The Red Light District - another PP EL84 amp

[SY]

No, the feedback from the 8 ohm tap takes care of the other taps, too.

[sds2000]

Fran

Awhile back you wrote

"Oh great jumping jesuses!

That looks far more complicated than I thought it would be! I'll never wire all that up!"

I see now instead

" Oh yeah, this will be a "nest-o-snakes" model! "

and

"I just can't see how you could do it a lot neater either!! "

All I can say is imagine how much worse it would look if you tack mounted your LEDs in, had built the screen regulators on turret boards, and run all your ground and filament lines in bright orange so you could spot them easily.

Fortunately, snakes or not as long as it sounds nice (and it will) you can flip the chassis right side up and it will look nice too.

Your build photograph does remind me of a question the about general tube amp wiring practice. When I did ham radio, back when "wireless" had nothing to do with computers, they used to discourage daisy-chaining tube filament connections in favor of multiple parallel connections. Does anyone want to weigh in on if this is relevant advice for audio?

[woodturner-fran]

I'm no expert, but its what MJ seems to recommend in building valve amplifiers - with a few tips like putting the input valves at the end of the chain and wiring PP valves in phase so hum gets cancelled out.

What I probably should have done is run 2 parallel connections to each channel. In any case, I power it up and see how it goes before rewiring it.

Fran

[EC8010]

Originally posted by sds2000
When I did ham radio, back when "wireless" had nothing to do with computers, they used to discourage daisy-chaining tube filament connections in favor of multiple parallel connections. Does anyone want to weigh in on if this is relevant advice for audio?

Either will work.

It's easier to daisy chain heaters, but they have to be done in the right order, input valve last. At each point, the wire must be rated for the current it carries, but as the current changes along the chain it's hard to have the optimum gauge of wire at all points, and you often end up with wire that's too heavy going to the input valve. The significance of the wire gauge is that heavier gauge wire doesn't twist too well, and you want as tight a twist as possible, especially near the input valve.

Conversely, if you use one twisted pair per valve, each going back to the transformer individually, the wire gauge can be much lighter (lower current), a tighter twist is possible, and only one gauge of wire is needed. This method has two further advantages:

From an RF point of view, any RF oscillation generated at one valve has to travel the full length of the heater wiring to the transformer, then out again before it can interfere with another valve. That's a lot of inductance. Add 10n capacitors from each valve's heater pin to chassis and from the transformer heater winding to chassis and you've got some very effective RF filtering.

You now have multiple pairs of wires going back to the transformer, so you harness them neatly together. Each pair of wires (despite its twist) generates a small 50Hz field. The twist of a given pair is unlikely to be perfectly aligned with another so the 50Hz fields add imperfectly. Sometimes they add constructively, sometimes destructively - they're not coherent. To sum sources that are not coherent, we must sum powers. As an example, if we had four pairs, each carrying 0.3A, P = I²R so we sum 0.09 + 0.09 + 0.09 +0.09 = 0.36. We then take the square root to find the individual current that would have given that power, and find it is 0.6A. This tells us that four twisted pairs each carrying 0.3A will generate the same hum field as a single pair carrying 0.6A. But we are carrying 1.2A over our four wires, so we've achieved a 6dB improvement. In practice, you probably won't quite achieve the theoretical improvement, but it's worth trying. A quick way to determine the improvement from using multiple pairs carrying identical currents is sqrt (n); where n is the number of pairs.

The short answer is that multiple pairs is a better way to do heaters than daisy chaining, but it does take quite a bit longer to do because longer lengths of wire are needed and the pairs all need to be laced together neatly in a loom.

[woodturner-fran]

On the power Tx I have, there are 2 heater windings. One is 6.3V at 3A and the other is 3.15-0-3.15 at 1A. At the moment I have all the heaters running off the 6.3V winding, which I know for 4 x EL84 and 2 x 12AT7 is pushing it a bit. Would I be better off sonically and electrically putting the 2 x 12AT7 on the 3.15-0-3.15 winding so that they are separate to the EL84s?


Seeing as I will have to wait for a little until the agilent LEDs arrive, I might as well do the work now rather than later.


What do you think?

Fran

[pengboon]

Sounds like a good idea! Even 4 x EL84 (4 x 760mA) is pushing the 3A winding.

[EC8010]

Definitely move the ECC81 to the 3.15-0-3.15 winding. While you're at it, use thinner (solid core) wire and twist it tighter. Add 10n from each side of the transformer winding to chassis and each ECC81 heater pin to chassis using really short leads.

[woodturner-fran]

Yeah, you know sometimes you need to say something out loud before you realise you don't need someone else to even answer!


So I've already done it. Moved the ECC81 to the other winding and wired each valve out of phase. EC8010: I can add the 10nF from each leg to ground (had already done that on the EL84s anyway).

Antoehr question, I'm sure its been answered many times.

All the circuit grounds go to one point ie what I understand as a star ground. The ECC 81 centre heater pin will now need to go to a chassis ground. Should the star ground point and the chassis ground now all be tied together, also at this one point?


sorry to be asking what must be stupid questions,

Fran

[SY]

Ignore the heater center-tap pins (pin 9); you've got a nice CT on the transformer already. Note that the ECC81 heaters are the ones which must be biased up by the DC string, so the transformer CT is the ideal place to pull it off. Since they're not run push-pull, the heater phase is irrelevant.

Star and chassis ground should be separated by a ground-breaker, generally a pair of back to back parallel diodes (in a 69 configuration).

[woodturner-fran]

OK, so is this correct:


EL84 heaters come from a 6.3V winding, no center tap. Each leg has a 10nF cap, centre point of which runs to circuit ground. Virtual centre tap created which goes to 1/4 B+


ECC81 heaters, 3.15-0-3.15 Tx, Each leg has a 10nF cap centre point of which goes to ground. Connection from centre of this winding to 1/4B+. Nothing wired from pin 9 of the ECC81 socket.

Is that right. I'm kinda getting confused!!

Fran