Jfet preamplifier

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
What do you think about this preamp? (Not designed by me)
 

Attachments

  • j-fet.gif
    j-fet.gif
    9.8 KB · Views: 3,000
diyAudio Retiree
Joined 2002
jfet preamp

A very competent design. It does have a large DC offset that needs to be AC coupled. Substitute some Toshipa jfets and bipolar transistors and swap the tantalum caps for some low ESR elkos and you should have a very good preamp circuit. You probably want more 1N4148s to bias the output stage for a few ma. The fets driving the output BJTs are source followers. Put the 470 ohm 100pF LP filter in front of the 50K pot. A little more complicated than the following (by the addition of the cascode gain stage andthe BJT ourput stage outside the negative feedback loop):

http://www.forsselltech.com/JFET Opamp.PDF

(which is shown without the feedback resistors)
 
Hi, Freddie!

This 50 k pot is a little bit high I think. Is it possible to use a 10k pot? If yes, do so. Why? The input filter won't influence so much when pot is at - 6 dB.

The -3 dB will otherwise vary from 34-720 kHz.

I think it's a matter of taste to have to pot at first or not. I would have chosen to have the LP-filter as close as possible to the real input. A tips though: Connect a pulldown resistor (100k-1M) from wiper to ground. Why? You can avoid noise if the pot wiper should loose contact with the resistance track, lane (or whatever you call it), very nasty if the amp has bipolar input stage!

Skip all 100 ohms res, add them if you must.

The 10 pF compensation cap is probably too small.

The output stage runs in class B. Put at least one more diode in or replace the whole thing with a Vbe multiplier, easier to trim the current.

I would also decouple the resiistor inbetween the bases of the output transistors.

As Fred says, the input offset problem is a real problem, can be fixed though.

The feedback is taken from the wrong point, yes? Use the output as feedback point but you can also have a little local feedback from the VAS stage.

Decouple the ref voltages for the cascodes towards the voltage rails, not to ground.

Lower the values of the feedback resistors. Use 1 k + 3.3k instead. It's allways an advantage to use low impedance feedback networks. Stray caps inluences less.

As current mirror, I recommend a Wilson mirror or a plain current source LP-filtered and possible a cascode on top of that. High impedance of the current source is important when use have a non-inverting amp.
 
diyAudio Retiree
Joined 2002
Per review and other unqualified opinions

DO NOT REMOVE THE 100 OHM RESISTORS!!! They are there for RF stability. This is really strange advice from a guy with as many parts as he uses in his designs. The input filter belongs on the input side of the pot. 50 K is fine for the pot and easier for the previous device or stage to drive. There is already a 1 Meg to ground at the pot wiper. Walt Jung once recommended making this 10 Meg to avoid distortion from current through the wiper. I won't speculate of 10 pF cap being too small until I made simulations and measurements. The 2N4416 and 2N5462 are biased much below their Idss (Vgs = 0). The 2N4416 source is biased positive with respect to the gate. Similar rational but negative voltages for 2N5462. The feedback is probably at the correct point for the 2N4416A Vgs under bais for being equal to the BC550's base to emitter drop and output bias current times 220 ohms. Since this is a follower stage the 1.5 K resistor does not need to be capacitor bypassed The feedback is not supposed to be around the output stage. Before someone says how idiotic an approach that is you might want to look at Mr. Pass's X amps......The 150 K feedback resistor is that large to allow sufficient open loop gain. I wouldn't bother with cap bypasses on the resistors for the cascodes since a preamp usually has a good supply with little current demands. The Jfet gate will not modulate this voltage like a BJT base will.

Like I said this a very competent design....... to someone who actually knows what the hell they are looking at! Peer review demands that the reviewers are actually peers of the designer. It is not a man on the street opinion poll. I would learn a little more before I picked apart a design, who's basic operation much less it's subtleties, were beyond my grasp. On second thought, press on. We could use the entertainment.
 
Pee-r review

I understand the gain issues, but nonetheless, the values given will compromise the noise performance. I don't offhand know the en of the input FETs, but the source impdance of the feedback network will have a noise figure something like 10-20 dB worse than that of the cartridge. Perhaps choosing a different point for taking off the feedback might have been a better design choice. If the input FETs have a worse en than the Johnson equivalent of those feedback resistors, they ought to be replaced with quieter ones. In similar circuits, I've had good luck with the lowly NTE458, though no doubt the Toshibas will do beautifully well. I miss the old CM860...

Agree that the 100 ohm resistors would be a good idea to leave in. If it were going into production, that would be mandatory, with maybe a bead or two. For one-off home use, one can try the circuit without those resistors, though to what advantage, I don't know.
 
Thanks for the suggestions Fred!

Fred, I think I will give this circuit a try..

I'm thinking of using 2SK170/2SJ74 fets instead of 2Nxxx, 2SC2240/2SA970 instead of BCxxx BJT's. And use Sanyo OS-CON instead of the tantalium caps. I will also add make room for a some extra 1N4148 diodes on the PCB to increase the bias as you suggested.

What do you think about that? Worth a try?



(I already have PCB layout for it, looks quite good. Double sided, the top layer is being used as a ground plane)

This preamp is designed by Mr. Molenkamp in the Netherlands.

/Freddie
 

Attachments

  • pre-layout.gif
    pre-layout.gif
    13.4 KB · Views: 2,016
More Per review........

"I understand the gain issues, but nonetheless, the values given will compromise the noise performance. I don't offhand know the en of the input FETs, but the source impdance of the feedback network will have a noise figure something like 10-20 dB worse than that of the cartridge"


ITS A LINE LEVEL PREAMP YOU ROCKET SCIENTIST!

There are plenty of tube circuits with higher impedances than this.
At output voltages around a volt or more you really think resistor noise is really an issue? You lower the resitor values and you knock down the open loop gain by the same ratio which may not be desirable.
 
Thanks for the interesting link. But what I got out of it, plugging in some WAGs for open loop gain (500), source Z (20 ohm), unity gain frequency (400kHz), and load capacitance (150 pf), was that this circuit would have little problem driving the load and remaining stable with a more conventional feedback arrangement. And the point of this demonstration was that tuning the noise gain can stabilize things without giving up the advantages of putting a loop around the whole circuit.

No data books here in my office. If you don't fill me in with better numbers, I'll run through my assumptions this evening.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.