Push-pull microphone preamp with UTC LS and Tamura transformers

[MaxDM]

Indeed, I mentioned this already in post #8:
"Moreover I like the idea of using opposite triodes correcting each other when wired in a long tail circuit. Idem dito for the 'crossed feedback' connecting the opposite sides."

In my opinion this is one of the qualities of the design...

Interesting paper!
Conclusion of the paper: " In a class A amplifier the use of a bypass capacitor generally reduces the intermodulation distortion, although it may either increase or decrease the harmonic distortion.
The decision as to whether or not to use such a capacitor depends mainly on the magnitude of the distortion. If it is very small, then it may be safely left off, with no possible audible difference."

The 2 charts (5kHz and 10kHz) of fig.9 show that when the output is low, - 6V with a charge of 15 ohm - the intermodulation distortion without bypass capacitor is lower in both cases. A low output of 2.4 Watt has lower intermodulation distortion without cap. Most of the music information played in a normal living room with reasonable high rendement speakers is within the first couple of watts.

In the microphone preamp the maximum output level used to record is low. Average level very low.

Not only the unbypassed common cathode resistor is playing a correcting role. The interplay with the cross-coupled NFB is important. This NFB is also influencing the behavior of the double j-fet. The voltage on the drain on both sides is constantly changing with the signal.

Whatever, the output connected to the RME AD convertor is measuring flat on scope. I'm pleased with the result when recording ...

And... of course it is easy to bypass the 16k5 resistor or to connect 2 separate 33k resistors on the cathodes and bypass both.

Paul

Yes, the IMD issue is probably not such a great deal at these voltages and currents.

I would still be concerned in driving an output transformer with a cathode resistor that large. The maximum current available is limited by it, in addition to the plate resistance summed.

Better would be to use a combination of a large cathode bypass, and adding whatever negative feedback from the anodes to the grids in whatever way you prefer, which could be cross-coupled as well, IMO.

 

[aazaa]

Yes, the IMD issue is probably not such a great deal at these voltages and currents.

I would still be concerned in driving an output transformer with a cathode resistor that large. The maximum current available is limited by it, in addition to the plate resistance summed.

Better would be to use a combination of a large cathode bypass, and adding whatever negative feedback from the anodes to the grids in whatever way you prefer, which could be cross-coupled as well, IMO.

The paper is rather convincing not to bypass the cathode resistor. The better values without the capacitor for intermodulation distortion below 2.4 Watt are not in the charts. Harmonic distortion is uncertain at the low frequencies and: "At the higher frequencies the addition of the capacitor quite consistentely increased the distortion."

I don't know the load you want to use. We don't have to drive a speaker.

Maximum usable current with the 16.5k resistor is 150/16.5=9mA, with or without the capacitor.
If more current is wanted a lower resistor can be used.

About the voltage needed: my RME converter is happy with 2V to have full output.
With a load of 600 ohms at the low impedance output I measure maximum 18V 20Hz-43kHz. Same limit with or without the bypass capacitor. At higher frequencies there is a lot more output.

Limitation is in the capacity for the lows of the output transformer. Raising the current is giving only little improvement. With the capacitor there is a possibility to go slightly higher at 46Hz, not at 20Hz.

With an output of 3V at the fixed output (no bypass capacitor, 600 ohm load): 8Hz-43kHz.

I don't use the fixed low impedance output to connect the variable output.
Input impedance of the RME is 10k. More logical and possible to use the other one. There is a better ratio, a lot more headroom.
Tamura transformer ratios: input 4/4, output 5.7/1

Thanks for the reaction, MaxDM!


Paul

(And: If you want to draw maximum current, it is a bad idea to use 2 separate 33k resistors.)

 

[MaxDM]

Well, I would look at square wave response with the 16.5K resistor and pay particular attention to the sharpness of the corners of the square wave, and possible ringing.

The maximum current flow is due to the sum of the cathode resistor, and the resistance of the plates as well.

But if you like the way it sounds, that's what's important.

 

[aazaa]

Well, I would look at square wave response with the 16.5K resistor and pay particular attention to the sharpness of the corners of the square wave, and possible ringing.

The maximum current flow is due to the sum of the cathode resistor, and the resistance of the plates as well.

But if you like the way it sounds, that's what's important.

No visible difference at square wave response with or without cap...

And yes, it's all about the way it sounds!

 

[aazaa]

The maximum current flow is due to the sum of the cathode resistor, and the resistance of the plates as well.

Agreed MaxDM. When choosing the tubes the internal resistance has to be low enough and the maximum dissipation high enough.

In this configuration, when capable tubes are choosen, it is not the sum of the cathode resistor and the resistance of the plates which is decisive for the maximum current flow.

In the preamp the common cathode resistor is functioning as a current source.
The choice of this resistor is decisive for the current trough the double triode.
In the preamp 150V/16.5=9mA.
If a larger 22k resistor was choosen: 150V/22=max 6.8mA, with 33k only 4.5mA, with 47k a low 3.2mA...

Same result with different tubes having different internal resistances.

Paul

 

[aazaa]

In a new experiment, the LSK389-circuit was built at the second channel with the UTC LS12 (1:14 ratio). To reduce partially the extra gain of the transformer, 390k resistors were added between gates and cathodes on both sides of the LSK389/ECC83. (To avoid the additional NFB the values of the 2k7 and 22k resistors can be changed.)

I suppose the lowest possible noise is reached now. There is 5dB more noise when a 180Ω resistor is connected at the 600Ω input than with an open input. (Visible on computer with the RME app)

No more 40kHz. A roll off starts at 16.5kHz. Was expected, the LSK389 has 25pF at the input. When a 1:10(Ls10) or a 1:14(LS12) transformer is used, the LSK489 (4pF) would be a better choice. The tiny bit more noise of the LSK489 is largely compensated by the gain of the transformer.

A 1:4 transformer at the input and the LSK489 as monolithic matched dual J-fet is probably the best choice.
No additional NFB or changing of the resistors necessary + a better possible performance of the transformer/preamp.

There is also the possibility to keep the UTC LS30 (1:1 ratio) and add a parallel dual J-fet for lower noise.

Paul

 

[aazaa]

Circuit of the hybrid version in post#70 got a minor update.

 

[aazaa]

The first hybrid 2-channel preamp is finished with 2x UTC LS30. Not the lowest noise possible, but low noise and performing/sounding very well. It will be my reference for experiments when building the second preamp.
- fixed 68k and 1M resistors
- connection of the RME converter at the variable output

I did not yet try the 200Ω input of the LS30. The microphone was always connected at the 600Ω input.

Project for a second 2-channel preamp with UTC LS51 output transformers.
As input transformers I have a choice between UTC LS30x, LS10 and LS12, all bought years ago.
LS10 and LS12 can be wired with parallel secondaries, so ratios 1:5 or 1:7 are possible. I only tried the LS12 wired at 1:14.

I'm taking a break before building the second preamp.

Paul

 

[solkatten]

Great design And a very intresting set of experiment... keep going...

 

[aazaa]

corrected negative feedback - intermodulation distortion

Connection of my new RME Babyface Pro FS - input impedance 2k - to the fixed low impedance output of the preamp. A 1k5 resistor is added at this output to reduce the slight rise at 20kHz . The preamp is driving a load of 1k5//2k=857Ω.
Below the frequency response 20Hz-20kHz.

Intermodulation distortion.
The intermodulation distortion is low with or without the use of a bypass capacitor for the cathode resistor.
With the bypass capacitor there is systematically a few dB's more high frequency noise. I tried a 2,2µF film cap, 22µF and even an overly large 220µF capacitor.

Measurement of intermodulation distortion of 13&14kHz at -6dBu level.
- with 22µF capacitor: 0,020%
- without capacitor: 0,0097%
Values are so low nobody will be able to hear the difference.

Corrected negative feedback.
Results are below audible level, so irrelevant but the experiment is interesting and was done only for fun.

The level of intermodulation and harmonic distortion can be controlled and further reduced by correcting the negative feedback. (Corrected NFB, not added NFB!)

While connecting on and off the bypass capacitor for measurements I had following idea: Why not use the tiny AC-signal at the 16k5 cathode resistor to correct the negative feedback instead of suppressing this signal by a large bypass capacitor?
The simple idea turned out to be very effective. All I needed was a trimpot, a resistor and a 100n capacitor.
The 1kHz intermodulation distortion peak could be turned down to 30dB(!) below the level of the already low 1kHz peak with the 22µF bypass capacitor.
This low level of distortion is fluctuating with changing temperatures in the tube. Adjustment has to be done carefully with sufficiently hot tubes.
Thanks MaxDM for your insistence on bypassing the cathode resistor! Without our discussion I would never have had the idea for the corrected NFB circuit.

Below IMD measurements with bypass capacitor, without capacitor and with corrected negative feedback.
Clearly visible: a few dB's more high frequency noise with the capacitor, different distortion peaks at 1kHz.

Paul

 

[aazaa]

harmonic distortion

measurements of the harmonic distortion at 1kHz, -3dBu level
- with 22µF capacitor: 0,033%
- without capacitor: 0,0087%
- with corrected NFB 0,0079%

Again low values, all below audible level.
The third harmonic distortion peak is a few dB's lower with the capacitor. The only pluspoint for the capacitor, unfortunately also not audible.


The measurements with 22µF capacitor, without capacitor and with corrected NFB.
The 2kHz peak is below -100dBu with the corrected NFB.

Paul

 

[aazaa]

second channel distortion

The corrected NFB circuit is integrated in the second channel.
IMD 0,0061%
THD 0,0077%
The level of distortion is comparable to the results of the experimental first channel.

Differences:
- IMD: the small peak at 2kHz is gone
- THD: only peaks at 2kHz, 3kHz and 5kHz, the first harmonic distortion frequencies.
peaks at 4kHz and 6khz are gone.

I suppose the additional small parasitic peaks in the measurements of the first channel were due to the experimental state of the build. There was a direct possibility to change capacitor/no capacitor/corrected THD. The 1M trimpot was not installed properly, the extra wires were too long. Strangely enough, no 5kHz peak was visible on the first THD measurement.

Paul

 

[abbey road d enfer]

- with corrected NFB 0,0079%

Sorry, I've lost track of the subject. What means "corrected NFB"?

 

[Winston OBoogie]

Sorry, I've lost track of the subject. What means "corrected NFB"?

I believe Paul's talking about the network from the 1M pot, through the 100nF cap and 1M R, and returned to the 16K5 cathode resistor. About 1/3rd the way down post 90.


Very nice design Paul. The only thing I don't quite grasp is where there are resistors from the drains of each J-Fet to the cathodes of the upper cascoding valve. Was/is this a means to degenerate the pair or define the open loop gain of the composite stage?
Normally I find that a stiff connection at the drains gives lower distortion. Degeneration and defining of open loop gain being done at the source end of the J-Fets.

For the record, I don't have the benefit of hands on and your insight with this design so I'm just curious as to the process or thinking.
Not at all criticising or questioning how it's been done, just hoping to learn something

Edit: There's actually another thing here that intrigues me: The DC feedback connections to the cascoding tube's grids. I see this as being a "driven cascade" which is a nice trick for keeping the grids and cathodes moving in tandem. An alternative to tying the cascoding voltage to the lower device's source pin.
However, here it's also defining closed loop gain. I don't think I've ever seen this topology before. My brain can't figure out the whole kit & caboodle of the circuit right now... Help Paul !

 

[abbey road d enfer]

I believe Paul's talking about the network from the 1M pot, through the 100nF cap and 1M R, and returned to the 16K5 cathode resistor. About 1/3rd the way down post 90.

OK. Thanks. Very nice idea, congrats, Paul!

 

[aazaa]

Sorry, I've lost track of the subject. What means "corrected NFB"?

Winston OBoogie said:
I believe Paul's talking about the network from the 1M pot, through the 100nF cap and 1M R, and returned to the 16K5 cathode resistor. About 1/3rd the way down post 90.

OK. Thanks. Very nice idea, congrats, Paul!

Thanks Abbey!

The preamp has evolved a lot since the start.
In the first post you can find where essential changes have been posted. (shown in bold)
This allows new people to follow the evolution without having to read the entire thread.

At the moment the information listed is:

+390V version in post #63
heater PSU in posts #64 & 68
hybrid version with UTC LS-30 and LSK389 in post #70
corrected negative feedback - intermodulation distortion in post #90
harmonic distortion in post #91
second channel distortion in post #92

Paul

 

[abbey road d enfer]

Thanks Abbey!

The preamp has evolved a lot since the start.
In the first post you can find where essential changes have been posted. (shown in bold)
This allows new people to follow the evolution without having to read the entire thread.

OK, I need to spend some time to catch up.

 

[aazaa]

Very nice design Paul. The only thing I don't quite grasp is where there are resistors from the drains of each J-Fet to the cathodes of the upper cascoding valve. Was/is this a means to degenerate the pair or define the open loop gain of the composite stage?
Normally I find that a stiff connection at the drains gives lower distortion. Degeneration and defining of open loop gain being done at the source end of the J-Fets.

For the record, I don't have the benefit of hands on and your insight with this design so I'm just curious as to the process or thinking.
Not at all criticising or questioning how it's been done, just hoping to learn something

Edit: There's actually another thing here that intrigues me: The DC feedback connections to the cascoding tube's grids. I see this as being a "driven cascade" which is a nice trick for keeping the grids and cathodes moving in tandem. An alternative to tying the cascoding voltage to the lower device's source pin.
However, here it's also defining closed loop gain. I don't think I've ever seen this topology before. My brain can't figure out the whole kit & caboodle of the circuit right now... Help Paul !

The anwer is partially in your intelligently analising post, Winston.

Our brain is used to read more traditional designs. Usually -when there are 2 or more stages in these designs- each stage is optimised individually. A capacitor or a transformer is connecting the stages. Maybe global NFB. If the individual stages are understood we add up things and that's it.

Not so in the hybrid preamp circuit.
In se it is a very simple circuit but the interaction between all the elements involved is complicated. Understanding the entire functioning in one overview is difficult.

There are 3 couples of 'amplifiers'. The monolithic LSK389 with 2 almost perfectly matched J-fets, the ECC83 and ECC82 with far from perfectly equal triodes (tubes are not selected). Furthermore the ECC82 is notorious for being "a tube with one of the highest intrinsic distortion levels on the planet." (Ian)*

The low distortion IMD and THD figures in the measurements illustrate the circuit's amazing corrective ability. There is interaction -directly or indirectly- between all 6 'amplifiers'.

The 20k pot between the anode resistors is trimmed to have 0.00V between the cathodes of the ECC83. Unusual choice, I know. Both J-fets of the LSK389 are almost perfectly equal. Trimming this way is choosing for the best possible performance of the LSK389.

I opted to connect the current source directly to the source of both J-fets. Gain control is done by the 2k7 and 68k resistors. (Control by the connected opposite sides.). The whole combination - Jfet, ECC83, NFB from opposite ECC82, direct coupling, gain control - is indeed unusual but functioning is excellent.
Perfect interaction between the two J-fets. Direct interaction between the cathodes of the ECC83 triodes, both connected to the 68k resistor.

The whole interaction of all the elements is too complicated to describe. And as if it wasn't complicated enough, there's the added NFB correction. This correction has the ability to reduce drastically the residual distortion caused by inequalities in the tubes.

Alas, I can't describe things better than the way you did it when you asked for help, Winston. I am an amateur with limited basic knowledge. And I'm not English.

There are more competent people here who can explain things better. (You, Abbey, Emrr, Ian, Merlin and many others...)

Paul

* Of course Ian was exaggerating to make his point.
I didn't change the ECC82. It is a very nice sounding tube when corrected properly. Excellent example is the superb Luxman Classic CL-38u preamplifier with also ECC83(3x) and ECC82(5x). THD at 1kHz is a stunning 0.006%.


But Ian is right... [edit: is he ??? see posts #118-121]

I intend to use the Sylvania 6SN7WGT tubes in the second build. I've used this tubes in a preamp probably 25 years ago. Tubes were bought at the time nobody was interested.

Intention is not to have less distortion. Less correction resulting in the same low level of distortion is indeed the better approach. And combining the 6SN7WGT with my 6SL7GT tubes will give me an added vintage feel.

 

[Winston OBoogie]

Hey Paul,
perfect answer my friend, I understand

I wanted to think some more about the subtleties and mechanisms of your circuit last night and reply with something constructive or at least something that sounded halfway intelligent.
But, as I said via PM, I've managed to get myself caught up in dealing with another issue with a tube compressor that's turning into a bit of a longer headache so...

There's a lot to admire in your design. No way are you an "amateur with limited basic knowledge". For what it's worth, I personally think it's the most original circuit of its type that I've seen on here since... I don't know when.

Later...

 

[aazaa]

Thanks Winston!

Improved trimming. Stabilisation of low DC and low distortion.

Trimming DC between both anodes of the ECC82 to the lowest possible level was difficult.
After trimming the fluctuating DC affects the distortion level.

The 16k5 resistor has been replaced by two 33k resistors. The trimmer is connected to ground.
- DC: separate NFB for each triode trough the 33k resistors.
- AC: a 2.2µF capacitor is connecting both resistors/ cathodes. Functioning is almost as before.
- Corrected NFB by two 1M resistors.

Trimming is easy now. More precision, a single turn has less impact.
DC fluctuation is not completely gone but is very limited.
Maybe distortion will be slighly higher (?) but it will be stable.

Time to replace the boards. There is not enough room left for the extra resistors.

Paul