cap values and boost freq in an srpp: how?

[duck_arse]

I currently have this on bb, and some questions follow from.



the input impedance of Q1 is set by R6. the gate is fed signal by C4, so it's working against that 1M to set the boosted/driving range, is this correct? fiddling the C4 value between 47nF and 680nF against the 1M, the boosted freq should be flat from 4Hz. but as far as my observations of the low freq response, it seems more to correlate to the value of C4 against R8.

is any of this correct so far? is my C6 connection in/correct, the whole cause/inconsequential? how do all those wildly variant values for the 'C4' in other's circuits get selected? should C8 up to 2k for j201?

additionally, has anyone else tried the omega front-end on this circuit?

[amptramp]

First, an SRPP stage takes its output from the source of the upper transistor, so move C6 to the other side of R8.  As drawn, the upper transistor acts as a current source - high gain but not true push-pull SRPP operation.  R8 should be sized to be equal to 1/g_m of the upper transistor.  That way you get equal gain for the upper and lower stages.  If you look at where the current goes, the SRPP stage operates as a power amplifier because the difference between the upper and lower transistor current has to go into the load - it has nowhere else to go.  The RV2 value of 250K looks a bit too high for this circuit.  When it is too high, the amplifier saturates to either the top or bottom rail, meaning a lightly-loaded SRPP stage might make a nice clipper with R8 varying the symmetry of the clip.

[Transmogrifox]

C4 small-signal frequency cut-off is defined also by approximately 1/gm of the upper FET, so your observation that it seems to be dominated by R8 in this configuration is correct.

That said you will notice the high-pass effect with C4 in the k-ohm range as opposed to the 1Meg that would otherwise be intuitively supposed.

Of course, for large signal swings this becomes nonlinear and adds to the overall effect if you locate that cutoff in a place where a change can be noticed -- some kind of a mild compression effect.

[PRR]

What is R7 for??

C4 is a bootstrap. It has to be much larger than you would think from R6's value. Its action may be very load-critical. Find your bass-cut somewhere else.

Everything interacts here. You may not have enough free parameters to get all the tone-shaping you might want. My thoughts would run to a buffer at the output.

RV1 can make input impedance as low as 22K. Are you sure this is what you want?

[duck_arse]

thanks for replies.

amptramp - dunno where I got that wrong connection, but it did work. 1/gm is good, more maths to play with. RV2 was 500k, didn't come to hand. just grabbed parts produced a fairly even clip, plenty gain, maybe because of wrong connections.


transmog - do you mean the 1M of R6?

PRR - the great man hisself, RG shows that resistor in his "foolin' with fets" dissection, so I left it until I had something working. shorting it produced some slight effect, but I need another go at this with changes to see what changes. RV1 is straight from ROG's omega. I was looking for some way of reducing what the stock circuit was producing on the bb. it has some tones in it.

[merlinb]

Why bother with all the biasing components around Q1 anyway? Q1 could be self biased with R8 alone, and adjusted to the desired drain current.

[amptramp]

duck arse

The connection you have to the bottom of R8 to the output will function and in this case will appear to function as a FET amplifier with a current source load.  Because the current in this stage has to go somewhere, it is a power amp stage that requires a load just like the SRPP stage does.  The true SRPP stage is a series push-pull stage.

merlinb

You could use self-bias but adding an additional gate bias network allows you to design the stage for transistors with varying Idss and Vgs ranges in transistors.  The biasing network allows the designer to separate these values from the operating parameters of the circuit.

[duck_arse]

the amended:



personally, with my cloth ears and 2 chords, I couldn't hear the difference (Jon's 'clipper ship' is wired "the other way" as well). I've made the changes all the same, as minor as they are.

the output buffer: if the SRPP wants to see a decent load, the buffer won't provide it. what aspect of the SRPP is being isolated from what by buffering? I did try a direct coupled jfet buffer, but thought the shift in clipping would suit a distorter rather than a booster.

and g_m: I see g_fs and |y_fs| both referred to as (forward) transconductance on the datasheets, is this the same/similar to g_m? and once R8 is equiv to 1/g_{m - or - fs}, does the freq response of the upper effectively become 1/2*pi*RC or close to?

[PRR]

|y_fs| implies that transconductance varies with frequency. It sure does, at frequencies 1,000X higher than the audio band. If there is a plot that extends down to lower freqs where |y_fs| flattens-out, it may be a clue. There's usually a low-freq number elsewhere in the sheet.

[duck_arse]

for the j201, both Vishay and Fairchild graph for g_m, but fairchild give a min figure for y_fs (without pipes), then don't mention it again.


http://pdf.datasheetcatalog.com/datasheet/vishay/70233.pdf
http://pdf.datasheetcatalog.com/datasheet/fairchild/J201.pdf