, and a regulator.

 
Yes, but too small a cap before the choke results in a reduced HT supply. This is a common ploy for reducing B+ without wasting power across a resistor. The old 'small capacitor ploy'. I like a nice small capacitor ploy, but not when it results in a reduced Pink Panther, nicht wahr?
 
w

 

[elliottstudio]

Quote:

 
Yes, but too small a cap before the choke results in a reduced HT supply. This is a common ploy for reducing B+ without wasting power across a resistor. The old 'small capacitor ploy'. I like a nice small capacitor ploy, but not when it results in a reduced Pink Panther, nicht wahr?
 
w

I too have used the small capacitor ploy in the past but it never reduced my Pink Panther

- Now I just have custom transformers wound so I end up with the correct B+ in the first place! And in the case of my preamps regulators are used.

 

[wakibaki]

Had to order some TO220 insulator kits, I'm waiting for them...
 
w

 

[elliottstudio]

 

[wakibaki]

Sorry, I went downhill again a bit with the chest infection, but should be finished tonight~tomorrow. Had a routine follow-up chest x-ray a couple of days ago tho', still in the clear with the big C. 

 
w