J201 JFET - fabulous friend, sneaky foe (comment/discussion)

[amz-fx]

The following is an extract from Donald Tillman's article that popularised the J201 amongst guitar pedal builders.

Do a search in the "First Stompbox Forum" archive and it will show that the J201 was in widespread use on this forum (at its previous site) several years before Don decided to change out the 2N5457 for the J201. I would contend that the J201 popularity came from here instead of his revised article that came along a few years later. 

The J201 has a greatly reduced output swing on the Tillman preamp because the drain voltage biases high.
(No problem with the 2N5457, the original part used, despite what the website now shows.)

As I recall, at the time the 2N5457 was quite popular but the available devices began to vary all over their spec range and made it difficult for beginning builders to get the best performance from their circuits. The J201s were more consistent, had low noise and gave a bit more boost for what was being built here at the time.

There are still some TO92 jfets available on Mouser at fair prices and anyone trying this circuit could test it with the J113 or the J112 and see what kind of results those fets give.

Best regards, Jack

[FSFX]

The following is an extract from Donald Tillman's article that popularised the J201 amongst guitar pedal builders.

Do a search in the "First Stompbox Forum" archive and it will show that the J201 was in widespread use on this forum (at its previous site) several years before Don decided to change out the 2N5457 for the J201. I would contend that the J201 popularity came from here instead of his revised article that came along a few years later. 

I was really just rephrasing what Teemu Kyttälä said in his book

"For example, J201 is a commonly used FET - probably because it was once used in the famous "Till"
guitar preamplifier and "FET Preamp Cable", both designed by Donald Tillman."

[Clint Eastwood]

The J201s were more consistent, had low noise and gave a bit more boost for what was being built here at the time.

There are still some TO92 jfets available on Mouser at fair prices and anyone trying this circuit could test it with the J113 or the J112 and see what kind of results those fets give.

Best regards, Jack

I can confirm that the J113 works very well in this and many other guitar related circuits. They are pretty consistent, have lower noise than J201 (not that I noticed the difference) and a relatively high amplification factor. I have absolutely no beef with the J201, but
do wonder where the myth comes from that it is low noise

[duck_arse]

https://www.interfet.com/jfet-datasheets/jfet-j201-j202-interfet.pdf
https://pdf1.alldatasheet.com/datasheet-pdf/view/121328/VISHAY/J201.html

I have absolutely no beef with the J201, but
do wonder where the myth comes from that it is low noise

2 possible sources linked.

[PRR]

...no beef with the J201, but
do wonder where the myth comes from that it is low noise

If your source impedance is over about 3kOhms, source self-hiss will exceed J201 hiss. By 5k or 10k, any clean FET you can find (say post-1970) will be "low-noise" relative to the source.

This means E-guitar, MM phono, most tape heads, any rational step-up input-transformer mike input, many-many line inputs.... it's not about the JFET. You can have biasing and overload troubles, but if you get it happy, the FET noise is not a limit.

MC phono and 1:1/1:2 mike transformers do want a fat (high Gm) FET.

[Clint Eastwood]

https://www.interfet.com/jfet-datasheets/jfet-j201-j202-interfet.pdf
https://pdf1.alldatasheet.com/datasheet-pdf/view/121328/VISHAY/J201.html

I have absolutely no beef with the J201, but
do wonder where the myth comes from that it is low noise

2 possible sources linked.

Well, that's interesting. Interfet says 5nV/Hz, Vishay says 6nV/Hz, and Fairchild (application note 6609) states 10nv/Hz. Looking closer, the latter is measured under different conditions, so that probably explains the difference.
So let's say it is 5nV/hz. This is 'very low noise' according to Interfet and Vishay. But compared to what?
To make a fair comparison, let's look at data for different jfets from the same manufacturer, Interfet.
2n5457  - 7nV/Hz, described as  'Low noise'.  This is noisier than the J201, but not a huge difference
2n5484  - 4nV/Hz,  'Low noise'. This device is suitable for VHF applications, equivalent to MPF102, BF245. It has a lower noise figure than the 'very low noise'  j201 
J113  -  1,2nV/Hz,  'Low noise'. Marketed as mainly for switching and chopper duties.

Clearly, if you call a J201 very low noise, it would be logical to say a J113 is like  'ultra low noise'.
Conclusion: it seems that the confusion about jfet noise performance is caused by inconsistent descriptions in datasheets.

By the way, I looked at the data for the LSK170, made by Linear systems and a very highly regarded (and pretty expensive) audio device. It is has a noise figure of 0.9 nV/Hz, not that much better than the ordinary 50 times cheaper J113

[PRR]

.... confusion about jfet noise performance is caused by inconsistent descriptions in datasheets.....

JFET hiss in the audio band at sane impedances is almost all 1/Gm resistance thermal noise. So makes sense that fat switches rate well.

You can't test noise for-sure in production. It takes a closed chamber and time (if you are going to test, some markets expect 1/f zone guarantees). Unless you pay a fortune for slow tested parts, you get lot-tests and a huge margin to cover bad-day variation. And there is considerable more variation in JFETs than in BJTs. So JFETs find few niches. So get sloppy specsheets and pass/fail testing.

[Rob Strand]

Do a search in the "First Stompbox Forum" archive and it will show that the J201 was in widespread use on this forum (at its previous site) several years before Don decided to change out the 2N5457 for the J201. I would contend that the J201 popularity came from here instead of his revised article that came along a few years later. 

Jack, that is absolutely the case!

I put it out there, if a newbie presented a design, which was the same as the Don Tillman Preamp with a J201, and put up the tests voltages there would be quite a few people flagging the bias point is way off.   The design with the 2N5457's is textbook.

As I recall, at the time the 2N5457 was quite popular but the available devices began to vary all over their spec range and made it difficult for beginning builders to get the best performance from their circuits. The J201s were more consistent, had low noise and gave a bit more boost for what was being built here at the time.

And now a lot of J201's are dodgy.   As the fake (and out of spec) devices come in we see more and more problems.

Well, that's interesting. Interfet says 5nV/Hz, Vishay says 6nV/Hz, and Fairchild (application note 6609) states 10nv/Hz. Looking closer, the latter is measured under different conditions, so that probably explains the difference.

To first order the Jfet noise depends only on the drain current bias point, and not on the type, that sets the transconductance (yfs, gm).   The high Yfs0 JFETs (ie. low resistance) generally produce less noise because they *can* be operated at a higher transconductance.   Low Yfs JFETs are limited by their maximum Yfs.   The point is, the operating point can have an effect and datasheet don't compare apples to apples.

On top of that is a whole lot of secondary effects.  A major one is 1/f noise.   If you have a higher 1/f noise and the JFET noise spec is a 1kHz then it's quite possible that this alone can determine the published noise.  That requires the operating transconductance is high enough.

BJTs have similar issues.  Some BJTs are quoted as low noise.  Some are low noise *for high impedance circuits* because the secondary effect which cause noise in high impedances are reduced.  Similarly there's BJTs which are low noise *for low impedance circuits*.  These tend to have low base resistance (rbb') which appears in series with the base and adds noise.    This is on top of the fact that to first order all BJTs produce the same noise when the operating current is the same.

The source impedance used in spec can have an effect on the published noise as well.


This document gives a reasonable overview,
https://www.vishay.com/docs/70599/70599.pdf

[Rob Strand]

I pulled up some quite old J201 (/J202/J203) datasheets from Motorola.  They are quite sparse.  The datasheet states: see 2N4220 for graphs.  This was common in the Motorola data, it was good because it leaked info about commonality between parts.

When we look at the noise plots for 2N4220 and 2N5457 (Motorola), it's the same plots!

One caveat is the J201 specifically has parameters outside of the 2N4220 range.  The J201 have a high channel resistance, in fact they have a higher channel resistance than most JFETs.

[brett]

That would be good, but it was at 0.25 V.  I "squeezed" it a little further and got 0.27 V without throwing Vd too far away from the often-recommended Vs/2.

The 0.6V is the JFET parameter VP and in-circuit the JFET operates at a gate-source voltage less than that, like 0.25V.

If you measure across the gate and source the meter loading can drop the measurement by a factor of 1/2 for a 1M ohm input DMM.  However, if you measure the source voltage and the meter loading has (virtually) no effect.

Good point.  My DMM is 10M input, but thinking back, the difference was there.   
Thanks. 

[brett]

The results of others might vary, but I live in a quiet apartment and was recently disappointed to find that this J201 project resulted in a really noisy booster (ie I could hear the amp when I turned it on.  I always aim for on-off to sound the same).  This is despite the gain of the booster and the preamp+amp being quite low. 
What?
As it turns out, these circuits pour ripple through.  With a basic plugpack supply, unfiltered, I was getting 0.03mV of output ripple, and with gain of 5 x, then 12AX7 triode boost of 60x, and 6J1 boost of 5x that was about 0.05V output (and 0.05W) into a speaker with 95 dB/W@1m.  Around 80dB !  I needed an RC filter of 1k ohm and 220uF to calm it down.  It's still noisier than I'd like it to be (maybe 40dB).

I haven't calculated anything concerning JFET noise at the speaker, but I'd be surprised if many people go as far with power supply filtering as my RC of 1kohm and 200uF.  Even with this, there's still no chance of me hearing the JFET noise.  Yes, my plugback is probably poor (transformer, not switching, Moen brand IIRC).

Here's a similarly filtered booster/buffer.
cheers

[PRR]

> these circuits pour ripple through

Well, yeah. It's got about no PSRR. (The FET 'plate resistance' is over 100k, so ripple is hardly attenuated.)

You can't power simple preamps with crap power.

[FSFX]

I pulled up some quite old J201 (/J202/J203) datasheets from Motorola.  They are quite sparse.  The datasheet states: see 2N4220 for graphs.  This was common in the Motorola data, it was good because it leaked info about commonality between parts.

When we look at the noise plots for 2N4220 and 2N5457 (Motorola), it's the same plots!

One caveat is the J201 specifically has parameters outside of the 2N4220 range.  The J201 have a high channel resistance, in fact they have a higher channel resistance than most JFETs.

It seems that Fairchild never actually published true graphs of J201 characteristics. The J201 graphs on the datasheet bear no relationship to the numerical date. In fact they are the same as the J211 graphs.

[Rob Strand]

As it turns out, these circuits pour ripple through.  With a basic plugpack supply, unfiltered, I was getting 0.03mV of output ripple, and with gain of 5 x, then 12AX7 triode boost of 60x, and 6J1 boost of 5x that was about 0.05V output (and 0.05W) into a speaker with 95 dB/W@1m.  Around 80dB !  I needed an RC filter of 1k ohm and 220uF to calm it down.  It's still noisier than I'd like it to be (maybe 40dB).

Without using regulators, it's not uncommon to use second order filters to get more attenuation.  They are very easy to whip-up when you are testing.

Here's a few common circuits you come across to clean up power rails.

[Rob Strand]

It seems that Fairchild never actually published true graphs of J201 characteristics. The J201 graphs on the datasheet bear no relationship to the numerical date. In fact they are the same as the J211 graphs.

The J201 data was never good but that one sure looks all over the place.  Do you know what year that data is from? 
There are few eras of National Semiconductor and Fairchild, as they went through different owners.

I thought the National Semiconductor 1997 FET databook might be the go.   That has graphs but they cover the whole process and the low Vgs(off) case for the J201 isn't represented well.

[FSFX]

The J201 data was never good ....

It is not just the J201 as can be seen here too.

[FSFX]

Do you know what year that data is from? 

I am not sure of the dates as the same data has been re-published a few times and even persisted when ON took over. The National FET databook that I have (1977) seems to only have numeric data.

The graphs for Process 52 in the ON Semi application notes are probably a good representation of the J201 characteristics. 

A few years back, I emailed ON Semi to point out the datasheet errors.

[Rob Strand]

It is not just the J201 as can be seen here too.

I can see how the link to the J111 etc. graphs works.  There's no intent to match the tabulated data.    Think of the PF5102 as *another entry* in the J111/J112/J113 tabulated data.   The PF5102 covers Vgs(off) from -0.7 to -1.6V, the J113 covers Vgs(off) from -0.5V to -3.0V, etc.   The PF5102 has tighter tolerances than the J113 and sits somewhere the near middle of the J113, you could even think of it as a tight tolerance J113.     So the graphs for the J111 etc. family still apply.    The tabulated data for the PF5102 is fully specified with min and max (no typ.).  The tabulated data for the J111 etc are very loose in that there are no maximums on IDSS and no minimum on Rds_on.

I am not sure of the dates as the same data has been re-published a few times and even persisted when ON took over. The National FET databook that I have (1977) seems to only have numeric data.

The graphs for Process 52 in the ON Semi application notes are probably a good representation of the J201 characteristics.

A few years back, I emailed ON Semi to point out the datasheet errors.

The NS FET Databook (1977) I've seen has all the process info.   It's actually *very* similar, if not identical, to the Fairchild/ON info.  In the front of the book were tabulated data, not quite complete datasheets.

In the past I've spent a lot of time matching the info on the processes with the JFETs data they correspond too.   It's a time consuming and frustrating task.  Things don't match up.   Datasheets like the J201, J111 which have incomplete tables which don't help.   I've also tried to use datasheets from multiple manufacturers for cross-checking and filling in some of the blanks.   There's simply not enough info to fill in the blanks, and there's so many things that don't match up between the graphs and the tabulated data.   Every time I've done this type of thing I end up tell myself not to do it again now I just chant "Insanity is doing the same thing over and over again and expecting different results."

[Rob Strand]

FWIW, back in reply #19 I posted a J201 version of the Don Tillman preamp which behaves like the original 2N5457 version.
Here's a lower parts count version of the circuit.



The gate bias shift trick can be use to help any low VP JFET (which are now common on ebay) behave like a higher VP JFET.    However, if the JFET also has a low IDSS you may have to increase the resistor values to allow the JFET
to clip correctly on the -ve output swing, as I have done in this circuit.

[Paul Marossy]

Do a search in the "First Stompbox Forum" archive and it will show that the J201 was in widespread use on this forum (at its previous site) several years before Don decided to change out the 2N5457 for the J201. I would contend that the J201 popularity came from here instead of his revised article that came along a few years later. 

Jack, that is absolutely the case!

I remember being over at Ampage forum when Aron said he was going to start his own forum. I guess that was around 2001-2002? Anyway, yeah I built quite a few of those J201 based circuits that people were presenting here. They were coming out of the woodwork back then. I still use some of those circuits today.