Jfet pre

[Emmathom]

I think I forgot a gate-stoper for the 1rst Jfet (1KΩ ?) and no need of 100nF at input because of the transformer upstream
Revisited schematic below...

 

[Hubbub]

Looks like you forgot the drain load resistors for the FETS, and an NPN with base and collector shorted is nothing more than a diode. I also think the XTX653 collector currents will be rather unpredictable, since the base is purely current driven rather than biased by a voltage divider. No need to use bipolar coupling caps as both operate with about 12V DC across them.

 

[Emmathom]

Looks like you forgot the drain load resistors for the FETS, and an NPN with base and collector shorted is nothing more than a diode. I also think the XTX653 collector currents will be rather unpredictable, since the base is purely current driven rather than biased by a voltage divider. No need to use bipolar coupling caps as both operate with about 12V DC across them.

yeah, you're right about R drain ! I've corrected...

As said at the start of my post, I don't have experience with jfet circuits so I relied, on the advice of a forum member, on the Scott Hampton's proven schematic. I posted his article previously.

What changes do you think would need to be made?
Do you think it can be done without blocking cap between the 2 stages? at output ? There is no risk to get 12v DC in the modulation?

 

[rock soderstrom]

@rock soderstrom I just read your intervention in https://groupdiy.com/threads/looking-for-recommendations-preamp-a-perfboard-edit-fetboy.70359/page-2
May I ask you how you get "tons of colour" ? by different R source capacitors values ?
You said : My advice is the Hamptone JFET Preamp aka Fetboy. Easy to build, tons of colour as micpre and DI for instruments.

Yes, I mean it the way I wrote it, this circuit can go from clean to heavily distorted. A sound machine, ideal for rock and similar stuff. If you want the recording to have a certain sound, the Fetboy is the right choice. If you want to record classical music or other clean stuff, this is the wrong preamp IMHO.

BP Muse Nichicon are very good quality for coupling caps in audio and often used

Don't go too fancy with the components and first build the circuit as described in the article. It is best to simply build the Fetboy on a test board and make your first experiences with it. Then you'll quickly see if you like it - the component effort is super low.

Then you can mod the amp to your liking. My favorite mod for the Fetboy are switchable source capacitors, then you can flexibly adjust the sound and especially the THD.

Hampton insists on this value but a 10uF MKP is big, almost bigger than the whole board

10uF is right, you don't need a dielectric strength for 400V as this is not a tube circuit. Wima has nice 50V MKPs that are nice and small and take up almost no space

 

[Emmathom]

Thanx for your add !
Ok to go as the schematic with 10uF intersatge blocking cap.
According to my calculations, with a R to ground of 10KΩ (the pot.interstage) -3dB at 3Hz is reached with 5,3uF... maybe Scott put 10uF in case of 5KΩ pot ?
but the main thing is the output : with 100KΩ to ground, -3dB at 3Hz is reached with only 1/10 so 530nF > why putting such value as 10uF ? 1uF should be sufficient... ?

 

[rock soderstrom]

As said at the start of my post, I don't have experience with jfet circuits so I relied, on the advice of a forum member, on the Scott Hampton's proven schematic. I posted his article previously.

What changes do you think would need to be made?
Do you think it can be done without blocking cap between the 2 stages? at output ? There is no risk to get 12v DC in the modulation?

As already mentioned, first set up the amp as designed, especially if you have little experience.

The Fetboy is a tried and tested design with a few known quirks, everything is documented here in the forum. The crux of this amp is that you need the right (selected) FETs so that all voltages are correct and no components downstream gets too hot.

If you are successful, the result will be an amp that delivers what it promises. Very flexible sound tool and very popular with recording engineers. The DI input is also very useful, not only for stringed instruments but also organs, synthesizers and other sources that need drive.

I built four of them, three of which I passed on to friends and which have been in constant use for a decade.

 

[Emmathom]

Nice !
This circuit is gonna be "fed" by OEP X187B input transformer and will output asymetrical (no need to symmetrize as my pre is in a flight next to my audio card > 30cm / and I should get enough gain to not need extra +6dB)

 

[rock soderstrom]

will output asymetrical (no need to symmetrize as my pre is in a flight next to my audio card > 30cm / and I should get enough gain to not need extra +6dB)

Same here...

I think I have shown this device before. It was my (now in a box) test device and still works perfectly.

 

[Voyager10]

As @Hubbub mentioned, the circuit as shown doesn't have any real control over the output stage current (which flows through the MPSA14 and ZTX653 transistors). Basically it will be the current in the 47K resistor times the ZTX653's current gain, which could be huge. It's also unpredictable: a ZTX653 you buy now may well have a higher gain than a typical ZTX653 when the circuit was designed.

There's a simple fix, which is to put a resistor (say 3.3k) between the base of the ZTX653 and ground, which will fix the base voltage (at 1.5V or so) which fixes the current in the 100 R resistor (about 9mA, which is more than adequate for driving a line inout).

 

[Emmathom]

As @Hubbub mentioned, the circuit as shown doesn't have any real control over the output stage current (which flows through the MPSA14 and ZTX653 transistors). Basically it will be the current in the 47K resistor times the ZTX653's current gain, which could be huge. It's also unpredictable: a ZTX653 you buy now may well have a higher gain than a typical ZTX653 when the circuit was designed.

There's a simple fix, which is to put a resistor (say 3.3k) between the base of the ZTX653 and ground, which will fix the base voltage (at 1.5V or so) which fixes the current in the 100 R resistor (about 9mA, which is more than adequate for driving a line inout).

Thanx for the add ! (24v*3300)/(3300+47000) = 1,575v
So this schematic...

 

[Emmathom]

According to my calculations, with a R to ground of 10KΩ (the pot.interstage) -3dB at 3Hz is reached with 5,3uF... maybe Scott put 10uF in case of 5KΩ pot ?
but the main thing is the output : with 100KΩ to ground, -3dB at 3Hz is reached with only 1/10 so 530nF > why putting such value as 10uF ? 1uF should be sufficient... ?

Anybody has an answer about cap values ? my calculations are right ?
4,7uF for interstage with 10KΩ pot. / 1uF for output with 100KΩ to ground

 

[Voyager10]

4.7uF for inter-stage coupling should be adequate, it gives you well under 0.5dB down at 20Hz.

For the output, the relevant value is 100K in parallel with the input impedance of your line input. That's probably going to be in the 10-50K range, so again 4.7uF would be needed, or maybe 10uF-22uF to play it safe.

 

[GJB17]

Hi people... speaking of FET preamps... I have done a fully balanced unit that has noise and distortion that are unmeasurable (at least with my gear!) It uses a matched pair of (Toshiba) 2SK117's coupled to an LM4562. If anyone is interested I can post the circuit. BTW I have a large collection of NOS (genuine) Toshiba FETs if any body needs.

 

[Emmathom]

You're damn right ! How could I forgot the 10KΩ (sound card) in // with the 100KΩ... !!!
Ok for 4,7uF interstage + 4,7uF output (at least)

 

[Emmathom]

Hi people... speaking of FET preamps... I have done a fully balanced unit that has noise and distortion that are unmeasurable (at least with my gear!) It uses a matched pair of (Toshiba) 2SK117's coupled to an LM4562. If anyone is interested I can post the circuit. BTW I have a large collection of NOS (genuine) Toshiba FETs if any body needs.

Hi @GJB17 Yes your schematic must be interesting ! It'd be nice to show us. I've already LM4562 in stock and also K170 (4 of them / real ones same IDSS / bought in the 2000's) > but no K117 in stock... (J113 yes around 30 items)

 

[Mr. Moscode]

I think you can use other similar JFETs but I would stick to the 2N5457. Buy more than you need, because you have to select them by hand.

This is a good source:

https://www.musikding.de/2N5457

This one is cheaper, but I never bought from them:

https://www.amazon.de/Chanzon-2N5457-N-CHANNEL-Feldeffekt-Transistor/dp/B08M3YJR9Z

J310 is a pretty hi gain jfet.

 

[Emmathom]

2N5457 are also available at Banzai as MKS2 Wima 50v 4,7uF / 10uF / etc... and Nichicon Fine Gold 470uF 25V

 

[rock soderstrom]

2N5457 are also available at Banzai as MKS2 Wima 50v 4,7uF / 10uF / etc... and Nichicon Fine Gold 470uF 25V

Olaf has really great things and quite a lot of them.

 

[thor.zmt]

I would like to replace it with a jfet-based circuit, more "tube-sound-like" because the sound of the BJT circuit is not of much interest.

J-Fet's do not inherently sound "tube like", they have 10 to 100 times the distortion levels of Tubes in a simple common Source/Cathode circuit.

I know, however, that jfets are "capricious and delicate" to implement and I would like to get your advice and if possible a viable schematic.

I do not know who considers J-Fet's "capricious and delicate".

The only issue is that you need to design keeping part to part variation in mind.

I post below a schematic that I found on the net as well as the link leading to it. And also a link to the excellent article on Jfets by Rod Elliott.

I suggest Erno Brobeley's articles on J-Fet's.

https://audioxpress.com/assets/upload/files/Erno BorbelyJFETs The New Frontier Part 1and 2.pdf

I am unenthusiastic about anything shown in the thread so far. But then again, it depends what the goal is.

I redraw the Hampton schematic... but I have a few questions :
- is the 100nF as signal input blocking cap is ok ?

Do not use input blocking cap's. They are pointless.

- is the output at a sufficient low impedance to avoid any adaptative circuit ? it's gonna "feed" a Line In sound card (10KΩ)

The Emitter Follower is suboptimal at best.

* I think 100KΩ at output after R100 is better than 10KΩ (?)

Use 1M even.

I think I forgot a gate-stoper for the 1rst Jfet (1KΩ ?)

J-Fet's normally do not need gate stoppers. All they do is add extra noise.

Revisited schematic below...

Honestly, this schematic and the original Hampton are good examples how not make a design that is intended to be produced or to be built by inexperienced builders. Performance, well....

The J-Fet specified is process 55, which is "general purpose amplifier", where a process 51 part would be a better choice (Low Frequency, Low noise).

See more here:

https://www.onsemi.com/pub/collateral/an-6609.pdf

For most J-Fet's the part number is inconsequential, it reflects marketing, selection, basic performance is down to process.

The JFET specified also has a VERY LARGE variability of Idss or Vgs(off), meaning it will be hard to get the correct drain voltage for maximum output. On top of that it is run at very low current, where the variability is even larger.

The CCS relies on the Beta of the CCS transistor. Seriously? I plugged this into TINA (MPSA14 is a Darlington for which there is no model, but a discrete Darlington with BC550 will do fine) and here is what we get:


The current in the follower is way too high. We probably want around 10mA, not 40mA, but I will tweak the base resistor to get 40mA:



Ok, if we do not tweak the J-Fet Operating condition, what is the result?

Gain is ~26dB. Maybe that is what is needed, but seems low. A single stage 12AY7/12AX7 will have way more gain.


Harmonic distortion for 10mV in and 190mV out is ~0.5%!



I only know one tube that approaches these levels of non-linearity and that is not really a tube, but a VFD abused as gain device (aka Korg Nu-Tube).

If we increase the output to ~1.9V, we get 5% THD!!!!



At 3.8V out we get severe distortion :



Noise seems acceptable at 10mV in with 87dB:

Now perhaps such levels of HD are wanted, to me they seem very high and I am not a fan of low THD as such, but the standards of even a very basic non-feedback tube circuit, this level of performance is absolutely dreadful.

I would described especially distortion performance as derogatory caricature of a tube Circuit. In my view this circuit lacks ALL the true qualities of a Pro-Audio tube circuits and offers nothing of compensatory value. Ok, it produces gain.

It would be trivial to improve the ability of the circuit to be replicated more reliably, about that's kind of besides the point. Here something that is slightly adjusted to ensure it can be replicated reliably:


With optimised operating point distortion can be lowered. This circuit would have to be build and tested to be sure performance is optimised at 8.5V at TP1, but the models in TINA TI are pretty good.

Still the performance is pee-poor by open loop tube standards. Maybe not quite "tube screamer" levels of overdrive, but these levels of distortion will audibly colour the sound.

Thor

 

[thor.zmt]

Still the performance is pee-poor by open loop tube standards. Maybe not quite "tube screamer" levels of overdrive, but these levels of distortion will audibly colour the sound.

Being critical is easy, it behoves him who criticises, to show a better way.

If we actually want the kind of circuit that behaves like a REAL tube stage, something like this would do it:



First, use Process 51 Parts and run the JFET heavily degenerated with active load (and R9/R10 form the actual load that sets the gain), as shown we get ~26dB gain.

The frontend, as long as JFET's of the same type and from the same reel are used will balance out pretty well, even 100% differences in Idss only shift operating conditions by ~0.5V or so.

R2 degenerates (aka local feedback) and linearises the JFET's transconductance, to around 0.7mA/V from well over 10mA/V.R9/10 then set the gain with the net transconductance. Increasing Values will increase gain while lowering will lower it, BUT the distortion relative to input voltage remains constant!

The Buffer is a simple Mosfet (which does need a gate stopper and should have gate to source zener to protect it which I have not drawn so Khinese Kopy Katz trawling this for stuff to Kopy find their Kopy blows up into their face on every switch on).

Predicted HD at 0.2V out is around 0.0025% (-92dB), at 2V it is 0.02% (-74dB) and at 4V out (shortly before clipping) 0.05% (-66dB)



The Harmonic spectrum is pretty much Classic triode (this is just before hard clipping):



If all we drive is a pot (say 20kOhm) we can simplify the circuit to omit the buffer and add that as input stage, giving a complete preamp circuit with 52dB maximum gain. Let's adda 1:4 input transformer (I have been modelling something using that cheapo e-bay/aliexpress 600:10k so I threw that on):


For 2.5mV input and the gain control at -6dB (worst case for noise) and 2.5mV input we get +6dBu output, 0.015% THD and ~75dB unweighted SNR, at least in the Simulator:


I think this all makes a fairly acceptable performance. If our microphone is "bogey" 10mV/1Pa we are looking at 82dB for 2.5mV and thus the Preamp Noise would be somewhere equivalent to 7dB absolute for this situation.

TBH, the output limit at around +17dBm seems a bit weedy, I think a redesign to run on 48V (which also gives Phantom power) would fix that. I will leave that to someone else. At 48V we can then use a Wellerman, I mean Meanwell 48V LED power stick and voila, we are good to go.

Thor