Common Source Amplifier Questions

[sa230e]

Hi,

I asked about heavily overdriving a BJT a few days ago and now I have a (probably dumb) question about heavily overdriving a JFET. So I'm simulating this JFET common source amplifier with a voltage gain of 10 (picture below) and I'm driving pretty hard (10v p-p). The green signal is the input waveform, the blue waveform is taken on the gate of the JFET (after the 500k resistor) and the red waveform is the output, taken at the 1 MEG load resistor.



You can see the signal is clipped on the output and partly on the INPUT as well because the voltage forward biases the JFET junction when the signal swings upward (unless I'm mistaken).

Is this a problem? I've been wanting to design a little practice amp for myself and I've been trying to decide whether to use JFETs or BJTs for the preamp and I can't decide whether or not I have a problem with this or not. I basically lose almost the whole top half of the signal. And I've read asymmetrical clipping causes a strong 2nd harmonic and I don't know if that's what I want.

I was hoping to design the preamp by stringing together a bunch of gain stages like this one (common source or common emitter) - like a tube preamp. I realize I could just use an op-amp and clipping diodes and not have to concern myself with this stuff but (for me, anyway) that's no fun. I know it sounds dumb but I like to do things the "old fashioned" way, I guess.

[phatt]

Hi sa230e, keep chipping away it will all eventually fall into place. :tu:
I will assume here you are chasing that elusive valve sound in a simple preamp. 8|
I'll leave those better qualified folks here to explain the technical details of the Q's you ask, meantime here is what I've got sitting on a testboard right now if you want a cheap short cut.

Having spent many years now learning the hidden secrets of guitar amplification this is as
close as I can get without breaking the bank.

I've already got a hybrid preamp which of course uses an AX7 at high voltage and in an A/B
comparison this comes close to delivering that elusive triode transfer character. At least at real world live gig levels few ears would be able to tell the difference.
triodes have this magic ability to clip on big signals in a way that is hard to reproduce with ss gear. If all you want is distortion then it's likely easier to go with back to back diodes and tweak tone shapes to desire but it is all those in between sounds that are hard to nail and triodes do that well.

I've only got a few jfets on hand to work with but I have no doubt other devices will work also. Never be afraid to experiment and BTW you will never perfect a circuit with a mouse.

Yes sims are a huge boost to design for the hobbyist but it's unlikely to pickup every potential flaw so a few test circuits will be needed before it's perfected. You still need to know how to avoid ground loops and other pitfalls the unwary novice.
This circuit is not built yet as I'm still working on it so a few more weeks of tweaking before any boards are made.

If you do not use my tone circuit then you will likely need a buffer stage on the output of
Q2.  This tone circuit has plenty of top end so remove C8 1n8 if you do not have enough
treble or try a lower value to taste.

The final outcome will still very much depend on what comes after this preamp so speakers
play a big part as does the guitar pickups i,e, SC or HB. Over-wound and like pu's will of
course make it breakup earlier. Personally I'm not a big fan of hi wound pu's, I actually have trouble finding lower output pu's nowadays.
So if say you plug my circuit into the front of most modern guitar amps with hot guitars it will likely come out very harsh and likely brittle so it's up to the builder to make any adjustments needed. Ideally you maybe better served with a flat response poweramp stage.

OK that's my budget limited contribution to the cause but if you wish to take it to extreme
then search for the work of member KMG, all SS fet amplifiers driving real output transformers all at high voltage. Though I'm not qualified in this field I've been burning up electrons long enough to read between the lines of new ideas and pickup on what is worth pursuing and what maybe a waste of time. I've built my fair share of landfill :lmao:

KMG,, Very clever stuff well worth a look. :dbtu:
have fun, Phil.

[sa230e]

Thanks for the reply, Phatt. I guess I should have mentioned that I am a real newcomer and that circuit is (for now) over my head.

I'm not really trying to recreate the valve sound with transistors per se (I'm not a competent enough electrical engineer for that yet) I just wanted to design a simple SS preamp using circuit topologies analogous to the common cathode gain stages used in tube amps (common source, common emitter) as opposed to op-amps and clipping diodes. I was trying to copy the topology really, not the sound (although a nice smooth, bluesy tone was what I was shooting for but I want to be able to have enough gain for soloing too) - kind of like the people who make FET equivalents of tube circuits. Now that I think of it, maybe I should simulate some of those designs and see if they do the same thing.

Basically I wanted to follow these principles:

• Common source / common emitter gain stages
• Clipping against the supply rails instead of diodes
• No negative feedback

When I started reading about electronics I wanted to design tube amps right away and I read a lot of material on tubes. At first I thought I could string a whole bunch of these stages together like a tube amp and call it a day but it didn't seem to work out that way. BJTs saturate and JFETs forward bias and act like closed switches if you drive them hard enough which complicates things somewhat.

Basically what I wanted to know was how fact the gate forward biasing when the input signal goes high enough (and thus clips the input signal) affects your tone. Do you get that strong second harmonic because it clips the input signal or does it not matter since the output of the stage swings from rail to rail?

Would this effect make it unfeasible to use JFETs in such a design?

[Roly]

Lizzen, you need a bit of wood, a large plain kitchen cutting board or "breadboard" is ideal.

Next you need an oddment of ali L-section that you can mount sockets and pots on.

{phatt you had a pic of your doozy, please link in here to illustrate.}

Take a small bit of Vero/stripboard/dabboard/matrixboard, say a couple of inches square with some mounting screw holes, and build the first stage on it.  Build the tonestack you want on another and wire to the pots.  Connect and win.  Continue thusly.

One of the very early things you will need is a reliable (and foolproof) power supply for your new breadboard.  Mounting?  Ah, small wood screws or self tappers out of the "assorted mech" bin.

What gets developed here later gets repackaged in a proper chassis ... if you like.

The breadboard brings some order to the process of getting individual stages to behave, is highly traditional and tried and tested long before those 'orrible plastic prototyping decks.

A valve testbed;


Different one;


350V @ 200mA + heaters in a computer PSU box;



To try and address your questions...

You are never going to get 10V up the input, not even from a domestic CD player, so first stage overload isn't really an issue, and if it is we can easily make it become not so.

What is interesting about FET stages is that they have an intrinsic transfer curve that is similar to a triode.  They both have a log-law, but they have a different index or power term.  From memory it's x-squared for a FET and x-to-the-four-thirds power (but I'll stand corrected on the values of the curve.  Apparently there is a spot where a FET can be biased to produce the same curve as a triode.)

The point is that both triodes and FETs tend to naturally squash one half of the signal wave and stretch the other, what Hi-Fi-ists call harmonic distortion, due to the characteristic being curved.


Arbitrary triode, ECC81/12AT7


Arbitrary small signal FET, e.g. MPF102

The idea of going gain>tonestack>gain>volume>main amp is to distribute gain and loss so that the desired stages go into overload first.

In a clean straight amp with distributed gain the first thing that will go into overload/clipping will be the output stage - desirable in valve amps, undesirable in s.s. amps.

In a valve amp like the Lamington you have a gain control near the front end and a master volume control at the power amp input.  This allows the setting of the levels to give either power amp clipping first, preamp clipping first, or both occurring together.

Anti-parallel diodes across the signal path in the preamp are almost exclusively found in s.s. amps and never valve amps. This is a clue that controlled signal clipping in a s.s. amp is better than uncontrolled OP stage limiting.  With such a clipper you can play around with LED colours which have different voltages for each diode.

If you don't know how much effect you are going to need, make it adjustable - include a gain control, or three if that's what you need.

A possible reason for the subjective difference is that when a s.s. output stage gets to clipping it effectively shorts (and therefore electrodynamically damps) the speaker for the duration of the clip, while in a valve amplifier because di/dt in the output transformer goes to zero during clip the speaker is effectively open circuited and can do as it likes colouring the sound.

The Gate of a J-FET is simply a diode to the bar, so it's mainly just diode clipping.  In some designs a real diode is placed nearby in parallel with this G-S diode to conduct instead.



postscript:

Take a look at this;



This idea started out in a thread here discussing preamp design.

Now we need a gain cell, and the ideal amp will have an infinite input impedance (a FET close enough) and a zero output impedance (or an emitter follower, close enough), and an important factor was to try and preserve the triode-like characteristic curve distortion of the FET.

So with these two devices, one handling input, the other the output, we should have something a bit closer to an ideal gain element than either on its own.

We have gain control on the first gain stage, but the cell has otherwise been simply repeated for the make-up amp following whatever lossy passive tonestack you wish to use, and the emitter-follower now provides low-Z drive to the main amp, line, whatever.

{there is an improved power supply on the later version}

[sa230e]

Roly, I appreciate the effort you must have gone through with that last post, but like I said, I'm new. I didn't understand most of it, unfortunately.

But point taken on the breadboard. It's a good idea. I think I'll put the pre-amp project aside for a while and design a test bed. I've got a transformer, plenty of caps, rectifier diodes and an LM317 voltage regulator so I could easily whip up a passable power supply. I also have a small speaker and an LM386 chip amp so that should make it easy to swap in gain stages and test them out.

Thanks.

EDIT: I know I won't get 10v p-p on the first stage. It was the second or third stage I was trying to simulate and it was just easier to set the signal generator to 10v than to build two or three stages before it.

[Enzo]

If you want to make a circuit that sorta resembles tube format, really, look at the Peavey circuits in their Transtube series.  I think I mentioned it in your other thread.   The preamp part is made with transistors.  They are placed in pairs to give more tube-like gain.  The circuit has all your old favorite things, like "cathode" bypass caps, and "plate to grid" caps.

In the example below, the gain channel is an extra couple stages switched into the middle of the channel.   Same idea as the tube Classic 30.

[Roly]

No matter, I write for the "sandbaggers" as well, those who read now, or later, who don't comment.  Grab what understanding you can, maybe come back later and re-read and get more.


A bit more headwashing.  Let's take a worked example to illustrate "the distribution of gain along the signal chain", and let's assume that you are really whanging the guitar and getting 1Vpk signal out of it (for a nice neat round figure).


A typical first stage is a 12AX7 triode which will have a best-case gain of around x50 (or think FET or bipolar stage if you would rather, makes little difference, FET's ~= triodes anyway);

{Engineers are really lazy.  Gains and losses can be calculated by simply multiplying along the chain, but being lazy, if you convert these voltage gains to decibels you only need to add and subtract.   Accounting come to electronics!  But they are a handy way to get a handle on ratios, and very large or small numbers.}

{dBv = 20 * log10(Av)   ; where Av is voltage gain, x50 in this case.

20 * log10(50)= 33.9794001dB or +34dB}

1Vpk x50 = 50Vpk on the anode.

If we have a solid state first stage with a 24v supply we are obviously going to be running into clipping against the rails, but a valve input stage will have a supply of around 300V and therefore have an available swing at the anode of getting on for +/-150V, so no clipping there.

Now if we run this directly into a similar stage it is going to get "crushed" witless anyway;

1V x50 = 50Vpk, x50 again = 2500Vpk, obviously impossible, so we will have square waves coming out the second gain stage.

Now introduce a tonestack between the stages.  Typically a tonestack will have an insertion loss of around 20dB, a bit more than its control range, so now we have;

{20dB = 20 * log10(Av)

20/20 = 1 = log10(Av)

Av = 10^1.0 = x10

0dB +34dB - 20dB = 14dB overall to this point;}

1Vpk in, x50 = 50Vpk x1/10 = 5Vpk x50 = 250Vpk

Now this is driven hard with Gain and Master Volume set flat out, so even with valves this is going to clip a bit, but not nearly as badly as if we had two gain stages in cascade followed by the tonestack.

(Purely) by convention we expect a power amplifier to produce full output with an input of around "0dBm" or around 1Vpk, and this is also typically where you will find the Master Volume in the signal chain.

Calculating the gain after here is a lot easier with a s.s. amp than with a valve amp, but we can get a general idea from the two resistors in the Negative Feedback Network, and somewhere around +30dB is typical.

{say 50 watts into 8 ohms

P = E²/R

transpose, substitute;

E = root(PR)

(50 * 8 )^0.5 = 20V_RMS, * root(2) = 20 * 1.414 = 28.28Vpk

So for 1Vpk in this must have a voltage gain of x28

dB = 20 * log10(28)
20 * log10(28) = +29dB}

So our maximum overall gain from input to output is +14 +29 = +43dB

1Vpk x50 x0.1 x28 = x140

Clearly the amp is going to be overdriven by a guitar, but then that's what we expect if we have the front end Gain and mid-amp Master Volume both on 10.  But we don't often do that (well, I don't often do that, anyway, but then I don't play Anything-Metal).

I had a visit from a friend this week with his new second-hand Fender Blues Deluxe which he has bought for pub gigging as a harp player.  I was a bit dubious that the tonestack would match well with a mike, but it turned out to be very effective with quite tame acoustic feedback.  So I plugged in my guitar as well, set both the Gain and Master for 3, and we Bluesed out for a while.

Now taking our hypothetical amp above with these settings and a more realistic 100mVPk in (Fender specify only 4mV in for testing);

0.1 * 50 * 0.3 * 0.1 * 50 * 0.3 = 2.25Vpk at the Master Volume wiper.  (the power amp in the Blues Deluxe only has a voltage gain of about x7 (280mV in -> 2V out across 8 ohms)

Take a look at the signal distribution at the test points along the signal chain in the Blues Deluxe;

http://support.fender.com/schematics/guitar_amplifiers/Blues_Deluxe_schematic.pdf

In the end it comes down to "which stage do you want to clip first?", and that's a matter of taste, what sounds right to your ears given the context of the song and venue.  "Tonehacking" your amp is a very worthy and natural thing for any musician to do to their "instrument".  You don't need a PhD is Advanced Nurgling, just good safety practices, and the willingness to have a go, to "chance your arm", to learn stuff as you go and make (many) mistakes.

My favorite on this topic (and she plays bass too!);

http://youtu.be/xhQ7d3BK3KQ

As Enzo points out, Peavey have liked gain cells with transistor doublets or triplets that use both AC and DC Negative FeedBack (NFB).  My FET/BJT gain cell in the preamp above was certainly inspired to some extent by seeing Peavey's like this.