Boogie MKIV Lead channel JFET version, Oscillating like mad

[J M Fahey]

Roly, I'm sure you are not saying the heresy that these guys should .... *DESIGN* ?  :duh :duh :loco
Meaning they should grab their calculators, put new batteries in (the original ones must be flat from lack of use) and ... actually *calculate* some bias points ? 
What use would the trimmers be then? 

[stormbringer]

Now, getting rid of trimmers sounds interesting. I know i could use trimmers to bias and then replace with resistors, but actually solving the stuff with maths before my next build and modify the design accordingly sounds both interesting, meaningful and fun, and a good step on the way of actually learning to roll my own, instead of just building other peoples designs.

Got any link with information about bias point calculations off the top of your head? Although i'm quite sure i will spend the next few hours googling anyway.

[Roly]

Roly, I'm sure you are not saying the heresy that these guys should .... *DESIGN* ?  :duh :duh :loco
Meaning they should grab their calculators, put new batteries in (the original ones must be flat from lack of use) and ... actually *calculate* some bias points ? 
What use would the trimmers be then? 

Weeel hush mah mouth!   :duh


Okay @stormbringer, find, download and install LTSpice; join the Yahoo LTSpice group, find the model for your favorite FET (MPF102's happen to be very available where I am).

Take a sheet of paper (you remember paper, don't you?) and draw two lines across it, one representing ground, the other representing a +ve supply voltage.  Mark the latter "+18 volts" to give yourself a flying start.

Draw a jack socket symbol between the lines far left.  Draw a resistor from the socket hot to ground and mark 1 to 10Meg.  Draw a cap in series with the input and mark it 0.1uF.

Now it gets complicated.

Draw an N-channel FET.  Connect the 0.1uF to the Gate, and add a resistor from Gate to ground and mark it 1 to 10Meg.  Add a resistor and electro cap in parallel to ground from the Source.  Add another resistor from Drain to +ve supply.

Now at this point the designing starts, and it starts by guessing the stage current is going to be 1mA.  You now compute using Ohm's Law the value of the Source resistor to give you the desired bias voltage for the FET (obtained from the datasheet), which we will guess as being about 4 volts.  So we can write next to the source resistor 3k9 to 4k7.

Now given an 18 volt supply and the desire to have the largest possible swing before clipping, and accounting for the fact that we have just given up 4 volts biassing the FET, the available swing at the Drain will be 18-4 = 14 / 2 = 7 volts, but since it is sitting on top of the bias voltage the target resting Source voltage is 7 + 4 = 11 volts.  11 volts below the supply of 18 is 7 volt drop across the Drain resistor, and at 1mA this suggests a 6k8 Drain resistor.

Either in LTSpice or on the bench we can now knock up our first stage and see if we get the desired 11 volts on the drain, or something near it.  If we don't, we need to change something in the right direction, more sums, more measurements.

Finally we have the DC conditions where we want them and it's time to look at the AC conditions.

We can check that the reactance of the input blocking cap is small compared to the surrounding resistances so the low frequency cutoff is acceptable (Xc = 1/(2 Pi f C), Xc should be less than one tenth the surrounding resistances).

Now what about the Source bias resistor bypass cap?  How big should that be?  On the face of it you would do the same thing using the source bias resistor which might give you a few microfarads, but when you try it you find the low frequency rolloff is unacceptably too high, so what is going on?

The impedance in the Source circuit is actually a lot lower that it looks, in fact about the Source resistor divided by the gain of the FET, perhaps a hundred times lower than expected.  Simple, bump the bypass cap up by a factor of one hundred to say 100uF or 220uF 6.3V.

Next we generally go in to a tonestack.  Do you want to include a "follower" stage to drive it at very low impedance?  Simply rip off the Fender tonestack from somewhere, and the best place is to download Duncan's Tone Stack Calculator which will model it for you allowing you to play with the values and see what they do, and have a lot of fun in the process.

The recovery stage after the tonestack?  Well why not simply copy the first stage you just designed?

So, brick by brick you move down the signal chain until you have a complete preamp, then a whole amplifier.


Some instruments are very handy, and In Ye Olde Days of Yore an early and important task was to build your own audio oscillator and audio millivoltmeter (and perhaps even oscilloscope).  These days such stuff is pretty damn cheap, and you can also get software for an old clunker computer to provide signals, look at waveforms, and even do spectrum analysis.  But which ever way you go some instruments are vital to extend your senses past just your ears (which are vital too, but not enough on their own).

So how did this work out?  LTSpice result appended.  Not perfect and need tweeking, but not too far off the mark for a first stab.

HTH

[stormbringer]

Wow Roly, that was more than i could ever hope for! Thank you for taking time explaining and doing this!

Will start experimenting right away

An oscilloscope is high on the wish list, been reading about different DIY variants lately, even though buying good tools is important, i love building, You learn so much more that way.

Again, i feel incredibly grateful!

[Roly]

Well, now it's up to you to kick it along by lashing this stage up, birdsnest/"aeroboard", strip or dab board, or even my least favorite protoboard, record the initial voltages at Source and Drain, then tweek the Source resistor to make the Drain voltage rest in the middle of its available swing range, and let us know what you found, how one build reality compares to guesswork and a Spice model.

Once the DC conditions are right you can try feeding some signals through it, maybe from guitar to an amp, and confirm that it amplifies (about 20dB, in voltage terms);

dB = 20*log₁₀(Vout/Vin),

20/20 = 1

log(x) = 1, therefore x=10, so it should have a voltage gain of ten times.

If we assume a peak signal from a guitar of about 1 volt you can see why I started with a much higher supply voltage than the original circuit, but anything from 12 to 24 volts would be reasonable.

[J M Fahey]

Roly , congratulations.
Excellent explanation and design.
Plus values found are confirmed by *real World" testing.
My typical FET stage (which is roughly similar to the one above) gets 20X gain, but is fed around 30V and uses a correspondingly higher load resistor.
Just for kicks, redo calculations/simulation with only 9V supply available and check the *dismal* performance available.
Even more because instead of a rational, calculated "cathode"(source) biasing resistor, they use "the same value as in a Marshall or Fender"  :duh :lmao:
WHY, oh WHY!! Is there some Magic involved?
So in practice *all* those "converted from tubes" circuits are *terribly* UNDERbiased.
Of course, "trimmers solve it all", don't they?  :grr
I'll never understand how somebody "clones" a 50x gain stage (12AX7) , which has actual 3x to 5x (no kidding) signal gain, which clips (ugly) with 1/10th what a guitar pickup provides, and feels happy because he "built a Twin in a matchbox" or some similar nonsense.
Oh well.

NOTE: not forgetting the 3:1 spread in FET parameters, even if pulled from the same box. 

[Roly]

Aww shucks.  :crazy2:   It's just something I knocked up out of some old fruit crate I had kicking around.   

What am I trying to say?  You can take a highly math-y approach with hybrid-y parameters and all, grind a lot of numbers, but back in the real world we have the spread of FET characteristics (MPF102 I_dss spread is 10:1 IIRC) which tend to laugh at our spurious accuracy.

The seat of the pants "design" above might be a bit dubious for production, but if it's going to be reproduced in ones by individual builders who can easily tweek source resistor values (or try a different FET) if required, then "near enough is good enough" (and the operational quality of the final hand-crafted product may well be better than one off a production line).

The 18 volt supply assumes a mains powered amp, and with the MPF102 I think you could even take that up to 24 odd volts to good effect.  But if you really must operate from 9 volts then just grab a suitable op-amp (but yeah, we are back into having to actually design something, do a couple of multiplications and divisions  ).

I really can't see a valid excuse for all those trim pots when it is so easy to mock up something on the bench and try some different values - and hopefully recognise the basic problem is not enough supply voltage.  And it's even easier now we have things line LTSpice free for the downloading - no more burnt fingers. 

The initial preamp above is not just lazy, it's failing to learn by doing - and then you publish your "design" on the net!  It's not that there weren't some terrible bloopers in the days of paper magazines, it's just that it is so much easier these days to stamp something "verified" while bypassing any sort of error checking.

Personally I'd be thrilled if a credible "StormBringer" preamp design emerged from this - great name.   :dbtu:

[stormbringer]

Yes, the stormbringer amp will see the light of day some day. That's the goal.

Currently experimenting with that one gain stage with the 3 different jfets i got a bunch of in stock, j201, 2N5457 and MPF102. Drawing it and tuning in LTspice, then mirroring the changes on breadboard listening and recording. as i dont have a scope at the moment, I simply record through a Zoom R16 and comparing waveform amplitude in audio editing software. This is really interesting!

[joecool85]

Personally I'd be thrilled if a credible "StormBringer" preamp design emerged from this - great name.   :dbtu:

:tu:

**edit**
Oh, also I would suggest starting a new thread if this StormBringer continues in development.

[Roly]

Personally I'd be thrilled if a credible "StormBringer" preamp design emerged from this - great name.   :dbtu:

:tu:

**edit**
Oh, also I would suggest starting a new thread if this StormBringer continues in development.

:dbtu:

[stormbringer]

Yeah, i figured when i get started in a proper way that i should create a development thread, where i write some kind of "diary"-like thing about the development process.

I have been reading alot, experimenting and learning. and the fetzer valve over at runoffgroove had lots of equations (as i'm working as a programmer, i love equations ), and analyzing that got me started writing a piece of software to aid me in the biasing quest.

Right now the app works like this:
you enter min Idss and vgs_off, and max Idss/vgs_off found in the datasheet, the application then plots a chart with the selected amount of curve points for minimum, maximum and estimated intermediate values using shockley's equation.

And then it also calculates optimal bias resistor values and voltages according to the formulas on that page. Now, i'm not entirely sure how accurate those equations are, but it's really helping me understand. I'm also making a small USB HID device using a PIC18F2550 (which i have previously used to make a gameport/midi to USB-adapter). The device will be controlled from the software and read actual Idss/Vp from the socketed Jfet and draw the graph + calculate the estimated numbers.

So just insert the fet, press a button, and bam - charts and numbers based on actual values appear.

Now this might seem like quite an overkill for doing something as simple as a gain stage, and it probably is. But i love exploring ideas and experimenting. and with a little luck, it might actually help my designs in the future. if not.. Well.i had lots of fun making it.

But i will most likely start a new thread shortly, with that development diary from a novice's perspective.

Thanx again for your help and interest!

[Roly]

 :dbtu:

The three phases of a new idea;

1. "That will never work."

2. "It's a good idea, but impractical/expensive/unpopular."

3. "I always thought so myself."

In my early 20's (in the days of two step wet photocopies) I invented the office fax machine; called it the Remote Photocopier.  My boss told me I was nuts, nobody needed such a thing.  {That's how I lost my first million bucks}.

[joecool85]

:dbtu:

The three phases of a new idea;

1. "That will never work."

2. "It's a good idea, but impractical/expensive/unpopular."

3. "I always thought so myself."

In my early 20's (in the days of two step wet photocopies) I invented the office fax machine; called it the Remote Photocopier.  My boss told me I was nuts, nobody needed such a thing.  {That's how I lost my first million bucks}.

Did you mean 60's?  For you to be old enough to do it in the 20's would mean you would have to be like...110.  How old are you?

[Roly]

:dbtu:

The three phases of a new idea;

1. "That will never work."

2. "It's a good idea, but impractical/expensive/unpopular."

3. "I always thought so myself."

In my early 20's (in the days of two step wet photocopies) I invented the office fax machine; called it the Remote Photocopier.  My boss told me I was nuts, nobody needed such a thing.  {That's how I lost my first million bucks}.

Did you mean 60's?  For you to be old enough to do it in the 20's would mean you would have to be like...110.  How old are you?

Egad - MY early 20's, not THE early 20's - I'm 63 and nominally retired (with a pile of organs and valve amps awaiting my attention  ).  Yes, THE 60's before office fax machines and dry copiers arrived, and modems ran at 300 Baud.  {TTL was still replacing DTL, the microprocessor had yet to be created, and my first computer build with a whole kilobyte of memory was still in the future.  These days I still pull out a mini SD or flash drive and just look at it in wonder; picked up some 4Gb flash drives at a local shop for two dollars each!!!}