Complimentary-symmetry SUSY?

[vossie]

Below is a circuit comprising a push-pull arrangement that is close to Nelson's SUSY patent - see Fig 2 in the document. Although I don't think it qualifies as X (see below for reason) I would like Nelson's input on that.

First off, let me say that the circuit is by no means optimised. Choice of transistors, biasing values, shield transistor voltage conditions, etc are not optimum.

At first sight it looks fairly complicated when compared to the current circuits floating around, but once you get the basics it is fairly simple.

Added are the cascode shields Q1, Q2, Q7 and Q8 providing a nice stable environment for Q3, Q4, Q5 and Q6 to operate in. This allows you to run these input transistors at a low voltage, biased in their ideal operating bracket.

Q4 & Q5 is biased by the voltage sources V5 & V7 which produce a constant voltage and therefore a constant current over R5 and R6. R2 and R3 allows for cross-talk between the halves. When driven balanced the whole circuit balances around the point where R2 and R3 is connected. (A different cascode arrangement could allow a self-balancing act in the differential pairs and could possible make active component selection of lesser importance - I've tried it and it seems to work but the gain is unity)

Tansistors Q11, Q12, Q15, Q16 are there to give full voltage swing but does not add current. Reason for this config is to stay as close to Nelson's patent circuit as possible.

The four current sources I7-I10 are set at a rather high current. The reason for the high value is because the current is needed to drive the 1k resistive load (just a value I pulled out of the air) at each output to ground (well part of it anyway as it is combination of R2, R3 and the biasing current of the four input transistors). This brings me to the FIRST problem I have not figured out yet. The GAIN is dependant on the value of the load resistance. (Circuit here is for 1k per side). This is not ideal at all as the small currents makes it quite sensitive the the load. Seems that these resistors R1 & R10 provides some form of feedback into the system which lowers the gain - I've gone blind to this problem {anyone?} I think the solution is to change configuration of the voltage gain transistors (Q11, Q12, Q15, Q16) to give current gain as well.

The second problem is the circuit does not respond well in both sides to a single-ended input (although the side being driven gives a perfect response). This is the reason I don't believe it to be SUSY compatible - Nelson? Does the particular circuit in your patent (Fig 2) work when driven unbalanced - I'm sure it does so from here my reasoning that the circuit represented here is not realy X? I have not tried to build that circuit so I was wondering...

Currently my reasoning for the lack of response in the other side when driven single-ended is the two cascode voltage sources connecting at the revolving point between R2 and R3. Just got an idea - will config that differently and see what it does...

The circuit shown is just up to the VAS stage - to drive a speaker load an output stage needs to be added (BTW, the complimentary symmetry JLH type works very nice in my ORCAD simulation.) Performance of the circuit attached is quite good - as is it gives full power (into the 1k load) up to well over 200kHz with very little distortion. (bit irrelevant at this stage...)

Below is the spice simulation file if you want to give it a go. I would love get some help from your guys to get this one a bit more "matured".

Cheers,
Vossie

[vossie]

Response of 20kHz square input (0.5V p/p)

[vossie]

Response of 200kHz square input (0.5V p/p)

[vossie]

Response of 200kHz sine input (0.5V p/p)

[vossie]

The simulation file attached...

[vossie]

Here is the circuit with a MOSFET output stage... Much less dependant on load for gain. Part of the feedback nw now returns from the VAS stage to ground through 4 1.6k resistors.

[Henrik]

Hi Vossie

I don´t think this is X, it looks like a bridged amp to me.
In my opinion you need a diffrential inputstage, like in the Aleph-X.
In order to get X you need some communication between the two halves of the amp, in your setup they are isolated.

The diffrential inputstage also makes it possible to drive Xamps with unbalanced input and still get the X-effect at the output.

Nelson from the X-SOZ thread:
"The essence is still using a symmetric feedback arrangement
around a differential pair which isolates error from signal
and uses it to make distortion on two halves identically
in phase and thus cancelled."

You also need some resistors in series with the input in order to get the wanted feedback.

[Henrik]

Hi Vossie

If you release the R2 and R3 from ground and just connect them, and put some 1K inputresistors in, you then get some diffrential behaveiour. If it will give you X I don´t know, then it is certainly intresting.

You can try the changes, i have attached my suggestion in a simetrix file and pdf.

[vossie]

Henrik,

I share your doubts - that's why I said that the imbalance in the opposite stages makes me wonder if it qualifies.

However, I think you missed R8 & R9 serving as the feedback to the inputs... and R2 and R3 which coupling the stages. The grounding at this point does not isolate the two stages from each other.

Any further comments?

[vossie]

Nice, now we're making progress...

Thanks. I actually had it like that on a older circuit but it got changed getting the biasing right. Now it's very much like Fig 2 in the patent...