EL84 Amp - Baby Huey

[Soonerorlater]

rhone,

I vote build it as is, and if we are lucky we will have closure by building something else....

[UnixMan]

Hi Ian. Sorry in case the question was already answered. Way too long a thread.

Originally Posted by gingertube

The schematic is pretty much like that of post 602 on page 61, feedback mod as stated above,

for the sake of curiosity. Given that you use local NFB a la a "Schadeode", why have you connected the g2 to the UL tap instead of a fixed voltage as is done in original Schadeode?

Have you tryed it that way too?

[TheGimp]

The discussion begins around page 61, post 609.

If you look in the url for this page, the last portion is ...-104.html

Replace 104 with 61, hit enter, and scroll down to post 61.

[gingertube]

UnixMan,
I haven't tried any of the Baby Hueys withe Pentode Mode Output Tubes, I have always added Ultralinear connections into the mix.
That is purely so that I can avoid using any global feedback.

The original circuit on which all the BH's are based is by Yves Monmagnon (Yvesm) and he did use pentode mode with global feedback.
See post 504 on page 54 for Yves circuit using 6GW8 (ECL86).

Cheers,
Ian

[UnixMan]

I wonder if there is any difference. With Schadeode, the closed-loop gain of the output stage is fixed by the ratio of the divider formed by the NFB resistor and the output impedance of the driver stage. Using UL, you reduce the gain of the output tube WRT the same tube used in plain pentode mode, so basically you add some local NFB via g2 (UL) but remove some (same amount?) from the "Schade" loop. The other way 'round if using pentode mode.

I have not done any math to check wether there is any net difference in the total amount of (local) NFB in either case, but I guess there is likely none.

On the other end, avoiding UL (via OPT tap) should avoid the phase-shifts introduced by the OPT, thus in principle the Schade local NFB loop should work better than the UL... :-?

[gingertube]

A bit "tongue in check" - Both methodologies linearise the output tubes and reduce rp. So using both together should be twice as good as either alone - right?
Cheers,
Ian

[UnixMan]

Originally Posted by gingertube

A bit "tongue in check" - Both methodologies linearise the output tubes and reduce rp. So using both together should be twice as good as either alone - right?

well, no. It's quite the contrary!

Both UL and Schadeode are just two different methods of applying (local) NFB. UL applies some local NFB via g2, Schadeode does the same via g1.

From NFB theory, the amount of linearization and output impedance reduction gained with NFB is proportional to the amount of NFB applied; that is, to the amount of "gain reduction" devoted to NFB.

With Schadeode (as far as you have enough open loop gain) you reduce the overall (closed-loop) gain to a fixed ratio, which is set by the divider formed by the feed-back R and the eq. output impedance of the driver stage.

If you use UL, the open-loop gain reduces to the gain of the tube in UL mode, which is much less than the gain of the same tube used in pentode mode (that gain difference is what gives you the gain in term of linearity and rp reduction of the UL).

So, what you get with Schadeode+UL looks like a couple of nested NFB loops.

BUT! the total open-loop gain available is always the same: it is the gain of the tube when used in pentode mode.

Again from NFB theory it can be demonstrated that, for a given amount of total gain available, a single overall loop (in this case plain Schadeode) is more effective than a number of nested loops (in this case UL+Schadeode).

Thus UL+Schadeode is definitely worse than just plain Schadeode (pentode mode). At least from a "numerical" point of view.

Whether that's also better from a sound quality point of view is another story. Parameters like THD and output impedance do have little (if any) correlation with the perceived sound quality.

To determine which way does actually sound better you need to try and listen for yourself...

[Ian444]

Originally Posted by gingertube

The BH design has evolved over time by pure listening. Sure I looked at the theory and what might happen if I did this or that but the determining thing for what stayed in and what was rejected was based on just listening tests not measurements (this was a deliberate policy - I could have done all the measurements, after all I'm a Electronic Design Eng. in the day job, but I went this way based upon the fact that there appears to be very little correlation between what sounds good and the sort of measurements which are "traditionally" done, i.e. THD etc).

Cheers,
Ian

Edit: from post #1015

[Rene]

I was curious after the Baby Huey EL84 so I build it. But without global feedback. I have removed R1 and R2 and connect the lower 1K resistor to earth. See picture. I hoop this is all right.

But nevertheless I encounter a problem. Ik have an oscillation of 250KHz on the anode (and UL connection) of the EL84. Input signal is 0V. If I load the LS output with an 8 ohm resistor or lower the oscillation is gone. To me this is not good and the amp seems to be a bit unstable.

Have I done something wrong? Must the grid of the lower ECC803 connected to earth with a resistor of 1M?

BTW, in triode connection I don't have this oscillation problem.

Thanks in advance.

Regards,
René

[gingertube]

Rene,
Your changes should be fine.
To get rid of the oscillation in Ultralinear Mode you will need to increase the 150 Ohm screen resistors. These have a grid stop function and the body of the resitors need to be hard up against the tube pin for the screen connection. Try 470 Ohms or even 1K in lieu of the 150 Ohms. That should fix it.
Cheers,
Ian