mosfet treble booster?

[MartyMart]

Small analogy: It has been said that the human genome is something like 98% identical to that of the Chimpansee. It's the last 2% that makes the difference...

True, but it's actually closer to 99%  !!! ( so my Doctor buddy once told me )

My "Tornado" circuit is 25% BSIAB .... so who am I to "grass"  !!

oh well .... onwards  !

[Ronsonic]

Small analogy: It has been said that the human genome is something like 98% identical to that of the Chimpansee. It's the last 2% that makes the difference...

Well, 70% of that is identical to yeast. A lot of basic common elements are needed.

Ron

[WGTP]

I usually stay away from these discussions, but sometime it is a little frustrating to think there are some interesting and significant circuits that are forbidden knowlege because a certain builder uses them.  The Hotcake is another example.  I can see both sides of the issue, but for me only building on a breadboard, it would be nice to know some of that stuff.  Heck, I don't even build the stuff into a box for myself.   

[markm]

I usually stay away from these discussions, but sometime it is a little frustrating to think there are some interesting and significant circuits that are forbidden knowlege because a certain builder uses them.  The Hotcake is another example.  I can see both sides of the issue, but for me only building on a breadboard, it would be nice to know some of that stuff.  Heck, I don't even build the stuff into a box for myself.  

I agree.
Doubtless somewhere is an electronics/transistor manual has a schem that pretty much looks almost like this circuit or, at the very least, resembles it. Check out some of the Russian DIY sites and see what they were doing in the '70s and '80s......it's quite interesting.

[petemoore]

  Just getting it close to something biased'd be nice for now...
 I'm not all good with these mos bias things, yupp, I'll go read 'n stuff,
 anyone who would like to observe that I'm getting
8.2v supply from good 'ol battery
D:4.12v
G:3.52v
S:1.82v
 I have what's shown there, minus the protection diode, I used 10k pots for the drain and source resistors, Drain adjusts to just under 1/2v up to near V+ smoothly, and the source can be pulled to ground, or go...too high
 The gate is like hooked on the Drains coattails', I wondered if this wouldn't be the case starting off, with the two = value divider resistors pulling the gate bias toward 1/2v.
 ...about mosfets, do you thing I should start lowering the base to ground resistor, cutting the base to V+ resistor, should the schematic work.
 Any suggestions as I go through and retest the circuit against the unverified schematic and digup on the mosfet reads at GEO and AMZ...for like biasing this mosfet most easily from where it's sitting?
 input cap gets a gate, a 10m to ground, a 10m to the drain [5k from  V+]
 Drain has, 10kpot to V+, output cap, and ^10m
 Source has 10kpot to ground.
 Could the diode shown be an LED? I think I have a 5.1v zener down in the 'long since I visited dungeon' area...lol, lotsa boxes.
 Perhaps if I add that in there things will change...I don't * think I'm popping Mosfets with each try...probably because I've used 2n7000's many times without protection. Course now again, typing, maybe I'm missing something that biases the gate by leaving the diode out.
  I'm learning the hard way about mosfets again, maybe the trick is to go back and readup on them againagin.
  I've got a bunch of 'em, always I run into the doh, how do I bias a mosfet thing...bipolars and jfets I can always whip into some resemblance of bias.

[R.G.]

Oddly enough, there is a whole article on biasing your own MOSFET boosters at GEO. 

Go look at http://geofex.com/Article_Folders/mosboost/mosboost.htm for how to bias a MOSFET.

[petemoore]

  Goes right over my head again.
  I can't follow the math, so when I don't follow where it says 'wrong'...then try reading below to pick up where I can't refer to...etc.
  What'd be cool is to actually show voltages, and resistances, on any particular way of hooking up a mosfet, perhaps even diddle a resistor and show the voltage difference, yes I could and should have done that, or perhaps increased my comprehension of texted math problems, which I'm trying to do, sort of...
  I'd ask questions but that would be most of the bottom half, the meat and potatoes half of the article.
  I'll go back and read through terms, try to grasp enough to form a decent question.
  Sometimes another approach is needed for a dolt like me, I'm just putting something on the board and using trimpots to cover a range on S/D biasing, as you can tell I'm trying to figure out what I'm not figuring out.
  The article is just a top heavy reading/comprehension load for me. I can't hack the math, and don't really understand the terms and equations.
  I don't know what I want even, what are operational bias voltages for DSG, or should I be figuring it with current instead?
  I don't think
D 4.12
G  3.52
S  1.82
  is going to produce any kind of near linear amplification, the gate is too high/close to the drain, source is also using up 1.82v [at least] of the supply. I need more than about .5v difference between gate and drain.
  If I could bias a mosfet, I'd post voltages, resistances and performances, we're talking about 8 components, either it works or it don't, mine don't, perhaps it's not wired as shown, I'm trying to determine this too.
  I'll go through again with the dogma/DMM measureing of R's connections, non connections to see if I've got what is shown on the unverified schematic.
  I get the idea it works or should work.
  R values and voltage at pins is what I do understand and have been trying to apply to this circuit.

[petemoore]

When you put a gate-to-source voltage on the gate
  This is about hooking 9v D to S, and adding a voltage to gate,
  so this is the gate in a neg gnd circuit being 1v *higher than the source?
  Any takers on how the voltages I posted look ?
  Ok, maybe I should try to get a handle on this big chunk
  I = Gfs*[Vgss-vt]
  I : current? [should mean impedance being I?]
  =: equal to
  G: Gate
  f:  frequency ?
  s: source
  *: multiplied by
  { : one of these with one of these '}' means work this part of the equation first
  V : Voltage
  G: Gate
  ss: [solid state !ha] or...
  s: source
  - : minus
  Vt: voltage threshold
  }: see above '{'
  yupp, schematic with resistance values and typical or just your voltage measurements, plus resistances would make a great study for me, yupp...I'm lazy, well when it comes to running repeatedly into brick walls, I mean I'd write up more about what that's like but it'd be useless...if I figure something out about mosfets, I'd be more than happy to share actual resistances, voltage measurements, what happens to the voltage at drain when 'this' [say a source or gate resistor value change] resistor goes this way, how to get the gate closer to ground so the drain has some swing room...whether or not to try another transistor if the gate won't move down, whether the protection diode 'needs' to be there, if it alters bias. Typical 'median' values and where to look for bias tweeks.
  I should just know mosfets biased are beyond my comprehension, and increase my comprehension or reduce the number of mosfet circuits I blindly attempt..lol.
  I guess mosfets are best left to those who are on the other side of the mosfet biasing comprehension hurdles. I've learned something from experimenting with and trying: I don't know mosfets. I guess until the math is fully understood, there's no way to demonstrate or explain it beyond that with texts or actual application notes, like resistances/voltages....I see current mentioned alot in the GEO article, and surmize it's my lack of terminology and math comprehension that prevents me from understanding it.
  I should probably de-mosfet other than the Mosfet Booster, and stick with Jfets and bipolars until I can get a handle on the mosfet bias.
  These devices have proved a very different learning curve than bipolar, viewing thousands of bipolar schematics, and understanding them pretty well in a voltage sense...well I can hook up a bipolar ten different ways off the cuff, I haven't a clue how to get a mosfet to even get close.
  All I can do is figure I didn't read where the source is higher than gate, have seen mosfets plunked into FF's so figure they probably work if source is @ low voltage, gate a bit higher, and drain about 1/2v.
  Just cutting and bridging and altering values of resistances...long way around getting to decent bias points.
  I've been doing too much typing and not enough measureing on the circuit board, or perhaps I should hit the brick wall of reading a few more times instead of this blind diddling.

[Gus]

Little history

As far as I can tell no one had a small signal mosfet gain stage like the SHO on the MARKET or WEB before the SHO.

There was the Blackberg press(sp) book designing with VMOS or something like that from the late 70 or early 80's IIRC.  I have the book somewhere at home I will look for it.  I have a few power mosfet gain circuits I never shared from about 95.

Plus there is more going on with the SHO circuit than say Jacks posted design Jacks is a more "standard" design IMO .  Different circuits 

SHO has a different bias, feedback as the gain goes up the input Z should go down due to the   and a gain control that messes with the source R and there for the bias.
  The SHO is cool little circuit.

[R.G.]

OK, let's try this the simple way.

MOSFETs have a channel current that is a multiple of the voltage across the gate/source. For most modern MOSFETs, that's something like 0.1A/V to 1A/V.

There is also a turn-on threshold you have to apply before any current flows. This is the Vthreshold. It's kind of like the forward drop of the base-emitter in an NPN. Until you give it enough voltage to start up, nothing happens. After that, the channel current increases by about 100ma per volt of increase to 1000ma per volt of increase; this difference in current ratio is much like the hfe of a bipolar.

Generally, the voltage->current gain is so big that unless you want pure on/off switching, you'll use a source resistor to provide some feedback. So the current that flows in the channel is always Vsource/Rsource.

Here's the trick: In a MOSFET circuit, the gate will always be just a little more than Vt above the source. In a bipolar transistor, the base will always be 0.6V above the emitter. In a MOSFET, the gate will always be Vt above the source. So much current can flow for little increases above Vt that the source pin rises on its source resistor and takes the extra voltage away, always coming to a balance with the gate-source nearly at Vt. The only real differences from bipolars is that Vt is bigger than the 0.6V of a silicon base-emitter, and that Vt can be different from MOSFET to MOSFET. It's something you look up on the maker's datasheet.

So if you connect a drain to +V, a resistor from source to ground, and raise the gate on a pot wiper between ground and +9, nothing happens on the source at all until Vgate gets up to Vthreshold. Then Vsource follows Vgate up as you turn the pot up, volt for volt. The current that flows is always Vsource/Rsource.

That's a source follower. Now if  you put a resistor in the drain, the resistor in the drain makes the drain voltage drop by ohm's law: Vdrain = +V - Id*Rd. But we know that Id and Isource are the same because the gate is an insulator. So  the voltage at the drain goes down when the voltage at the source goes up.  If you do the same trick of putting a pot on the gate and watching both source and drain voltages, nothing happens until you get the gate to Vt. That is, the source stays at 0V, pulled down by the source resistor, and the drain stays up at +V, pulled up by the drain resistor.  When Vgate gets bigger than Vt, the source follows it up, causing source current of Vsource/Rsource to flow.

That means that the drain voltage comes down as the source goes up. How much? by the ratio of the drain and source resistors. If Rd and Rs are the same, the same current flows through them, and Rd drops as much voltage off V+ as Rs raises off ground. When they meet in the middle, the party's over and the device is saturated.

If Rd is zero, you have a source follower and the drain stays up at V+. If Rd is equal to Rs, you have equal and opposite signals at drain and source. If Rd is bigger than Rs, you have gain. The gain is always Rd/Rs, because the voltage at the source follows Vgate, and that causes a current to flow through the MOSFET of Vs/Rs, and the same current of Vs/Rs flows through Rd, so the voltage at RD is +V - (Vs/Rs)*Rd. For AC signals, this is an AC gain of Rd/Rs because Vs follows the gate, but one VT lower.

So to bias a MOSFET:
1. Figure out what voltage(s) you want the drain and source to sit at for the right amount of signal swing.
2. Figure out what current you want to flow thorugh the drain and source.
3. Calculate Rd and Rs to satisfy 1 and 2.
4. The source must follow the gate. So you want a bias voltage on the gate of the Vs you calculated that you want, plus one times Vt. All done. The gate is a piece of glass twenty volts thick, so it has a voltage on it, and 0.00000000000000000000000000000ma of current through it. You can make the biasing network of resistors be anything you like. The gate does not care. It cares only about the voltage on it.

Did that help?

[MartyMart]

Pete, I'm with you on the math !!
Yup your Drain/Gate and source seem way too close ( to my untrained eye ) for proper amplification/bias

OK, similar setup ( NOT SHO ) and here's my voltages :
BS170
D  6.8v
G  0.84v
S  0.15v

Obsidian type setup using 3M3 from gate to 9v and 1M5 from gate to ground
Drain R is a 5k6 ( depends on the mosfet! ) you have a 10k trim, source R is 1k8
again you have a trim pot, give that a whirl ?
Oh and a 3mm red LED ( normal ) between S/G will protect it fine enough.

MM

[petemoore]

  Yes it did and will.
  The source must follow the gate.
  You probably said exactly this in the article, reading it now brings to my bias issues some 'light bulbs' already, I'll see now if I can light this up.
  I'm going to say thank you again and re-read then re-debug.
  I think the gate is being pulled up by the resistor to V+, I'll try getting the gate voltage down and watch for the source to follow, while trying to observe vt.

[Mark Hammer]

The dumbass way......

When there is an easy path for AC from source to ground, gain for that portion of the audio spectrum goes up.  That is why the gain goes up as the resistance between the 100uf cap and ground goes lower.  The 100uf value is selected to apply that gain boost equally across the entire audio spectrum.  As Processaurus notes, if the cap value is smaller, then that extra gain gets applied to a narrower range of frequencies - i.e., audio content above a certain frequency.  Assuming the usual rule of thumb, chopping the cap value by half moves the corner frequency up an octave.  I'm not sure what the current low-end rolloff is with the stock value.  But say it was 10hz with 100uf.  A 47uf cap would bump it up to just over 22hz or so, a 22uf cap would bump it up to just under 50hz, and a 1uf cap would bump it to 1000hz.

A fairly interesting booster could be achieved with the basic AMZ MosFet structure by doing the following:

• stick a 100uf cap and 1k8 resistor in series
• stick a .22uf cap and 2k pot (variable resistor) in series
• stick those two networks in parallel with each other
• run a 5k pot from their ground-side junction to ground

This would permit you to adjust overall gain with the 5k pot and adjust the balance of full vs treble-boost using 2k pot.  You would not be able to get as much overall boost with a "balanced" gain, but then you could easily cascade two of the AMZ circuits (one modded as per above, and one stock) in one box to ge lots of boost with variable treble boost.

[Ben N]

When there is an easy path for AC from source to ground, gain for that portion of the audio spectrum goes up.  That is why the gain goes up as the resistance between the 100uf cap and ground goes lower.  The 100uf value is selected to apply that gain boost equally across the entire audio spectrum.  As Processaurus notes, if the cap value is smaller, then that extra gain gets applied to a narrower range of frequencies - i.e., audio content above a certain frequency.  Assuming the usual rule of thumb, chopping the cap value by half moves the corner frequency up an octave.  I'm not sure what the current low-end rolloff is with the stock value.  But say it was 10hz with 100uf.  A 47uf cap would bump it up to just over 22hz or so, a 22uf cap would bump it up to just under 50hz, and a 1uf cap would bump it to 1000hz.

That sounds very much like the shelving effect of the cathode cap in the input stage of a tube amp--which is of course analagous to a source cap--I can tell because they look the same n the picture, er, I mean schematic.  But that contour is very different from that of, say, a Rangemaster, with its steady increase in volume proportional to frequency right up the spectrum, no?  Or is a RM just a really high shelf?

Ben

[WGTP]

R.G. when are you going to get around to finishing that?   

[John Blund]

wow. this thread turned out to be quite a bit more informative than I ever hoped. Can't wait till I get hold of the parts I need so I actually can try to make this thing work! 

[markm]

wow. this thread turned out to be quite a bit more informative than I ever hoped. Can't wait till I get hold of the parts I need so I actually can try to make this thing work! 

I'll bet!
I propose that you name your circuit the "MosFet TROUBLE Booster" !! 

[tcobretti]

I propose that you name your circuit the "MosFet TROUBLE Booster" !! 

Very nice.

[petemoore]

  I just haven't worked with Mosfets much and have a terrible memory.
  However, once again I feel RG has provided enough 'angles' to the same thing I think is on the mosfet article [for those who can read] so that even a dolt like me can think I have a much better handle on it. Only after the third re-read did I make out that '/' means divide, as is used to show the gain relationship between drain and source resistor values.
   And as a result of my newfound confidence, I believe there must still be some silly thing wrong with my circuit, I cut the gate to ground 10m, and dropped the gate toward V+ 10m down to 330k, this dropped the gate and source .2v. less than 330k kind of starts defeating the input impedance.
  I tried 2k2 source R, and 4k4 drain R, and got volts something like D4.4v G2.3v S .8v, no boost, maybe I static'd the Q or something.
  Thanks for the info once again RG, I think I'll try putting a little R on the S, 2 - 4 x as big an R on drain [I don't think I really need much gain], the fiddle with the gate and see what I can get.