Jfet-Mosfet Gain Cell

[wrenchone]

Attached is a schematic and response results for a gain cell using a cascoded N-channel jfet plus a p-channel small signal mosfet. There is feedback, but the feedback path is short and direct. The cell as shown is set up for a gain of 40x. Distortion results are impressive. I've laid out an RIAA preamp using a pair of these cells with a passive RIAA network in between, and will add it to the growing pile of things to be built/evaluated. By using the mosfet, you can use direct coupling between stages and still have some leeway to bias both stages properly.

[wrenchone]

One of the things I want to do going forward is some juggling so that the input jfet has a little more current. I'm also looking at the possibility of using a Darlington as the second stage rather than a mosfet.

[wrenchone]

Here's a ~5X lineamp based on a similar cell structure, but with a "helper" current source for the input jfet, my old friend the PN4391. Food for thought...
You could also do a low level gain cell with the BF862 in front end duty with a helper source. I haven't characterized my little roll of BF862s yet (lazy, I guess), so I don't know how much help they'll actually need.

[wrenchone]

Here is a low level sim using the BF861, which has a model in PSpice. Gain is 30X (needs tweaking), but distortion is nice, so similarly fruitful result could probably be expected for cousin BF862. If someone wants to take this schematic and tweak it with an available BF862 model, I'm sure the results would be interesting to more than myself.

[wrenchone]

As promised quite a while back, here is the basic gain cell configured so it will run with a Darlington. To get this to work, I used the equivalent of 3 X GaP green LEDs (2V each, ~ 13 ohms incremental impedance if you use the HP variety) in the emitter of the Darlington (6V+ 39 ohms equiv. resistance). I also put a current source from the Darlington collector to ground so I could run some extra bias current through the Darlington + LEDs and still get reasonable output voltage centering. It'll be interesting one day to look at the actual THD and harmonic distribution vs. the sim.

Another beast for the menagerie.

[Calvin]

Hi,

as always I like Your little circuits alot ;-)
BTW. The BF862 is a NXP part and NXP has models of their parts downloadable on the Website.

jauu
Calvin

[wrenchone]

This is true, but getting them to work in the current version of PSpice is a royal pain in just about every nether part you can imagine. Maybe I should give LTSpice a try or unearth and upgrade my copy of Beige Bag Spice ( or just buckle down and figure out how to work with PSpice). There are models from IXYS and Supertex (their depletion-mode stuff) I'd like to be working with as well...

[grufti]

Very interesting. Thank you. A bit like a SIT, like you said in the RIAA thread.

[hazard500]

Originally Posted by wrenchone

Attached is a schematic and response results for a gain cell using a cascoded N-channel jfet plus a p-channel small signal mosfet. There is feedback, but the feedback path is short and direct. The cell as shown is set up for a gain of 40x. Distortion results are impressive. I've laid out an RIAA preamp using a pair of these cells with a passive RIAA network in between, and will add it to the growing pile of things to be built/evaluated. By using the mosfet, you can use direct coupling between stages and still have some leeway to bias both stages properly.

Actually this looks like a "double" cacode to me - you have drawn a folded cascode where the the bottom (nFET) device is itself a cascode of 2 nFETs.
A simple folded cascode (1 nFET, one pFET) will give the normal cascode benefits of reduced Miller capacitance and also provide the DC level shifting, is there any reason to add complexity? ie it appears to me that the additional nFET is not required, can you please explain rationale for adding this in?

[wrenchone]

These are all complementary feedback pairs of some sort, with cascode on the input stage (not totally necessary in some cases) to reduce the effect of the miller capacitance on the first stage - certainly a concern if used as the first stage of an RIAA preamp - maybe not for other applications.I have looked at simulations where I've use a folded cascode (not in this thread, I don't post everything I do by a long shot). This is possibly an option for the first stage of an RIAA preamp where the signal amplitude is small, but not for a second stage or a low gain line amp where the signal/current swing is higher. If you differ, please posit a circuit as an example, and we can talk about it.

The basic premise of what has been presented here is a complementary pair where the excess gain is fed back to reduce overall distortion - all the other little bows and ribbons are secondary to that purpose. I have another thread where the two sections of the circuit are AC coupled to allow independent bias of input and output stages - the important concept here is DC coupling between input and output stage along with decent biasing/centering and good overall THD.