diyAudio

diyAudio (http://www.diyaudio.com/forums/)
-   Analog Line Level (http://www.diyaudio.com/forums/analog-line-level/)
-   -   If You Don't Have a SIT, Fake It... (http://www.diyaudio.com/forums/analog-line-level/206532-if-you-dont-have-sit-fake.html)

wrenchone 12th February 2012 03:59 AM

If You Don't Have a SIT, Fake It...
 
1 Attachment(s)
A week or so ago, I decided to do a simulation of a lineamp circuit using a mosfet with drain-to-gate feedback, a configuration that can look like an extremely linear triode. I wanted to try something triode-ish that could be done with readily available parts rather than busting my a** trying to score some SITs. I was shooting for a gain of ~5 or so, and the values used in the simulation shown below get me there. I was expecting THD in the tenths of a percent or so, and was shocked at the simulation results.

Attached is the circuit I simulated and the simulation results. The jfet buffers the mosfet gate, allowing a high impedance feedback network and also providing some heft to bull past Miller capacitance. Using my trusty favorite the PN4393 or the 2SK117 for input duty nets me similar results, with a little bit of tweaking necessary to center the output mosfet drain voltage.

In retrospect, the setup looks very much like the guts of my "Half Nelson" Class A amp, described in the Pass Amps forum. In real life, I plan to use a depletion mode mosfet current source of some sort to replace the perfect current source at the mosfet drain. The pair of resistors at the input was used to see how the distortion would fare when the circuit is driven with something besides a perfect voltage source. I'll use a 10k volume pot in front of the real deal. I'm pretty sure someone has already done a circuit very similar - it's too simple. I really don't care if I happen to be re-inventing the wheel. Am I gonna build this? Heck, yeah! My only quibble is that the circuit is inverting, and I like a non-inverting line amp.
I'll probably replace the IRF610 of the simulation with a smaller mosfet like the MPF960, a nice little fet in a TO-92L package (sadly discontinued, but I still have some). The TO-92L will handle 1/2W or so of dissipation with no problem.

Salas 12th February 2012 04:06 AM

Brilliant.

wrenchone 13th February 2012 06:51 PM

1 Attachment(s)
Such a simple circuit was easy to build up on perf/vero board.... It needs a little tweaking to exactly center the output voltage, but then, I did a quick and sloppy job of selecting the input jfet anyway. It's good enough to test.

wrenchone 13th February 2012 06:53 PM

1 Attachment(s)
Here's the basic in/out waveforms, showing a gain of -5. Next I should break out my HP8903A, which needs some exercise...

Ken Newton 13th February 2012 09:53 PM

I share your surprise at the THD figure the sim is predicting. Makes me wonder if there is some complementary distortion cancellation taking place between the JFET and MOSFET. Nice little circuit. With sufficient bias current it would probably make for a good headphone amp as well.

buzzforb 13th February 2012 10:36 PM

The 2sk2013 might improve things even more as I believe it could possibly be a better device than the 610.

wrenchone 13th February 2012 11:52 PM

I'm using an MPF960 in the prototype, as it has about 1/2 the input capacitance of the IRF610, and it's in a TO-92L package that will dissipate the 300mW or so required by this application. The TO-220 device you see on the breadboard is a Supertex DN2540N5 depletion mode fet, used as the current source load for the output mosfet. If I get within a factor of ten of the THD shown in the simulation, I'll be one happy camper.

Nelson Pass 14th February 2012 12:32 AM

1 Attachment(s)
The SIT simulation thing seems to have originated with Schade, who was
simulating triodes with pentodes using this sort of feedback. With your
input Jfet, the input impedance is high enough to be usable.

Here is a graphic that one of our members posted (can't remember who,
sorry).

:cool:

buzzforb 14th February 2012 12:52 AM

Quote:

Originally Posted by Nelson Pass (Post 2905498)
With your
input Jfet, the input impedance is high enough to be usable.
sorry).

:cool:

Is this one of the keys to this circuit?

wrenchone 14th February 2012 01:19 AM

It was certainly what I had in mind when I used a similar circuit for the "Half-Nelson" amp, (something I ginned up in 2007 before I found out about all this Schade stuff) and here as well. The jfet allows a high impedance feedback network and a respectable frequency response, due to the low jfet capacitance and the source follower drive to the mosfet gate. I'm going to use my results here to do another spin on the Half Nelson, possibly using one of the Ixys depletion mode fets on top of the output mosfet. I'll post circuit and results for that back in the Half Nelson thread when I get around to it.

Yep, I saw the simulation (or something similar) shown in post 8 in one of the tube threads. The "Schade Fet" has cropped up several times, once as a ringer in a "name that tube curve" threrad.


All times are GMT. The time now is 10:45 AM.


vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2014 DragonByte Technologies Ltd.
Copyright 1999-2014 diyAudio


Content Relevant URLs by vBSEO 3.3.2