Push-pull microphone preamp with UTC LS and Tamura transformers

[aazaa]

Optimised voltages and currents. The third harmonic distortion peak.

When the current of the ecc82 is increased slowly, you come to a point where the third harmonic distortion peak is very low. The dissipation of the tube easily reaches 2W/triode (max. is 2.5W). So the same experiment was done again with less voltage across the tube.

Surprise! With 150V across the test tube and a current of about 8mA/triode, the third harmonic distortion peak drops below -100dBu. Idem for the 12kHz and 15kHz IMD peaks. Once over that point, the distortion increases.
Dissipation is a very reasonable 1.2W/triode.

New B+ 325V. Stabilised B+ ! Fluctuation of B+ has to be kept as low as possible.

The high tension of the unregulated PSU fluctuated 15 à 20V depending of the hour of the day.

Regulation with a few zeners and 4 transistors: bc550c, bc560c, 2sb649ac and stp9n65m2. The zeners are fed by a very stable current source. No more chokes.
Current is limited to 100mA. Fluctuation is less than 0.7V.

Adapted anode and cathode resistors.
A 100k trimpot is adjusting the current drawn by the j-fets / the ecc83. The voltage across - and the current drawn by the ecc82 is set by the changed voltage at the anodes of the ecc83.

Nice to see that adjusting a trimmer at the very beginning of the chain reduces considerably the distortion of the tube at the end.

The adapted circuit:

Below IMD and THD of the old Philips and Mullard ecc82 tubes. Incredible performance.
Tests with a french 12au7WAH Cifte tube of recent production: IMD 0.003% THD 0.004%.

Paul

 

[abbey road d enfer]

Very impressive, considering there's no global NFB.

 

[aazaa]

Very impressive, considering there's no global NFB.

Indeed Abbey. The little ecc83/ecc82 tubes surprised me.
Getting equally good performance with 6sl7/6sn7 will be a real challenge.

Paul

 

[aazaa]

Playing with harmonic distortion.

The transparent sound of a tube preamp with almost no distortion is nice, but it can also be fun to add some extra colour. With the hybrid PP- preamp it is possible to increase the even harmonic distortion while keeping the odd harmonics very low.

A little experiment.

Overview of the adjustments:
2k trimpot: unbalance between the anodes of the ecc82: 0.00V
20k trimpot: unbalance between the cathodes of the ecc83: 0.00V
1M trimpot: second harmonic distortion turned up to -60 dBu
100k trimpot: odd harmonics lowered to below -110 dBu


THD is 0.13% almost pure even harmonic distortion.
And yes, fans of an uglier sound can do the inverse, low even and more odd harmonics.

A great control not only over the level, but also over the character of the distortion.

Paul

 

[Fdieck56]

I am curious what your drain to source DC voltage is for the FET Pair input as I believe you are operating with less than the 10 volts on the data sheets for the given capacitance figures. The 2.7K resistors are gobling up more the Vds and the increased impedance seen by the drains is negating the cascode effect of the tube cathodes for the FET drains. 500 ohm gate resistors will really help reduce oscillations RFI susceptibility, and interaction with transformer leakage inductance. I would reference the grids of the ecc83 to get 12 to 15 volts Vds for the fets And get rid of the 2.7K resistor.

https://www.alldatasheet.com/view.jsp?Searchword=2SK389
https://audioxpress.com/article/JFETs-The-New-Frontier-Part-1

 

[aazaa]

I am curious what your drain to source DC voltage is for the FET Pair input as I believe you are operating with less than the 10 volts on the data sheets for the given capacitance figures.

The psu has been rebuilt very recently. The 3 Talema pot transformers were quiet but not dead quiet. The only downside of the preamp.
I decided to put the primaries of two transformers in series and run the third at a lower voltage. Voltage doublers both for heaters and B+. Maybe not ideal, but better than having to buy new transformers.

Now the psu is dead silent. Even when switching on, not the slightest vibration is audible. Only inconvenience: the high voltage has dropped to 288V. I will have to do a little revision to have the same ideal balance as before.

I can't give you the exact measurements for the posted preamp at B+ 325V.

The experiment with the increased second harmonic distortion is done at B+ 288.5V. Changing slightly the cathode resistors resulted in the same low distortion figures as before.

measurements:
cathode ecc83: 23.35V
Vds LSK389: 14.15V

You can add gate resistors if you want, but in my build it really isn't necessary.
Functioning of the 2k7 and 68k resistors is explained in post #98.

Paul

 

[Fdieck56]

I really don’t understand how you get a DC cathode voltage of 23 volts out of a grounded grid ecc 83….. I mean I REALLY don’t understand it. With the stated drain voltages for the jfets a 2.7K drain resistor might near the “sweet spot” for distortion. I can understand not wanting to explain all the intricate details of the design of a nested feedback circuit, but please don’t take my comments as just putting my two cents in. I am a BSEE and have been designing with jfets since the mid 80s and interested in your design. Thank you for your reply.

Fred

PS I went back looked at the circuit again because I figured I MUST be missing something….. I see the grids are not grounded but lifted by DC through the plate connected feedback network. “The whole interaction of all the elements is too complicated to describe“ might be closer than I thought to summing things up. Thanks for stirring up the gray matter, I think following current events has turned my brains to mush.

 

[aazaa]

I really don’t understand how you get a DC cathode voltage of 23 volts out of a grounded grid ecc 83….. I mean I REALLY don’t understand it. With the stated drain voltages for the jfets a 2.7K drain resistor might near the “sweet spot” for distortion. I can understand not wanting to explain all the intricate details of the design of a nested feedback circuit, but please don’t take my comments as just putting my two cents in. I am a BSEE and have been designing with jfets since the mid 80s and interested in your design. Thank you for your reply.

Fred

PS I went back looked at the circuit again because I figured I MUST be missing something….. I see the grids are not grounded but lifted by DC through the plate connected feedback network. “The whole interaction of all the elements is too complicated to describe“ might be closer than I thought to summing things up. Thanks for stirring up the gray matter, I think following current events has turned my brains to mush.

Indeed Fred, you're right. The grids are not grounded.

Forgot to specify that the LSK389 used is the LSK389A

Maybe also this:
I always tune my preamps to the lowest distortion!

Paul

 

[aazaa]

Tune the ecc83 - or not, a different approach.

First this: there is no necessity to adjust the unbalance between the cathodes.
Without the 20k trimpot it is still possible to turn the distortion down to the same low levels with the 100k and 1M trimpots.

I knew it from the start. Nevertheless - driven by some obscure intuition(?) - I prefer to trim the cathode unbalance to 0.00V.

With my tubes - not selected ones - the 20k trimpot was turned almost all the way to one side. The anode resistances were near 180k and 200k.

A different approach: Why not adjust by means of a small correction of the 56k grid resistor on one side? Since the 20k trimmer was already installed, I combined both ways.

First channel: I adjusted the 20k trimpot to have equal anode resistances. A 10M resistor in parallel with one of the 56k resistors reduced the unbalance to only a few millivolts. A small adjustment of the 20k trimpot was enough to set the unbalance to zero.
Correcting this way resulted in an exactly equal level of the mirrored signals at the anodes of the ecc83.

Second channel: 6.2M parallel with 56k (4.7+1.5M). Same result.

But - as mentioned - you can forget about it, omit the ecc83 unbalance correction and reach the same low THD level by trimming simply the 100k and 1M pots. (The negative feedback correction with the 1M trimpot has the ability to reduce the unbalance distortion to almost zero.)

I like my approach more, but have no convincing argument for doing it my way.

Paul

 

[aazaa]

The circuit adapted to the new approach.

The correction of ecc83's unequal triodes is done by adjusting the 2k pot at the grid resistors.
I added two 10k resistors to protect the LSK389A from overvoltage in case of a defective trimpot.

Paul

 

[groovyloop]

Very interesting schematics! What is the best place for the "gain" potentiometer?

 

[aazaa]

The "gain" potentiometer is at the output in my preamp.

 

[groovyloop]

Wow! But what if the signal is too loud? I thought 68k between cathodes ECC83 or just after input transformer stereo pot, no?

 

[aazaa]

There is a lot of headroom.

I would keep a potentiometer at the output. I prefer to use a stepless one, as in old days.
A rotary switch (2 or 3 steps?) to change the value of the 68k resistor could provide an additional option to choose the effective gain of the preamp.

Paul

 

[groovyloop]

Thank you, sir! It is very helpfull! I think pp schematics gives better response some time

 

[analag]

The circuit adapted to the new approach.

The correction of ecc83's unequal triodes is done by adjusting the 2k pot at the grid resistors.
I added two 10k resistors to protect the LSK389A from overvoltage in case of a defective trimpot.

Paul

Nice FET input pre. Why not let a higher transconductance tube sit on the FETs head.

 

[aazaa]

analag said:
Nice FET input pre. Why not let a higher transconductance tube sit on the FETs head.

Thanks and... indeed, why not experiment with other tubes, transformers and a changed value of B+?

Sticking with the ecc83 and ecc82 tubes helped a lot to evolve the preamp to the circuit as posted in #110.
The performance of the preamp can be seen as a kind of rehabilitation for the ecc82 / 12au7 tube.

Paul

 

[aazaa]

Sylvania 6SN7GT and RCA 12AU7A

A noval-octal adapter (6V) was made to test two magnificent Sylvania JAN CHS 6SN7GT / VT-231 tubes (black plates). Tests with several different cathode resistors (values between 11k and 22k).

The distortion is consistently low at all operation points. Intermodulation distortion can be turned further down with the 1M pot, but the 12 and 15kHz peaks cannot be lowered by changing the operation point with the 100k trimpot, nor can the third harmonic distortion peak be turned down.

Sylvania 6SN7GT IMD - best possible value - 0.007%
With the IMD tuned to 0.007% THD is 0.01%.
Similar results with both tested tubes.

The Philips ECC82 tube - IMD 0.0015% THD 0.0011% - outperforms the 6SN7 easily when it is tuned to one of the optimal operation points.


RCA 12AU7A
Beautiful tubes with long plates, recently bought on ebay.
IMD 0.0027%, also better than Sylvania 6SN7GT.

The preamp is tested on linearity with the RCA tube. Before testing the 2.2µF between the cathodes was replaced by a 3.3µF cap.

Conclusion:
- Am I wrong to suppose that differences cannot be heard at such low distortion levels?
- The 6sn7 is performing very well at all operation points.
- When the ecc82/12au7 tubes are tuned to one of the optimal operation points, distortion is exceptionally low.
All the ecc82/12au7 tubes tested could be tuned to outperform the 6sn7. Best results (marginally better) with tubes of the Philips family: Philips, Valvo, Mullard.
- Excellent linearity of the preamp with RCA 12au7a.

Paul

 

[emrr]

Interesting, this contradicts commonly held beliefs about 6SN7 versus 12AU7. Most in fact seem to think of 12AU7 as horribly distorted.

I wonder if 5692 fares any better.

 

[emrr]

I figure transformer distortion is easily higher and swamps the tube contribution. That's my takeaway from years of listening to many varieties of old things.