[Patrick Turner]

I mentioned that ........

I have very nearly finished re-engineering a pair of Dynaco MkIV monos.

** Question for Pat.


The Dynaco Mk IV uses two EL34s in UL mode with a 470V supply - OK.

The published data for that UL of operation includes THD curves that barely
exceed 1% below clipping - with NO feedback applied .

Well, a number of things affect published data which we always must question about conditions for the test unless fully stated. The only time anyone might achieve 1% THD at near full Po with any UL% taps between pure pentode/tetrode or triode will be with PURE class A1, and usually with only one load value, and with matched OP tubes and a distortionless driver amp.

In nearly every brandname PP amp I have tested, class AB operation is always used to boost the Po ability to get better amp sales. The most "honest test" is done using the rated load connected to the labelled output ohms, so with say 8r0 connected to an outlet labeled "8r0".

Consider MkIV Dynaco. I recall B+ = Ea = +440Vdc. If the EL34 have Pda = 22Watts each, then Ia = 50mA, and if Ig2 = 6mA, then total Pda + Pg2 = 24.6Watts, which I consider a bit too high. But anyway, Total Pda = 2 x 440V x 0.05A = 44Watts, and Class A anode efficiency = 45% so max Class A1 Po = 19.8Watts.

The approximate RLa-a for pure class A1 = 15,840 ohms. The MkIV has OPT with nominal RLa-a = 4.300 ohms for 4r, 8r, 16r secs, and let us say we choose to use the 4r outlet, where ZR = 1,075:1. The load we would need to use to get RLa-a of 15k8 = 15,840 / 1075 = 14r6, ie, more than 3 times the labeled outlet value. Load line analysis will back up everything I say here. Now with 14r6 at 4r0 outlet, perhaps THD might be under 2% at 19.8Watts, with no GNFB.
But with 4r load at 4r0 outlet, the pair of EL34 struggle to have THD under 4%
at clipping, and at whatever rather high Po that might be. Using lower B+ and KT88 lessens the THD for same or slightly higher Po.

However, Dynaco claim virtually the same THD figures, with circa 20dB of
feedback applied ??

I searched around for data on MkIV and I didn't find any info saying THD was the same with or without NFB. Perhaps you found a more informative source of info. The info I have talks about IMD, not THD, and there was nothing surprising, but then also no exact statements about how tests were made.

Is the 7199's pentode gain stage mainly responsible for this anomaly ??

The pentode section of 7199 has to make Va slightly exceeding Vg of one EL34.
This is about 26Vrms at clipping, and THD will definately be over 1%.
And many H are present and its adds to THD of class AB EL34, so expect 3% or more without GNFB. The 20dB NFB reduces this to about 0.3%, and this is typical real world behaviour.

For comparison, the Quad II use two, low cost, EF86 pentodes in
differential mode to provide gain and phase inversion. The output stage uses
local (cathode) feedback, plus overall feedback but is otherwise
conventional PP with KT66s or 6L6GCs performing equally.

The EF86 of Quad-II are not quite in pure differential mode. In fact, the pair of EF86 are set up as a paraphase pair where output from Va at one EF86 is resistored down to feed the grid of other EF86, and common cathode R is bootstrapped by the GNFB resistance network. Its all so very clever, this arrangement, and hardly anyone fully understands the several things going on simultaneously, to achieve the use just two little piddly EF86 to drive a pair of KT66. The Quad-II set up feeds the distortion of one EF86 to the other where the odd order H add while some even order cancel. The Quad-II driver amp has more THD than if twin triodes similar to Williamson, or Leak, or anything at my website.
In one pair of Quad-II I re-engineered, I set up the EF86 as a true differential amp. I set up a -400Vdc supply from which I had a large value R from -400Vdc to commoned cathodes. The GNFB is fed only to the grid of secons EF86, and there is no paraphase connection. This halved the THD. But in 1950, the idea of -400V or even -40V and a CCS to cathodes was just not on - too complex, too pure, too good, and too damned expensive while the rest of the world's amp industry colluded to make the lowest common denominator product for maximal shareholder returns.

The Quad II achieves 0.1% THD at rated power and 0.05% THD at circa 10 watts
or less.

Well all that depends on load conditions, and perfect tube matching, and almost no Quad-II amps I've tested achieve these wonderful figures you quote.
Usually, with perfect test conditions, the tube matching makes sure the 2H current production is so close that whatever difference there is remains less than the amount of 3H which is always the major H in what you see in PP amps
while still working in class A1. When the action becomes class AB, then you get far more than 1% and an an increasing number of odd and even H and PSU noise.

I very much liked Quad's idea of cathode feedback windings on the OPT. I recall making my first 8585 amp and I originally had 4 x GE6550A in each channel with Ea = 400Vdc, Ia at 55mA, and I saw 50Watts at 0.7% THD without GNFB but with partial UL connection of screens and with 12.5% CFB
I found I could raise Ea to 480Vdc, and used a fixed Eg2 like Quad, but lower Vdc than Ea, and still get excellent THD results which were consistently better than triode, or any % of UL.

Patrick Turner.