New Line Stage

[fitzfish]

I wanted to share a new line stage idea that developed out of some work I was doing with the Zen IV converter. I posted it here instead of in the analog line forum as it is more in line with this one. I looked for a Balanced to SE conversion stage that was as simple, elegant, and performed as closely to the Zen IV as possible. No ICs for this one obviously, and every discrete opamp I modeled that performed well became too complex to satisfy the “simple and elegant” requirements.

Here is what I came up with - It is similar to an Aleph P, only built with JFETs. Unlike simple followers, it can support gains reasonably up to 10 or possibly higher, can support voltages of +/-10 to +/-20V or higher, and can operate of a single supply of between 10 and 40V or so.

If you have a problem with capacitors in the signal path, you probably should look away… Just kidding, no more than usual for something this simple!

J2 and J4 are current sinks for J1 and J3. The gain is defined as the parallel combination of R4 and R5 divided by R3. R6, R7, and R8 serve the same purpose for the other side. Gain Resistor R9 is feedback between the sides and is sized to set the gain accurately. The rest of the circuit is pretty much self explanatory. The HF rolloff caps C1 and C2 are optional based on your taste. The input and output are necessary to prevent offset concerns with other equipment.

I tried changing resistors R1 and R2 to into current sources, but the circuit suffered for it. The gain of these devices is jut to low from what I can figure. Likewise, reducing the current in J2 and J4 with source resistors also degraded the performance.

The circuit as drawn starts to clip at about 5.5Vrms, or 15V p-p. Should be enough for most needs. Distortion from the models looks to be below 0.005% at 1Vrms output. It rises as you add gain, but still looks to be below .01% at a gain of 10...

To increase the gain, first decrease R3 and R6 and R9. R3 and R6 are set at 10K for unity gain to keep the input current low (and the input caps small!) G=5 is set by R3, R6 = 1K; R4, R5, R7, R8 = 20K, and R9 as 320 ohm. To set the G=10 R3, R6 = 1K; R4, R5, R7, R8 = 40K, and R9 as 160 ohm.

Back to where I started, if you connect your SE amp to the positive output and ground, all appearance are that it will work as a nice balanced to SE converter. if you need to invert 180 degrees for some reason. It is there for you in the negative output.

I welcome any comments and feedback. The circuit is fairly well optimized, although I may have missed something. I may not get back quickly, so please be understanding if I do not reply promptly. I also hope that someone with both SE and balanced input amplifiers will test this somewhere along the way to see if it compares well for that purpose.

Dave

[Zen Mod]

we made something like this previously ;

dunno how's named - maybe Jfet BOSOZ or something like that

full SUSY , DC coupled inputs etc

in any case - you made nice work

[fitzfish]

Originally Posted by Zen Mod

we made something like this previously ;

dunno how's named - maybe Jfet BOSOZ or something like that

full SUSY , DC coupled inputs etc

in any case - you made nice work

Hi Zen Mod,

Thanks for the compliment!

Also thanks for the original project name. A search led me right to it... Some of those discouraging sim results look very familiar. I never could get the CCS replacements for the resistors to work well enough either... It can work but the distortion goes up considerably. Never got to a cascode, need to try it.

I will also go back and look at the input caps to see if they are really necessary. I started with fairly low gain resistors at the input. With the final values, input caps might not be necessary at all. Need to check at different gains and input levels as well.

Maybe there will be some interest in a JFET version again.

Dave

[nar]

Yes, definitely ! Keep up the good work,

Brds,

nAr

[fitzfish]

Thanks nAr.

I also wanted to point out for everyone that the gain for SE output is 1/2 that of Balanced output. The gain figures in the original post are for balanced out. The clipping number for the output is per side, so for balanced it would be twice as much. I just want to make sure there is no confusion...

Enjoy!
Dave

[qusp]

would be nice for wolfston voltage out or headphone amp

[EUVL]

Some suggested reading for you :

Distortion in JFET input stage circuits


Patrick

[fitzfish]

A few more details for those interested -

On the input capacitor question, removing the caps causes a modest rise in distortion. Offset current at the input I believe. I suggest you could take them out or leave them there, depending on what you think when you listen to it.

As drawn the circuit bandwidth is limited by the input capacitor. At unity gain, the -3db point is ~10Hz. when increasing the gain to 5, it is at ~20Hz. Increasing the input cap to 2uF keeps the -3db point the same, ~10Hz. The output caps can be reduced to 2uF as well with a 10K load on the output with about the same -3db point. If you want to drive into 600 ohms differential, leave them large...

Patirck,

Very nice information there. This appears in line with the changes you made to the F5 and F5X diff pair to more closely match the parts. Reading through it also is eye opening when it comes to the performance of a discrete opamp-like implementation with a conventional jfet diff pair and current source. Without more gain (more stages), current sources, degeneration, cascodes, etc, I would not have gotten where I wanted to be... It answered just about all of the questions I had! Thank you.

Interesting reading in the Blowtorch threads as well. I did not make it far enough to see if anyone made a PCB for DIY use. I like the JC80 circuit, it reminds me of a certain xamp in the works...

Dave

[EUVL]

I was trying to point out to you the difference between single ended and complementary diff pair.


Patrick

[fitzfish]

Hi Patrick,

Thank you. I will be doing some complimentary work in the future I believe! It was enlightening for SE pair performance as well...

Dave