JFET SRPP RIAA Preamp

Then it's not an SRPP. I will continue with this project as an an SRPP so that I can suss out the implications. Juma has already done the gyrator thing and tacked it on to another one of my threads. We'll see about output impedance when I measure it.

Suit yourself. I wasn't attempting to hijack your thread, only to a suggest a solution to the problem of metastable quiescent drain voltage that you had raised.
 
ken:
i think what you're describing is what was done for the phono preamp used in the Hafler 915 jfet preamp, page 26.
yes?
they add complimentary symmetry and use the unobtanium jfets, of course.
:)
mlloyd1

Perhaps, the most reliable means to address this problem is to apply fixed-biasing to the upper JFET, turning it in to a gyrator. You could create a bias reference voltage by dividing the supply in half with a string of two very high value resistors. Tap the midpoint of this string for biasing the upper JFET's gate. Then, A.C. couple the upper JFET gate to the bottom of it's source resistor, at the lower JFET's drain). You can use a small value capacitor because of the high impedance terms involved.
...
QUOTE]
 
ken:
i think what you're describing is what was done for the phono preamp used in the Hafler 915 jfet preamp, page 26.
yes?
they add complimentary symmetry and use the unobtanium jfets, of course.
:)
mlloyd1

Hi, mlloyd1:

Since you ask (no hijacking is intended, Wrenchone :D), yes, the front half of that Hafler circuit is it. A fixed bias gyrator allows us to have our cake and eat it too. We obtain both a high A.C. impedance - giving us maximum gain, and a low D.C. impedance - giving us a stable and predictable D.C. bias voltage across the FETs. I know an good idea worth adopting when I see one. ;)

I would not worry too much about having complementary symmetry, along with sourcing the scarce P-channel JFETs that would require. I would build this circuit single ended utilizing only N-channel parts. I would cascode those FETs, and also implement the indirect feedback technique on the upper FET, per the below linked paper from A.N.T. Audio. My best sim of just such a circuit predicts a gain of around 46dB. It will produce a 1kHz 2VRMS output signal with a sim'd 2H at around -75 to -80dB, and 3H at nearly -120dB. All higher harmonic are beneath -140dB. All without resort to loop feedback, only low value (10 ohm) source degeneration resistors and the indirect feedback technique cited above. Okay, unplanned hijacking over. :rolleyes:

http://www.ant-audio.co.uk/Theory/Application_of_Indirect_Negative_Feedback.pdf
 
Last edited:
To put it bluntly, I'm not in love with gyrators butting up against triode equivalents. A current source works better against a "triode equivalent" than a voltage source (at least in my book). Anyway, I want to investigate what can be derived from a simple SRPP before I make a leap (if I ever do) to some other alternative. It's my time (I don't have a lot of it, for various reasons, none of which I have to justify to anyone here), my resources (ditto).

At any rate, I've already figured out how to compensate an SRPP (the position for the capacitor was not where I initially thought it would be, and I got a nice, overdamped response with 15 pF of compensation)). I suspect quite strongly that most hot-doggers here have let fly with SRPP designs here, SRPP there, without any sort of time domain/frequency domain investigation of stability. To be brutally honest, I never did a time domain test of a simple SRPP stage until the results I've posted here. Others might benefit from such a simple investigation...

Maybe a simple unquestioned SRPP topology works for vacuum-state triodes that have all their intrinsic parameters built-in ( maybe not, who knows?), but it appears to be a different situation for triodes synthesized from enhancement mode devices (with feedback) that have their own agenda regarding parasitics. At any rate, unless someone tells me differently, it appears that I might be forging some limited new ground. You are totally free to build actual circuits and post results. People who will actually do this (a mere schematic doesn't count) appear to be rather thin on the ground.
 
To put it bluntly, I'm not in love with gyrators butting up against triode equivalents. A current source works better against a "triode equivalent" than a voltage source (at least in my book). Anyway, I want to investigate what can be derived from a simple SRPP before I make a leap (if I ever do) to some other alternative. It's my time (I don't have a lot of it, for various reasons, none of which I have to justify to anyone here), my resources (ditto).

To also put it bluntly, I'm a bit mystified by your defensiveness.:confused: Who here has suggested that you must like or dislike one topology or another? Do you really feel it's necessary to declare to us that you do not have to justify the use of your time or resources? Really?? Sure, do or don't do whatever you wish (do I really need to say that?). Yet, can no one else submit directly pertinent thoughts and suggestions intended only to be helpful? Should you personally not find such thoughts and suggestions useful or interesting, so be it. However, there are others reading this thread who might. You are reacting almost as though this thread were your private blog. We are here as a knowledge community, each of us teaching and learning as we interact, aren't we?
 
Take a time out Gentlemen. Wrenchone stated twice that he was continuing research on the SRPP circuit as a purely acedemic exercise. Even if it is for personal edification there are people following this thread with interest. This is not a thread started as a group design project unless I missed something somewhere, and I have been flollowing this thread since it's inception many years ago. The written word, even at it's best, does not convey nuances that personal interaction does. Trying to ferret out what emotions a person wants to evoke is an exercise in futility.
 
Take a time out Gentlemen. Wrenchone stated twice that he was continuing research on the SRPP circuit as a purely acedemic exercise. Even if it is for personal edification there are people following this thread with interest. This is not a thread started as a group design project unless I missed something somewhere, and I have been flollowing this thread since it's inception many years ago. The written word, even at it's best, does not convey nuances that personal interaction does. Trying to ferret out what emotions a person wants to evoke is an exercise in futility.

Forgive me, but I don't believe a referee was required to either interpret, mediate, or call a timeout for us. Nothing ugly was taking place, nor about to take place. We were simply negotiating through an understanding of personal perspective, upon which we could more effectively communicate. From my perspective, it was as simple as that. :)
 
I've sort of fallen out of love with the concept of jfets used in an SRPP context, even though I could sort of force it to work.
As an analogy, it's like trying to make a pair of pentodes work as an SRPP, whereas all the SRPPs you see around are implemented with triodes. Think of dueling current sources..This explains why anything that made the jfet characteristic even more like a pentode (such as cascoding) made the SRPP hard to balance in practice, though it would look peachy in simulation, where the bits are all exactly alike.

At any rate, here's an SRPP design using a mosfet with feedback to emulate triode characteristics . The actual gain cell uses a P/N channel pair, so that I can get a high impedance input that will be far less sensitive to the characteristics of the driving source. This thing might be useful as a line amp with gain, especially if a follower is used at the output. As shown, the gain is ~5X. Distortion scales fairly linearly with input amplitude, and is overwhelmingly 2nd and 3rd order.

One thing I plan to do (though it'd not a really urgent priority) is to run a series of plots to determine the sweet spot for distortion vs. output load. I suspect that the 10k load I used is not optimum, and SRPPs are touchy about the output load. Ahem, another less endearing aspect of SRPPs.
 

Attachments

  • SIT_SRPP.gif
    SIT_SRPP.gif
    25.3 KB · Views: 728
Hello , Wrenchone ! - I saw this topic late . Thank you for your work ! This is a very interesting topic . I read it with great interest . If you don't mind, I'll try to make a phono preamp for myself according to your scheme from message 222 . Among my phonos, I now have one JFET SRPP RIAA-made according to the scheme from the Elektor publication . And it sounds good . DIY RIAA preamp with FETs | Audiokarma Home Audio Stereo Discussion Forums . And I am very interested in your experiments with field-effect transistors . Good luck to you !
 
Last edited:
For perverse reasons, I've re-opened this investigation to see what I could accomplish with an SRPP design using a relatively grubby, readily available jfet like the J113. Audio Research and Creek have both used this fet in their RIAA preamp designs (not 100% certain for the AR design, but strongly suspect - they house number their parts...).

I won't post the results until I get the rest of the circuit sussed out, but using the J113 at the input as an SRPP gives me a gain of about 100X in the first stage. This would be a disaster for phono cartridges that are sensitive to input capacitive loading (like most of the Audio Technica MM designs), but adding a cascode to the bottom fet fixes that. It also disturbs the stage balance, so I had to mess with using different resistor values for top and bottom loading. I also jiggered things around to find the sweet spot for resistive loading, and got simulated THD of ~0.005% as a result, most of it 2nd harmonic. Whether I'll actually be able to use the cascode and balance the first stage output voltage in an real circuit remains to be seen. If this design actually plays out, I envision using a ~100X first stage, and either a 12X second stage or a 6X line driver stage plus a 2X step-up transformer. The transformer would be useful for inverting the output to get overall non-inverting gain, as all the low-distortion line output designs I'm currently using are non-inverting, and the SRPP input stage is inverting.
 
Last edited:
I've developed SRPP and feedback-based gain blocks using the J113, and am about 3/4 of the way through a layout throwing everything together. It will be interesting to see if I can actually balance an SRPP stage in the real world (as opposed to simulated) with the bottom fet cascoded. I have plenty of J113s, so finding appropriate matched pair won't be too big a problem. If it works, I may try the cascoding and balancing dodge on an all-SRPP design using the PN4303.
 
I tried jFETs in SRPP but soon abandoned due to difficulty in balancing them. Anyway, the attached spreadsheet calculator might be of help. Just input the amplification factor u (mu, μ) and the transconductance gm, set the resistor in the source Rk, the load resistor RL and go. I hope it is useful.
 

Attachments

  • SRPP.zip
    16.3 KB · Views: 85
Since I haven't sent out for a PCB yet, I'm thinking to alter the layout so it can be stuffed as an SRPP for both gain cells of the RIAA preamp section, with the object of also being able to populate the 2nd stage as a non-inverting feedback amp. I think there's enough room for the extra components that would be required.