TILLMAN: HOW TO INCREASE OUTPUT LEVEL?

[JDoyle]

If it's not too much to ask, please pm me your telephone # so we can take this off the forum and get this straightened out since this is turning into a personal attack against me.

You are kidding.

Right?

Your ONLY responses to my points are to insult my integrity, temperament, and mental state, just as you did at the top of the post quoted, instead of any actual, oh, you know, PROOF that what you suggested was right, or even a HINT at how I was wrong, which makes YOU the first one to level personal attacks against the other.

And now that YOU feel insulted we need to take this offline?


No.


Regards,

Jay Doyle

[Dragonfly]

Who could've guessed that a thread discussing the Tillman preamp would be one that justified "locking" ?

Really...and this is just my opinion... each person has valid points. The question has been answered in the thread, however, the information is now overshadowed by all the rest of the stuff.

So Brian wants to make things as easy as possible, even if the results won't be the "best results".

So Jay wants to do things the "proper way", which gives great results, but is far more intimidating for beginners.

This is the problem when people start arguing opinions.

It is entirely possible that, even though it's the "wrong" way to do it, the OP might have found a point in the trim pots travel that gave him enough output and was still clean enough. He also might not have been able to get adequate results. He also might be intimidated by the though of having to do the math, so the project would sit untouched.

Regardless, the bickering serves nobody...especially the beginner that clicked on this thread hoping to learn something.

When I started building I did lots of things the "wrong" way. I hate math. So I avoided it and spent my time on the breadboard and using my ears. Now I use the math when I feel it will save time, but thats after a whole lot of builds. If I was a newbie interested in getting started in the hobby and was immediately pummeled with mathematical equations I never would have even started building...but because I was able to get some builds successfully completed I eventually learned how the math can help me.

Are we to discount all the hard work at www.runoffgroove.com because they regularly use trimmers on the jfets drain ? Of course not. Should we chastise them for publishing designs that utilize bias setups that could be improved ? No. Do the designs at ROG "work" and give great results. Yep. If someone builds a Thor and is having some biasing issues and asks how to fix it can we point out a better way to set up the bias ? Absolutely.

People want simple ways to get started in something and get some success as soon as they can. If they fail because they cannot understand something, then they get discouraged and go away.

In short, I agree totally with Brian. And I agree totally with Jay. You are both correct. However, the responsibility is with those of us who have been around for a while to determine what the best way to make a persons pedal work properly, taking into account their experience level, tools, desire, and needs.

========================================================================================


In hindsight, the answer to this question COULD have read something like this, which would have served all purposes, ESPECIALLY the person asking the question :

Renegadrian ....

here is the "quick and dirty" way to get a bit more boost from the Tillman, though you may or may not be satisfied with the results...it's worth a try though. Try connecting a 20k trim pot in place of the drain resistor, and see if you can adjust it and get a sound and output level that suits your needs. You could also try connecting a large (22uf or larger) cap from the J201's SOURCE to ground, though it will introduce some distortion.

If that doesn't result in what you want, then here is how you can design a preamp that will have a very similar sound to the Tillman, but have more output : Look over these application notes (Fairchild 202), and look at the series of articles by Ray Marston referenced in THIS thread and that should give you all the info you need.

=======================================================================================


Anyway, thats my completely opinionated take on things. Don't hate me for it.

[frank_p]

[MikeH]

This is getting pretty good...

[Dragonfly]

This is getting pretty good...

Nah...the first couple pages were "good"...this is just "discussion" 

[stm]

Very nice summary, Dragonfly.  I like the unbiased presentation you made about the different points of view discussed.

You mentioned both the "quick and dirty" as well as the "engineering" approach.  I feel my solution presented in posts 3 & 6 of page 1 is missing in the summary, and this one would correspond to a "practical" approach which answers *exactly* the original question, so here I go again:

Problem (rephrased by me):
How to *get some more gain* from an *already working* JFET preamp, whose owner *already likes its sound* and *doesn't want to alter it*.

Practical Solution:
*Double* the supply voltage to 18V (using two batteries or whatever) and *double* the drain resistor from 6k8 to 13.6k (I mentioned initially a value between 10 and 15k, but using two 6k8 resistors in series is the exact value needed).

--------------------------------------------------------------
(no need to read further unless you are interested in the justification for the above)

Justification:
The 'gm' (transconductance) remains the same, since it only depends on the intrinsic JFET parameters 'VP' and 'IDSS', and the source resistor 'RS' which remains the same.  Math proof follows:

The quiescent current 'idq' of the stage is given by (it is operating in the constant curren region):

idq = IDSS * ( 1 - vgs / VP )^2   [1]

But:

vgs = -idq * RS   [2]

Substituting [2] into [1]:

idq = IDSS * ( 1 + idq * RS / VP )^2   [3]

Eq [3] is a quadratic equation, so the actual value for idq and can be solved from here.  The symbolic solution is omitted as it becomes involved, but it includes a square root, as you migh anticipate.  The relevant point here is that 'idq' only depends on RS and the intrinsic JFET parameters 'IDSS' and 'VP'.

The transconductance of the stage is given by:

gm = 2 * sqrt(  IDSS * idq ) / |VP|   [4]

Again, we see it only depends on 'RS' (by means of 'idq') and the intrinsic JFET parameters.

The voltage gain of the stage around the operating point is given by the drain resistor 'RD' and 'gm':

Av = RD * gm   [5]

Thus, gain depends directly on 'RD', remembering that 'gm' remains constant as long as we do not change the JFET or 'RS', and it is totally independent on the supply voltage 'VCC'.

So, going back to my proposal, if RD is doubled, gain is also doubled, as predicted by equation [5], thus obtaining 6 dB of additional gain.

What about clipping?

Clipping to the positive supply rail is produced exactly when the input voltage equals -|VP|, which is a limit independent of RD, RS or VCC.  (As you apply a more negative voltage at the gate, idq goes down until 0 is reached, where 'idq*RS' is zero as well, so the gate-to-source voltage is exactly the input voltage or -|VP|.)

Clipping to the minimum value occurs when 'vgs' equals 'vd+VP', which is the point where the JFET enters into the triode region.  If you double VCC after doubling RD, the point occurs at the same positive input voltage for the gate.  Don't have the math proof at hand, as it is pretty involved, but I have also verified this with many different JFET SPICE models, both in AC Analysis and in Transient Analysis with Fourier harmonic contents.  The result is that the harmonic content is not affected when increasing the drain resistor by a factor, as long as the supply voltage is increased by the same factor.

In addition, one could argue that the miller capacitance increases and will reduce high frequency response.  Well, yes, it actually doubles, but as the gain is still pretty low we are talking about 10pF or so in case of the J201 in question, so still not relevant for the purpose.  A guitar cable may add 500pF or more easily.

OK, but what about the increased output impedance?  Not relevant either.  Changing from 6k8 to 13k6 will not be relevant to the sound, unless you intend to drive very small loads instead of guitar effects.

Corollary:
An intermediate supply voltage like 12V or 15V could be used, provided the drain resistor is scaled accordingly to 9.1k or 11.3k, thus increasing the gain by 2.5 or 4.4 dB, respectively.

Closing Argument:
Changing gain by adjusting supply voltage is not the preferred method, which is why a more educated approach to the problem is highly desirable.  Nevertheless, in this particular case presents an exact solution to the original problem within the bounds of the restrictions imposed.
-------------------------------------------------------------------------

[petemoore]

  Here are some questions::
  Are the answers all not right there ?
  Haven't we all said 'it is there all along' ?
  Is there something wrong with them because I don't understand them to the degree I would like to ?
  Is there a difference between those who can't get the math and those who don't want to get the math ?
  [ not that I get it, to the contrary, I somehow figure out how to ballpark around it, using 'safe' or 'hack' designs to get what I want...that or simply buy someone elses design, but I've had mixed luck with that too.
  As far as the mucka muck about who didn't answer right, and I'm no expert in english or electronics texts, but they looked like great reads to me, and I had a couple of those 'oh, there it is' moments, the kind where you say to yourself 'it Was there all along, all through the other 11 explanation approaches I read through, it's just this time it caused a spark  "
  Great reads, thanks guys !
  I'd be glad to help word questions about answer explanations, though most of that as it turns out tends toward suggestion to read the pages which are more concise and accurate, other pages if necessary to comprehend...
  Other good reads elsewhere on it, ROG has the 'Fetzer' and RG of course has the Jfet info there.
  About all there is to it besides that is work on the Jfet's operating parameters.
  Prephas choose a design that does ~what you need and doesn't require setting up Jfet operating parameters. 

[Dragonfly]

Thats a really nice, easy way of getting from "point A" to "point B" STM.  Thanks for the excellent summary, and its a option that I initially missed in my skimming of the thread. Also - thanks for explaining it so well...


The two things I think that we all need to keep in mind...

1) This should be a fun hobby
2) we need to do the things that "keep" it fun for everyone.

[JDoyle]

Great proof STM!

It didn't occur to me that gm wasn't dependant on Vds! Thanks for clearly laying that out.

However - I don't think you should set the triode region as the brick wall of the negative swing's clipping. There is actually a reasonable amount of swing left in that area and it is exactly that area, and the squashing it brings, which gives us the JFET sound.

Because one thing that DID change was the proportion of time the output signal spends traversing the constant current zone vs. the triode zone. The triode region is fixed but by increasing V+ you increased the size of the constant current region (I am sure I brutally mangled the terminology there), in terms of the load line. I have to think this will change the output characteristics. Basically, you will have less squashing. You can see this if you draw both loadlines on the transfer curves (Id vs Vds). (Edit after initial post: the difference IS pretty small, but it does exist.)

Plus, by doing a quick load line drawing for 6k8 and 13k6, the 'optimal' bias point appears to move.

I have no idea what that means, if anything.

Again, Thanks!

Jay Doyle

[aron]

While it is great that there have been very informative posts here. I cannot help but feel that this thread took way too many detours to arrive at the destination. In addition, what started out as a simple question turned into a long thread. It's fine if a "simple question" requires complex answers, but the original poster should not have to feel like they "started something".

The #1 rule here is to keep this place friendly. It serves us no good if beginners feel like they can't ask what is a seemingly simple question to them.

[stm]

However - I don't think you should set the triode region as the brick wall of the negative swing's clipping. There is actually a reasonable amount of swing left in that area and it is exactly that area, and the squashing it brings, which gives us the JFET sound.

You are correct, when the output goes "low" there is a soft transition between the end of the pentode (or constant current) region and the beginning of the triode (or ohmic) region. (Now that I think about it, using this "valve" terms contribute nothing but mojo.)

Because one thing that DID change was the proportion of time the output signal spends traversing the constant current zone vs. the triode zone. The triode region is fixed but by increasing V+ you increased the size of the constant current region (I am sure I brutally mangled the terminology there), in terms of the load line. I have to think this will change the output characteristics. Basically, you will have less squashing. You can see this if you draw both loadlines on the transfer curves (Id vs Vds). (Edit after initial post: the difference IS pretty small, but it does exist.)

I think I see your point. I haven't been able to obtain a closed formula for the transition between constant current and ohmic regions for a general case yet.  My experience comes from the many many simulations done during the revision of the Fetzer Valve article, where it was found empirically that 'vd=2*|VP|' corresponded to the transition of regions when RS was chosen as 0.83*|VP|/IDSS, which is the "magic biasing point" that produces a three-halves square-law around the operating point.  This suggests, at least for a particular case, that the ohmic region is independent of VCC, thus your statement might be correct.

Plus, by doing a quick load line drawing for 6k8 and 13k6, the 'optimal' bias point appears to move. I have no idea what that means, if anything.

Well, referring again to the experience on the Fetzer Valve optimal biasing is given by:

Vd = 0.6*Vcc + 0.7*|VP|
Rd = 0.9 * (Vcc - 2*|VP|) / IDSS

showing dependence on both Vcc and Vp.

Again, Thanks!

My pleasure.  It is good to have the chance to explore, discuss and exchange ideas about this!

[Renegadrian]

Renegadrian ....

here is the "quick and dirty" way to get a bit more boost from the Tillman, though you may or may not be satisfied with the results...it's worth a try though. Try connecting a 20k trim pot in place of the drain resistor, and see if you can adjust it and get a sound and output level that suits your needs. You could also try connecting a large (22uf or larger) cap from the J201's SOURCE to ground, though it will introduce some distortion.

If that doesn't result in what you want, then here is how you can design a preamp that will have a very similar sound to the Tillman, but have more output : Look over these application notes (Fairchild 202), and look at the series of articles by Ray Marston referenced in THIS thread and that should give you all the info you need.

=======================================================================================

Anyway, thats my completely opinionated take on things. Don't hate me for it.

Andy, I really apreciated your words and how you described your POV on either side...
I take your words of wisdom and try to follow, also I have to thank everyone who submitted some ideas, both "entry-level" and "mad-scientist" 
I hope this topic will be useful not only for me, but for everyone who wants to learn or just have a quick suggestion...

[frank_p]

I hope this topic will be useful not only for me, but for everyone who wants to learn or just have a quick suggestion...

+5 

[Paul Perry (Frostwave)]

Well, I think we have ALL learned something here.
Maybe even about how to increase the output of the Tillman.

I know I have.

[DougH]

Practical Solution:
*Double* the supply voltage to 18V (using two batteries or whatever) and *double* the drain resistor from 6k8 to 13.6k (I mentioned initially a value between 10 and 15k, but using two 6k8 resistors in series is the exact value needed).

I like your approach, Sebastian. Simple, straightforward, and clear. For a "quick fix" it doesn't get much simpler than raising the supply and making the appropriate adjustments. Sometimes we seem to suffer from "Nine-volt-itis" around here.

IMO we need to explore higher supply voltages more in a lot of projects. It's been done in the past (e.g. Mini-Tubes, Shaka HV, etc) and it would be nice to see some renewed interest in it. I've designed a few things I like to run at 24-30v because it sounds so much better. There is more dynamic range available in a lot of these devices and etc.

[alanlan]

Folks, My friend likes the Tillman I gave him so much...After all, he chose it amongst all the other boosters he tried...
He asked me if it is possible to have a little more output without changing its tonal characteristics...Maybe with a different pot?
What do you think, is there anything I can do to have some more volume?? do I have to change some components?? THX

Firstly, I think from your layout you have the added volume pot directly across R4.  You can therefore remove R4 as it no longer serves any purpose - might increase output a tiny bit just by doing that.

Secondly, I'd just put another pot somewhere around 10K to 50K (experiment a little) wired as a variable resistor in series with a 10uF cap directly across R2.  You can then dial in extra gain.  With a lot of gain, you will get more distortion - can't do much about this other than increasing headroom as has been suggested but this involves using more batteries - can't see the point really in this particular case.  If you don't want the pot ultimately, use a trimmer or a selected fixed value.

Good Luck experimenting!

[Renegadrian]

thx for your contibutions.
I will be experimenting a little - the extra voltage solution cannot be used, as it is going to get power from a 9v daisy chain with a couple other boxes.

[petemoore]

- the extra voltage solution cannot be used
  A MAX1044 could be used to nearly 2x the voltage supply.

[gez]

If this were me, I'd probably just stick a cap and series resistor (preferably equal to or smaller than the source resistor) in parallel with the source resistor.  That way you're partially bypassing the source, which will increase the voltage gain of the circuit.

Will this affect the 'tonal response' of the circuit?  Perhaps a tiny bit.  Will your friend even notice?  Doubt it...

[wampcat1]

If it's not too much to ask, please pm me your telephone # so we can take this off the forum and get this straightened out since this is turning into a personal attack against me.

You are kidding.

Right?

Your ONLY responses to my points are to insult my integrity, temperament, and mental state, just as you did at the top of the post quoted, instead of any actual, oh, you know, PROOF that what you suggested was right, or even a HINT at how I was wrong, which makes YOU the first one to level personal attacks against the other.

And now that YOU feel insulted we need to take this offline?


No.


Regards,

Jay Doyle

my apologies jay, I should have said since our comments are getting personal, I would like to talk to you rather type.
Thanks,
Brian