Pseudo-Triode ST35 Notes: SET aside?

[Shannon Parks]

Late last year I built my first Rev C board and set it on my bench, ready for a couple hundred hours of 'burn in'. My old Rev B test amp was now freed up for some experimentation.

Here are the modifications from a stock diytube ST35 using a PA774 power transformer and Z565 outputs (should hold true for any true ST35 or derivative):
-Ultra-Linear taps are disconnected
-solder a 100 ohm, 1W resistor directly on the PCB between pin 7 & 9 (plate and screen) on the EL84s
-disconnect the 150K resistors in Hafler's linearization network (we WANT even harmonics)
-screen feedback capacitors removed
-leave the rest of the negative feedback circuit alone
-for extra 120Hz hum reduction, I replaced the 50 ohm power resistor with a Dynaco C-354 choke (1.5H, 56ohms DCR)

Note that the Z565 outputs have a 7100 ohm primary. I have a Hammond 1608 and 1609 coming which will be used in this same test. That being said, we are not exactly running the EL84s per datasheet spec.


OPEN LOOP DATA with 100mV INPUT

020Hz 3.50V +30.8dB 2.318%THD+N 1.531W @ 8ohm
025Hz 3.55V +31.0dB 2.529%THD+N 1.575W @ 8ohm
030Hz 3.58V +31.0dB 2.591%THD+N 1.602W @ 8ohm
050Hz 3.63V +31.1dB 2.758%THD+N 1.647W @ 8ohm
100Hz 3.65V +31.2dB 2.833%THD+N 1.665W @ 8ohm
200Hz 3.66V +31.2dB 2.863%THD+N 1.665W @ 8ohm
500Hz 3.66V +31.2dB 2.862%THD+N 1.665W @ 8ohm
01kHz 3.64V +31.2dB 2.809%THD+N 1.656W @ 8ohm
02kHz 3.58V +31.0dB 2.689%THD+N 1.593W @ 8ohm
05kHz 3.26V +30.2dB 2.266%THD+N 1.320W @ 8ohm
10kHz 2.59V +28.2dB 1.660%THD+N 0.839W @ 8ohm
20kHz 1.69V +24.5dB 0.996%THD+N 0.356W @ 8ohm
30kHz 1.17V +21.3dB 0.654%THD+N 0.172W @ 8ohm
40kHz .864V +18.7dB 0.559%THD+N 0.093W @ 8ohm
50kHz .658V +16.3dB 0.642%THD+N 0.054W @ 8ohm
60kHz .502V +14.0dB 0.806%THD+N 0.031W @ 8ohm
70kHz .376V +11.5dB 1.038%THD+N 0.018W @ 8ohm
80kHz .273V +8.80dB 1.368%THD+N 0.009W @ 8ohm
90kHz .195V +5.80dB 1.863%THD+N 0.005W @ 8ohm
100kHz .18V +5.20dB 2.045%THD+N 0.004W @ 8ohm


CLOSED LOOP with 400mV INPUT

020Hz 3.90V +19.7dB 0.708%THD+N 1.892W @ 8ohm
025Hz 3.91V +19.8dB 0.744%THD+N 1.901W @ 8ohm
030Hz 3.92V +19.8dB 0.772%THD+N 1.921W @ 8ohm
050Hz 3.94V +19.8dB 0.790%THD+N 1.931W @ 8ohm
100Hz 3.95V +19.9dB 0.802%THD+N 1.950W @ 8ohm
200Hz 3.95V +19.9dB 0.806%THD+N 1.950W @ 8ohm
500Hz 3.95V +19.9dB 0.811%THD+N 1.950W @ 8ohm
01kHz 3.95V +19.9dB 0.806%THD+N 1.940W @ 8ohm
02kHz 3.94V +19.8dB 0.826%THD+N 1.931W @ 8ohm
05kHz 3.90V +19.7dB 0.951%THD+N 1.901W @ 8ohm
10kHz 3.79V +19.4dB 1.257%THD+N 1.796W @ 8ohm
20kHz 3.49V +18.7dB 1.671%THD+N 1.514W @ 8ohm
30kHz 3.12V +17.7dB 1.709%THD+N 1.209W @ 8ohm
40kHz 2.67V +16.4dB 1.564%THD+N .8931W @ 8ohm
50kHz 2.27V +15.1dB 1.508%THD+N .6470W @ 8ohm
60kHz 1.87V +13.4dB 1.795%THD+N .4385W @ 8ohm
70kHz 1.49V +11.5dB 2.022%THD+N .2779W @ 8ohm
80kHz 1.11V +09.0dB 2.084%THD+N .1551W @ 8ohm
90kHz .807V +06.2dB 1.945%THD+N .0812W @ 8ohm
100kHz .75V +05.5dB 1.613%THD+N .0701W @ 8ohm


CLOSED LOOP with 700mV INPUT
This is max output before odd harmonics appear. See FFT screenshot below.

020Hz 6.74V +19.6dB 1.413%THD+N 5.678W @ 8ohm
025Hz 6.74V +19.6dB 1.329%THD+N 5.695W @ 8ohm
030Hz 6.77V +19.6dB 1.294%THD+N 5.712W @ 8ohm
050Hz 6.80V +19.7dB 1.229%THD+N 5.763W @ 8ohm
100Hz 6.81V +19.7dB 1.210%THD+N 5.780W @ 8ohm
200Hz 6.82V +19.7dB 1.214%THD+N 5.797W @ 8ohm
500Hz 6.81V +19.7dB 1.216%THD+N 5.780W @ 8ohm
01kHz 6.80V +19.7dB 1.209%THD+N 5.780W @ 8ohm
02kHz 6.78V +19.6dB 1.252%THD+N 5.746W @ 8ohm
05kHz 6.73V +19.5dB 1.551%THD+N 5.645W @ 8ohm
10kHz 6.58V +19.2dB 2.292%THD+N 5.396W @ 8ohm
20kHz 6.07V +18.3dB 3.720%THD+N 4.590W @ 8ohm
30kHz 5.39V +16.9dB 3.990%THD+N 3.618W @ 8ohm
40kHz 4.61V +15.2dB 3.790%THD+N 2.634W @ 8ohm
50kHz 3.83V +13.2dB 3.760%THD+N 1.824W @ 8ohm
60kHz 3.08V +11.0dB 4.150%THD+N 1.178W @ 8ohm
70kHz 2.35V +08.4dB 4.090%THD+N .6927W @ 8ohm
80kHz 1.74V +05.3dB 3.930%THD+N .3728W @ 8ohm
90kHz 1.25V +02.3dB 3.560%THD+N .1959W @ 8ohm
100kHz 1.2V +01.8dB 3.030%THD+N .1888W @ 8ohm

The spurs are the 2nd and 4th harmonic. The odd harmonics are in the dirt.



Epilogue

Not bad SET recreation for a PP amp with negative feedback. One of the sweetest sounding amps I've heard yet, but it doesn't cover the music in syrup. My Joni Mitchell HDCD discs are nothing short of phenomenal, goose bump producing performances. I am IN a small, private club with Joni doing a setlist from Blue. Speaking of which, I've gotta run - her band is coming back from break. Ciao,

Shannon

[dado5]

Hi Shannon,

Thanks for the info!

BTW, What are the gain figures for the amp in the conventional UL mode?

Thanks.
Rob

[Shannon Parks]

Hi Shannon,

Thanks for the info!

BTW, What are the gain figures for the amp in the conventional UL mode?

Thanks.
Rob

Hello Rob,
22dB gain typically - checkout the earlier post 'transformer shootout...' for more info.

Shannon

[EWBrown]

Interesting information, I'll have to give it a try, have some 1620s and 1609s arriving today.

I was doing a side-by-side between an original, recently built (from an old unopened Dynaco ST-35 kit I got back in November 2003) ST-35, and a DIY-35.

First thing I noticed is that the Dynaco ST-35 has a small amount of hum, barely noticeable, but it really stood out as compared to the DIY-35 (revB) that is totally hum-free. also, the ST-35 has some mechanical 60Hz vibration from the original PA774 mounted on the thin steel chassis. If I place the ST-35 on a rubber mat, this mechanical "hum" disappears.
If the Dynaco ST-35 is picked up ahd held, this 60 Hz vibration can be readily felt.

The DIY-35 does have better sonics, and more "impact" as well as no hum, due to the better power supply section, more capacitance, better PC board grounding, etc. The DIY-35 Rev B also has old Z-565s, and a new "Triode" PA774 replacement, so this be the most accurate comparison, as opposed to my other units that use Hammond 272FX and 1650Fs / HWTs.

The ST-35 is using original NOS tubes, the DIY-35 is using all JJ tubes.


/ed brown in NH

[Shannon Parks]

Interesting information, I'll have to give it a try, have some 1620s and 1609s arriving today.

Ed doesn't mention this, but he plans to use the cheap-as-dirt Svetlana SV83 tubes in a triode-strapped config.

Hmm - Ed, I just started thinking about what value resistor you will use here. What were you planning to do to keep the 250V rating on the screens?

Shannon

[sotoole]

with these mods you just have to go with the hexfred diodes they make the amp even cleaner. look at the difference with a scope you will be amazed.
also try a balanced power transformer for another awesome improvement. ( this improved freds OTL amps so much he could not tell they were on even with 100 db speakers. )

[EWBrown]

For the SV83 screens, there are two approaches, 1: add a resistive voltage divider, or 2: throw caution to the wind, as is done with the Decware SE84CS series, and just go from SG to plate with a 1K 2W resistor. This seems to work just fine, in my SE unit, and as they have the tubes overbiased anyway (using a 150 Ohm 5W cathode resistor for two), the tube life in this particular application is expected to be somewhat shortened.

If alll else fails, use EL84s.

/ed

[Shannon Parks]

For the SV83 screens, there are two approaches, 1: add a resistive voltage divider, or 2: throw caution to the wind, as is done with the Decware SE84CS series, and just go from SG to plate with a 1K 2W resistor. This seems to work just fine, in my SE unit, and as they have the tubes overbiased anyway (using a 150 Ohm 5W cathode resistor for two), the tube life in this particular application is expected to be somewhat shortened.

If alll else fails, use EL84s.

/ed

What's the power tranny and B+ are you going to use, Ed? Also, what's the plate voltage in the Decware units? Finally, what do you swag the screen current is - 5mA?

Shannon

[EWBrown]

Probably the same PT that I used in the DecWareClones: the Hammond 272FX. I've used that pretty much universally, except for the previously mentioned "wooden wonder" which is a B rev DIY35, with Z565s and new PA774. I have one 272FX left, I'll at least try it before any permanent chassis mounting takes place. I figure the easy quick-test method is to hang the 1K 2W resistors on the terminal strips between Plate and SG.

I did measure the SV83 SG current long ago, don't immediately recall. Somewhere between 2 and 5 mA?

I found I was 2 PC tube sockets short, so I made another AES order (along with 2 gray painted steel 12X8X2 chassis and bottom plates, and one 12X8 cage).

/ed b