Simplistic NJFET RIAA

diyAudio Chief Moderator
Joined 2002
Paid Member
For the time being I am dropping R17 just in case since at least in another simulator and NJFET modeling it showed some peaking. I am not after absolute 0.1dB actual RIAA conformity, because neither the vinyl records nor the best monitor speakers can follow that. Seen any other aberrations in your simulation? Or you just simulated the RIAA only out of my circuit, under low Zo input and high Zin load? Because the response of my circuit is rather limited due to Miller and no GNFB or local NFB. I still wonder how high it peaked.

Thank you for double checking. Appreciated.
:)
 
Don't you Helene's ever go to bed?

Sims can vary because of our different models, cable capacitance and sloppy wiring etc., etc. but it looks like your RIAA is +/- 0.15dB from 20 Hz to 20kHz -- pretty darn good. R17 plays a pretty important role in the ultrasonic range --

I put 2 bode plotters on -- one with the standard RIAA, the other with the enhanced RIAA -- note the difference in vertical scales.


An externally hosted image should be here but it was not working when we last tested it.


Parenthetically, the difference between Standard and Enhanced RIAA Curves is about 6.9 dB at 100kHz -- but we aren't going to hear that high anyway -- we just want to eliminate the possibility of problems down the food chain:


Jack
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Final Theoretical Schematic

So to sum up the circuit idea and cross checks up to now, the final schematic to be made for real life test is in the attachment. I have checked, and It can work fine with 36V B+ too, using 4X9V or 3X12V batteries in series, instead of a regulated PSU.
 

Attachments

  • screenshot.gif
    screenshot.gif
    17 KB · Views: 3,195
I had a look at the Sears phono preamp schematic - comments.

I tried a JFET SRPP passive equalization phono stage a while back (search threads here for "JFET SRPP"). Distortion performance (simulated) was impressive for only 4 devices/channel. I used the PN4303, which has fairly low gm (needed to keep the gain from going through the roof without totally strangling the FET) and is a moderately low noise device (good for MM, not MC). I ran into problems with variable gain on the first stage due to the variable load of the RIAA equalization network vs. frequency and finite/relatively high output impedance of the first stage SRPP, and canned the project in favor of another approach. It didn't occur to me to use a higher impedance RIAA network - duh! This will breathe new life into the first stage. The second stage had issues due to relatively high output impedance. I would not use Sear's solution of a common source second stage due to high distortion (0.3-0.5%, though it's overwhelmingly 2nd harmonic). Using an SRPP second stage, loading it with a fixed resistor, followed by a current source loaded source follower dramatically drops the simulated distortion (0.02 to 0.06%, depending on how you bias the stage and how much load you use).

These sorts of simulation results may make me break out the old SRPP RIAA board and try it again after a little surgery.
 
I'm pretty familiar with the Pearl - I wanted to try another design that didn't require the 2SK170 or 2SK389. The SRPP topology wrings surprisingly good performance out of a relatively proletarian JFET like the PN4303. With the perspective from a couple of years of "fallow time" and inspiration from the Sears preamp, I might be able to cajole the SRPP circuit into a very decent showing.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Note on JFETS used

With 10mA Idss we get 70dB gain in my circuit. With 12mA Idss G=73dB. Even with median 8mA Idss BLs we hit 68dB. Just keep your strongest matched pair for the first stage. Matching must be used for same stages between channels so to keep the gain together for L&R. Also 500R trimmers can be used instead of 220R so to trim best for drain voltage and consistency between channels for that.
 
wrenchone said:
I ran into problems with variable gain on the first stage due to the variable load of the RIAA equalization network vs. frequency and finite/relatively high output impedance of the first stage SRPP, and canned the project in favor of another approach. It didn't occur to me to use a higher impedance RIAA network - duh!

For the Sears -- with 255k -- you are going to have a bit (4.6uV) of broadband noise, no?
 
diyAudio Chief Moderator
Joined 2002
Paid Member
wrenchone said:
Why do you want all that gain? Are you trying to run with a moving coil cartridge? One of the reasons I went with the PN4303 was to have a resonalbe stage gain. The gm is olly about 1.5-2k mmoos, about 1/10 the value for a 2SK170.

I want it to be able to cover anything with minor changes, or by using switches, and without step up. A flexible situation. Use commonly measuring 7mA Idss 2SK170BL for instance, take out the second stage 1000uF bypass, and you get 55dB. Perfect for Denon DL160 that I can find for 100 Euro. Drop the first 1000uF cap as well, you get 44dB. Great for a Grado, or any 3.5-5mV.

*Better use 16nF second RIAA cap if dropping bypasses.

At circa 70dB I can compare it at my friend's set up with the low Ortofon and the Steve Bench tube stage we made, without the step up cost.

Plus if a phono stage does not hit around 1V output with a given cartridge and chosen gain combination is generally short for macro dynamics to my subjective experience.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
I have one DL-160 that I plan to use. Some tests on load resistor before input stage and it will sing. My circuit is well doing for bass it seems. But in general, if the RIAA is right in the lows, not loaded down by the second stage, and there is a buffer at the output, I think that most cartridges thought as ''thin'' sounding, would fair better.
 
to Jackinnj -

The impedance in the Sears circuit is less than a lot of common tube-based RIAA networks that are used with MM cartridges. Also remember that there are two stages of low-pass filtering after the initial resistor. This may not be an issue anyway, as I can use the same resistive loaded JFET SRPP/source follwer combination for the input section as well as the output. simulated performance for low level signals is quite sweet (0.002% THD). and the follower will drive a lower impedance equalization network with no sweat. I can even jigger up an old prototype I have laying around that has most of the essentials in place. It's two extra parts per stage, but that's still a lot less complex and fiddly than what I'm using now. Everything changes for low level MC cartridges, but I have no desire to use them until I get a much more sophisticated TT than what I'm now using.