Sustainer Ideas

[Fresh Fizz]

@psw & col

I want to share with you the schematic of the mosfet compressor I have.

I've found this one on the diystompbox forum some time ago. Could it be that this one is the original and that the ones that zfrittz6 and psw have posted are derivatives? That could possibly be toxic? Maybe zfritt6 only wanted to be helpfull and posted the schematic. But has he actually built that compressor or has he obtained the schematic from somebody who stated that he had actually built it? For instance have seen schematics for starved plate tube amps (with ECC83s) that couldn't possibly work. How do I know? Because I've have been messing around with ECC83s with low plate voltage. That schematic must have been the result of a thought process and not a reflection of something that has been constructed in reality.

Back to the schematic I posted.

In this schematic there is a cap in between LM386 output (pin 5) and the rectifying diode. Which means that the anode side is at a zero volt potential, just like the cathode side.

In zfrittz6' and psw's the cap is missing, anode side is at half-battery voltage (4.5V) times attenuation (depending on 4k7 resistor and potmeter) , and the diode is conducting (?!)

Which means positive voltage on the gate of the mosfet. (Resulting in which resistance measured between drain and source of the mosfet?) Anyway, the cap at the gate can only be charged more when the voltage at pin 5 goes above 4.5 volts.

Now to me it seems that the cap has to be included. And it has to be big compared to the cap at the gate of the Mosfet.

At least I would bet that of the three schematics the one I posted should be the correct version. And if the frittz6 and aussimart are functioning hey should be two completely differen beasts compared to the one I posted. But I HAVEN'T built it, so I'm not sure. But has anybody really built one of the other versions?

Cheers,

Fresh Fizz

[Hank McSpank]

To chime in on this 'Compressor' discussion....

Firstly, Col is right, the MOSFET in the compressor schematic recently posted is *not* being operated as switch but acting as a Variable resistor (sure, you can use a MOSFET as a switch, but in this particular circuit the results would be terrible!).

Personally, I think the compressor being discussed is fairly crude ....all it's really doing is 'limiting' the output of the LM386 . Therefore, the only way to be sure that smaller input signals will 'excite a string' sufficiently is to have a high gain arrangement either at the LM386 (or more likely - feeding into it), which is going to cause you all sorts of problems....think of it along the lines of nailing a car's engine at 10,000RPM with lots of torque readily available at the wheels, but then merely controlling the car's speed with the brake only - not good. What we need is a good cruise control! (AGC).

I guess 5 years is a long time for a thread to run & there'll be repitition along the way as new entrants mention stuff that was posted 274 pages ago! A quick search, reinforces this point, here's a perfectly good circuit posted up by CurtisA some 2.5 years ago....

(CurtisA's original post with circuit explanation here - http://projectguitar.ibforums.com/index.ph...t&p=295401)

That type of circuit is *far* more elegant. Unlike the, the aussiemart compressor - which is really just a signal shunter (ie when the LM386 output gets too high, it starts shunting the input signal down to ground), the type CurtisA posted up actually varies the gain of the amp. It's not a unique idea...there are all manner of similar ones around the net. I've used a similar 'concept' in my own design (though quite different to CurtisA's schematic)

Re the aussiemart compressor - IMHO, for all bar the most basic of sustainer solutions - it's probably a case of move right along folks , there's nothing significant to see here"

WRT my own explorations...well, all my separate sustainer 'building blocks' work fine now (ie preamp, PIC AGC/Threshold cct, power amp, driver) - I now just need to 'bolt them all together' & hone. On paper this is small beer, but PICs aren't too great at giving you 'in circuit' visual information (eg "I wonder what the PICs AtoD'ed input level is at present" etc), so I'm now having to knock up a way of getting the PIC to talk to my PC's COM Port so I can see what AtoD levels are going on inside the PIC (via Hyperterm) while it's running in conjunction with the rest of my circuit. I'd have to say at this point that I salute anyone who's designed an all encompassing sustainer AGC circuit using just discreet analogue components - you're all Electronics Demi Gods!

Last night's testing revealed something new (I think!) - that even with a good AGC circuit (ie wide output range & no visible distortion on the scope), that there's a tiny bit of 'fizz' just at the point where the power amp chip needs sometimes a little quick 'coaxing' to get the output up a little to get the string to vibrate enough. It's almost as if the output stage is groaning when a quick burst is asked of it ....to use the car analogy again, think of when you're doing 30mph in 4th at the foot of a hill & you put your foot down to try & accelerate quickly...for most cars, it doesn't happen - you just get engine groan. Therefore, I'm now pondering if the driver needs a supporting powerchip that can get supply enough current into it very quickly.

Col - I see your schematic uses a transconductance opamp....which is a pretty new area for me (I'm aware they yield a current output directly linked to the voltage input)...care to say why you went the transconductance route & what the win is wrt sustaining devices?

Edited April 15, 2009 by Hank McSpank

[col]

Col - I see your schematic uses a transconductance opamp....which is a pretty new area for me (I'm aware they yield a current output directly linked to the voltage input)...care to say why you went the transconductance route & what the win is wrt sustaining devices?

Yeah, that was my very first attempt at an agc. It worked, but there were many problems so I went back to the drawing board. My second circuit was very different, and much more successful heres a link to one revision of it - to save you the searching.

FWIW, I think that the approach Pete has described of using a switch to shut off the power and turn it on in bursts is worthy of some development energy. I would approach it afresh, with a different circuit topology... I guess building in some varaible hysteresis rather than(or as well as) using a decay time might help a lot. Worth some thought for sure.

It seems like a good way to improve efficiency.

Curtisas circuit is a good design. I tried to get it to work but failed - although I'm not sure how I failed. Must have been some silly mistake or other. Anyhow, that circuit was his reaction to one of those three articles I linked for you a few posts back.

cheers

Col

[Hank McSpank]

FWIW, I think that the approach Pete has described of using a switch to shut off the power and turn it on in bursts is worthy of some development energy. I would approach it afresh, with a different circuit topology... I guess building in some varaible hysteresis rather than(or as well as) using a decay time might help a lot. Worth some thought for sure.

It seems like a good way to improve efficiency.

I'm not sure I'm following the line of thought - do you mean power efficiency?

If so, it has merit - & it's is one of the (offshoot) benefits of using a PIC ...ie you can program it to say for example "when the input signal level drops below X volts, then make the output pin low"...this then feeds the Power Amp chip DC volume control pin - most of these power amp chips with DC Volume controls are designed in such a way that if DC level on its Volume pin goes below say 0.4V, the power amp shuts down (& then draws microamps). This potentially will save a fair bit of juice. (the trade off being of course that the PIC itself saps juice...but I'm hoping it'll pay it's way consumption wise!)

Thanks for your schematic...that's actually a very good circuit, though there's perhaps a bit too much component count going on there! Is this soley your design or is it a tangent off a group effort some few hundred pages ago?! What's your honest appraisal of where you're at with your Sustainer? (in other words, what does your design do well, & what still needs refining)

Edited April 16, 2009 by Hank McSpank

[Donovan]

What we need is a good cruise control! (AGC).

Hello gentlemen.

I've recently been doing some reading about pulse width modulation using 555 timers as a means of more efficiently controlling electrical motors. The duty cycle is modulated, rather than pushing 100% of the time, it pulses power in the form of a square wave at a (for the most part) undetectable frequency.

Just a shot in the dark, but could the analogy of the cruise control be realized as an AGC through the application of PWM to the output stage's power supply (or somewhere else in the signal chain) through a low frequency, say under 20Hz, with the signal amplitude modulating the duty cycle?

Forgive me if this is a daft idea for reasons beyond my knowledge level... just came to me when I read that quote. If the power amp stage would be too difficult, the signal could be sent to ground using PWM, but again, modulated, not a 100% shot at some threshold as has been recently discussed and tossed out. I do realize this is something like the "shunting to ground" that has been talked about already on this page, but was just thinking this would be a more translucent effect, in theory, or at least intuitively within the confines of my flawed psychi.

[Hank McSpank]

What we need is a good cruise control! (AGC).

Hello gentlemen.

I've recently been doing some reading about pulse width modulation using 555 timers as a means of more efficiently controlling electrical motors. The duty cycle is modulated, rather than pushing 100% of the time, it pulses power in the form of a square wave at a (for the most part) undetectable frequency.

Just a shot in the dark, but could the analogy of the cruise control be realized as an AGC through the application of PWM to the output stage's power supply (or somewhere else in the signal chain) through a low frequency, say under 20Hz, with the signal amplitude modulating the duty cycle?

That's pretty much the path I'm taking by using a PIC in my sustainer circuit.

The PIC 'monitors' the sustainer's preamp output level & ultimately adjusts the duty cycle of its own PWM output stream to suit - if you feed this PWM stream into a low pass filter, you end up with a DC level - this DC level is applied to a 'gain control' JFET in the preamp (which adjusts the gain applied to the incoming signal to ensure a constant predefined output into the power amp)

Edited April 16, 2009 by Hank McSpank

[col]

FWIW, I think that the approach Pete has described of using a switch to shut off the power and turn it on in bursts is worthy of some development energy. I would approach it afresh, with a different circuit topology... I guess building in some varaible hysteresis rather than(or as well as) using a decay time might help a lot. Worth some thought for sure.

It seems like a good way to improve efficiency.

I'm not sure I'm following the line of thought - do you mean power efficiency?

yes, power efficiency. if the sustainer can be off for some of the time while it is functioning, then there will be some saving. Not as much as a class-d.

Maybe my instinct is wrong (haven't done the sums), but I would guess that repeated equal bursts of full power then nothing would be more efficient than steady half (output) power when using a class a/b amp?

Unfortunately, there is another issue with the class-P burst mode amp - low level inputs will still be low level, it doesn't equalise the levels as much as a good AGC would. I suppose that you could use lots of gain, but I would imagine that this would give you clipping in the on bursts, and the off bursts might be long enough to make the on bursts noticable... just guessing. Anyway, tfor it to work as I would want, I would want to combine it with a standard AGC to get the low levels up without clipping the high levels, so it becomes kind of redundant, also for it to work well I think some sort of shmitt trigger would be needed to provide hysteresis in order to prevent jittering around the on/off point. this requires even more circuitry.

If so, it has merit - & it's is one of the (offshoot) benefits of using a PIC ...ie you can program it to say for example "when the input signal level drops below X volts, then make the output pin low"...this then feeds the Power Amp chip DC volume control pin

Yes, that is certainly an advantage, you can experiment in software

What I don't fully understand is why you need a pic to provide what can be got from a dual op-amp set up as a precision full wave rectifier ?

Thanks for your schematic...that's actually a very good circuit, though there's perhaps a bit too much component count going on there! Is this soley your design or is it a tangent off a group effort some few hundred pages ago?! What's your honest appraisal of where you're at with your Sustainer? (in other words, what does your design do well, & what still needs refining)

It's my design. Based as I've stated on two of those articles I posted links to. I then used a full wave rectifier with a dc bias (feeds into the second op-amp in the rectifier) to provide a DC control from the input. So I didn't create it from scratch, but I found the sources, pulled them together and wrestled it into shape.

As far as performance, it does most of what it was intended to. Distortion levels are low enough so that there is no fizz, and only a little low frequency distortion heard occasionally on very low notes (compromising response time vs distortion).

The response times from the real circuit were not as good as I had been hoping based on simulations, and I couldn't understand this. I was also not completely happy about the maximum drive - it produces a good clear sustain, but not as lively as is possible.

The harmonic modes work, and there is some variety, but after the initial fun wore of I got a bit bored with them.

My initial intentions were to create something that allowed the natural sound of the guitar through, and I felt that this system was a little sterile. Although a vast improvement on the basic non AGC versions.

Since designing that circuit, I have discovered an embarrassing error in my calculations/measurements that explain (I hope) the not so stunning response times.

I measured the voltage level of the pickup output when sustain was strong at around 150mV. So I set up the circuit to start pulling back at around this point. I've recently discovered that 150mA as measured by my DMM equates to around 220mV as a setting on the virtual sig-gen in the simulator! I messed up some where between peak-to-peak and RMS.

Aside from that, recently a better understanding of the more subtle details of how it all works has allowed me to make a few little changes that have improved the performance. I've also decided to go a completely different route with the harmonic modes (some experiment that's not been tested yet)

You're right about the component count, but every time I've tried to reduce it, it's had a really drastic effect on the functionality.

In truth, its pretty honed down for what it does (apart from the harmonic modes, which are compact for the number of modes, but don't all deliver value for money IMO).

Input buffer => 1 op-amp

virtual ground buffer => 1 op-amp

Harmonic modes => 1 op-amp

Precision full wave rectifier with built in DC bias. => 2 op-amps and some extras

Jfet based VCA => 1 op-amp (and lots of supporting circuitry)

Power stage.

Thats all there is to it.

It can be compacted reasonably onto veroboard

most of the samples here were recorded using it (combined with a broken amp sim - thats where the occasional clicks and pops come from)

(the clip at the top is nothing to do with sustainers, the one at the bottom was made using the LM13700 based circuit, the rest using the jfet based circuit)

I'm back at work now though, so much less time to play with this, I'll get the new version built and tested asap an post the results.

cheers

Col

Edited April 16, 2009 by col

[Hank McSpank]

What I don't fully understand is why you need a pic to provide what can be got from a dual op-amp set up as a precision full wave rectifier ?

Control & flexibility. A simple full wave rectifier cct, will act upon the level that it's set to - but a PIC can have some supporting 'ifs/logic' attached (for example - the PIC can be programmed , if the sustainer circuit input level is too small, and yet the JFET AGC gain is at its max, then increase the poweramp DC volume level pin up momentarily", etc - that'd make large circuit going discreet). Think of an ADSR VCA setup on a synth...the PIC allows similar degree of control. (whereas the discreet method only offers the A & R bit of an ADSR!)

Remember, I set out on this journey with six drivers in mind (fed by hex pickups). I'd envisaged different operating modes. For example a 'solo mode' - the idea being here that when a player is playing a solo, he'll bend the strings a lot - not a problem for a single six string sustainer, but for a hex sustainer....as he bends out of the single coil range, then he'll have some 'fade away' going on!

I'd anticpated that in solo mode, the PIC could get funky with it's switching & feed each string's input to the string driver/cct immediately above it ...for example, if a top E string note is coming in ....send it to the B string driver as well. (therefore as the E string bends out of reach of its driver, it comes into reach on the next driver up) I'm less focussed on the Hex route now, but the PIC still allows amazing flexibility. For example, PICS can sink 20ma on each of their output pins, so I could elements of the overall circuitry on/off to suit.

Then there's granularity of a digital AGC - it allows 1024 'levels' to act upon - it'd be pretty darned hard to have a discreet circuit accurately act on that level of detail (eg that works out at '0.088mV' when on a 9V supply ...or if on a 5V supply as mine is, then discreet would have to handle granularity down to 0.0045mV). I doubt I'll need that level of detail though - but it's on tap.

Also, as you well know, each & every JFET has it's own 'characteristics', so I can tweak the PWM to ensure it's working in the centre of it's range *very* quickly & easily (none of this having to match discreet components to suit each JFET). Another possibility , -once the data is 'digital', it allows all sorts of interfacing possibilities with sexy bespoke ICs. eg How about an 8 channel analog volume control (hex coils + EMI blocker coil) - controlled by digital input http://www.cirrus.com/en/pubs/proDatasheet/CS3308_F1.pdf ).

By going digital early in the cct, there ought to be less problems with level drift too.

Finally, not forgetting of course, a PIC allows plenty of 'LED bling-age!'

I've just spent the guts of two whole nights trying to get my standalone PIC to tell me what its variable contents are while it is running my little AGC program. (this is one helluva steep learning curve!). I've been trying to achieve this by having the PIC send such data to a PC so I can see this info onscreen - at last, I've just got it working!

For ease/speed I've previously been working on the AGC solely on the PICKIT 2 development board...but this board comes with a little pot on it that I can vary DC levels into a PIC input pin (essentially simulating the rectified input signal of a sustainer). I'd have to say it's very cool seeing what the incoming sampled DC levels are in real time onscreen as I twiddle the pot (& then seeing the DC level alter on an output pin to suit). This is a major hurdle now behind me & allows me at least now to start thinking about bolting my separate building 'blocks' together. (my driver development has been parked, pending the arrival of an inductance meter from those nice fells in China - to many whacky unknowns/assumptions without one...I need hard cold data!)

Edited April 17, 2009 by Hank McSpank

[Donovan]

What we need is a good cruise control! (AGC).

Hello gentlemen.

I've recently been doing some reading about pulse width modulation using 555 timers as a means of more efficiently controlling electrical motors. The duty cycle is modulated, rather than pushing 100% of the time, it pulses power in the form of a square wave at a (for the most part) undetectable frequency.

Just a shot in the dark, but could the analogy of the cruise control be realized as an AGC through the application of PWM to the output stage's power supply (or somewhere else in the signal chain) through a low frequency, say under 20Hz, with the signal amplitude modulating the duty cycle?

That's pretty much the path I'm taking by using a PIC in my sustainer circuit.

The PIC 'monitors' the sustainer's preamp output level & ultimately adjusts the duty cycle of its own PWM output stream to suit - if you feed this PWM stream into a low pass filter, you end up with a DC level - this DC level is applied to a 'gain control' JFET in the preamp (which adjusts the gain applied to the incoming signal to ensure a constant predefined output into the power amp)

Thank you for that excellent explanation. I feel like I just learned something important and I'm glad that what I suggested had some merit, even if it's not an original idea!

That is brilliant, the running it through a low pass filter portion. Am I correct in assuing this gives a nice smoothing effect, so the AGC doesn't act as choppy? Is this post-filtering a common technique and if so, what is it called or what other applications might I find it mentioned?

[Hank McSpank]

Thank you for that excellent explanation. I feel like I just learned something important and I'm glad that what I suggested had some merit, even if it's not an original idea!

That is brilliant, the running it through a low pass filter portion. Am I correct in assuing this gives a nice smoothing effect, so the AGC doesn't act as choppy? Is this post-filtering a common technique and if so, what is it called or what other applications might I find it mentioned?

Yes, it's common. I mentioned a couple of posts back that using the term 'Low Pass Filter' sounds grand - but all it is, is one resistor & one capacitor(!) connected to the PWM output stream.

Yes, a PWM stream would be *way* to choppy for an AGC control signal, so the LPF's purpose is to 'average out' all the PWM's 'ons/offs'. If there are more 'ons' than 'offs' (ie changing the PWM's duty cycle to have it higher for longer than it's low), the capacitor charges up more & the DC level raises - and vice versa. For my PIC, the LPF is simply a 1k resistor in series & a cap in parallel with the PIC output (at this stage, I've used 4.7uf cap - but bigger/smaller values can be used depending on how quickly you want this 'averaged' DC level to impact your AGC control)

Hank.

Edited April 17, 2009 by Hank McSpank

[col]

*Lots of ideas about using a PIC for AGC*

So, rather than using the pic solely for AGC, can you:

A/D convert the raw input signal

apply software AGC (plus harmonic modes etc)

use the output PWM as part of a poor mans class-d amp ?

So basically the input goes (through a buffer if required) straight into the PIC. Then from the PIC output, straight into a mosfet push-pull switching output stage followed by driver plus any filtering required?

I know the devil will be in the detail, but is this feasible?

If so, it could mean an efficient, compact driver circuit.

How much current does the PIC use ?

what frequency does the PWM run at ?

what are the part numbers for the actual devices you are using, so I can check out the datasheets?

As an aside, processing the main signal with the pic rather than just the DC control voltage should make experimenting with phase for harmonic modes much easier. It really depends on how much grunt the PIC has though.

We can work at a relatively low frequency, and shouldn't have to worry too much about aliasing distortion, but the more processing required, the more powerfull the chip has to be.

With some of the more high tech PICs, you have multiple ins and outs. You can also get ICs with multiple matched pairs of Mosfets that can handle ~300mA which would be plenty in a hex system.

cheers

Col

Edited April 18, 2009 by col

[col]

@psw & col

I want to share with you the schematic of the mosfet compressor I have.

I've found this one on the diystompbox forum some time ago. Could it be that this one is the original and that the ones that zfrittz6 and psw have posted are derivatives?

Yes you are absolutely correct. The version Pete posted is one of the successors to that one, the designer Brett was also involved in the Aussiemart thread, but changed the name of the circuit after he included ideas posted by others into the design.

I'm not sure about the one zfrittz posted, but my guess is it's taken from one of those and re-drawn, he may have lost the original pic. I would imagine that he wasn't trying to 'steal' or 'take credit', rather he was just posting these things as a way to bring them back into the discussion for those that missed them when they firs showed up.

You might be right about the output cap, I don't remember the details, however, I still don't think this circuit could provide the kind of low distortion level we're after for this project - there will be a lot of asymetry. Probably good fun as a stomp box though.

cheers

Col

Edited April 17, 2009 by col

[Hank McSpank]

So, rather than using the pic solely for AGC, can you:

A/D convert the raw input signal

apply software AGC (plus harmonic modes etc)

use the output PWM as part of a poor mans class-d amp ?

Unfortunately not - for starters the PIC doesn't have enough grunt, but more significantly, it has AtoD ability but doesn't have the reciprocal DtoA! (some folks get around this by faking an 'analogue output' with clever look up tables to replicate basic waveforms such as sine waves etc). So no hope of what you're proposing. I've not looked at the most Powerful PICs...they may do DtoA, but I've enough on my plate with this one!

How much current does the PIC use ?

what frequency does the PWM run at ?

what are the part numbers for the actual devices you are using, so I can check out the datasheets?

Re the current a PIC draws - I'm not totally sure the datasheet is vague (& as of yet, I've not measured it) - I guess being a processor with lots of IO, it'll be down to what load is asked of it. It's not a high draw though...I reckon something in the order of 20-30mA (as an aside, I had a 'doh' moment the other night when my bench PSU was showing that my sustainer breadboard 'mash up' was pulling 250mA . After after disconnecting the PIC & other stuff in my sustainer...I realised my coil winder circuit which shares the same breadboard was still attached to the rail - phew!!

Re the PWM frequency...not sure either (the reason I'm not sure, is because to all intents & purposes for what I'm doing - generating a DC level - it doesn't matter.

The PIC I'm using is a 16F690 - it won't be my final choice (there are many lower spec'ed versions which will be more suitable for the sustainer). The reason I'm using this variant is becuase that's the one Microchip supply with their excellent PICKIT2 starter kit (I can heartily recommend it - http://www.rapidonline.com/Electronic-Comp...source=googleps )

Edited April 17, 2009 by Hank McSpank

[zfrittz6]

1)Un amplificador inversor

Es un circuito cuya ganancia está dada por R15 y R13, y su impedancia de entrada está dada por el resistor R13 de 47K. Ésta es la impedancia de entrada del operacional. El capacitor C6 es un acoplamiento en alterna desde la entrada al AO (amplificador operacional). Éste sirve además para aislar la corriente continua de la señal de Baja Frecuencia propiamente dicha (señal alterna). El preset P1 de 100K, es un preset de nivel de señal, por medio de este ajuste se logra el nivel de entrada ideal, dependiendo de la salida anterior.

Sobre la salida del primer operacional encontramos al capacitor de paso o acoplamiento C9, entre la etapa de entrada y el circuito atenuador.

2) Un circuito atenuador electrónico con ganancia unitaria.

Consta de un circuito de ganancia unitaria ya que R4/R3 = 1 pero no necesitamos ganancia en un circuito cuya finalidad es producir una atenuación alrededor de 1 kHz (tomada como frecuencia central). El cálculo de este capacitor se realiza mediante la fórmula:

Fo = 1 / 2 x Pi x R x C

En donde Fo es la frecuencia de 1000 Hz, R es el resistor de 47K en paralelo y C es la capacidad a calcular. Entonces, realizando el despeje de la fórmula, obtenemos:

C [F]= 1 / 6,28 x 47000 ohms x 1.000 Hz

multiplicando el resultado por 10 elevado a 12 obtenemos el valor en pF ,siendo el valor normalizado: 330 pF.

Sobre la salida del circuito encontramos a C2, que es el acoplamiento entre dicha salida y el potenciómetro electrónico; mientras que C4 es el acoplamiento entre la salida del circuito y la entrada del amplificador final.

3) Un “potenciómetro electrónico”.

Este sistema consiste en un transistor de efecto de campo (2n5459, Q2), acoplado directamente sobre una salida inversora.

El FET 2N5459, toma de la entrada, mediante el circuito recortador a diodo, la señal de Baja Frecuencia, que produce variaciones en el gatillo (gate) del mismo.

Estas variaciones salen por el drenaje (drain), pero con su fase invertida, ya que está polarizado en fuente comun (souge) o fuente. Por lo tanto, en la salida es necesario colocar un inversor, realizado en este caso con un BC548 (Q1) polarizado en

emisor a masa, a través de D1, D3 y C7, donde D1 y D3 son la polarización en continua y C7 el desacople en alterna; de este modo la señal de alterna ve el emisor a masa.

Los resistores R6 y R7 son las polarizaciones en continua del drenaje y del colector respectivamente. Ahora, mediante Q1 obtenemos una tensión variable entre el punto de union de C2-C4 . Haciendo variar la RO (resistencia interna) de la juntura CE de Q1, mediante la señal aplicada a su entrada; tal cual como si entre el punto de union de C2-C4 y masa, colocáramos un potenciómetro y lo variáramos en forma constante, frente a las variaciones de señal, esto es lo que hace el circuito constantemente frente a variaciones en la señal de audio.

En el circuito de gatillo del FET encontramos, como dijimos, un circuito recortador que “dispara”el gatillo del FET con las variaciones de la señal de Baja Frecuencia. Este circuito está formado por la polarización del gatillo R5, el diodo D2, y el capacitor C5.

C5. R5 representa la ZI del FET, o sea, la impedancia que ve la señal BF, ya que el Fet, de por sí, tiene una impedancia de entrada (ZI) muy elevada.

Saludos

[col]

Interesting schematic. What does the Q1, D1, D3 network do? Is it some kind of error correction to reduce AGC distortion? or something much more mundane? will have to investigate further.

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

I've had some success with a vastly simplified version of my own circuit that I've been working on over the last few days.

Rather than apply a bias to the dc control voltage, I've used a divider to bias the jfets working 'position'. This meant I could scrap the op-amps and use a simple 1 diode half wave rectifier, but still get the tailored response. I got it up and running on the breadboard tonight and it is extremely promising.

Still work to do, there is some noise at lower frequencies, and the overall noise levels are not as good as my old system. However that was fully soldered up and installed, rather then a big mess on a bread board. I have also to add a zobel network to the power amp to reduce the load in the higher frequencies where my mixed mode amp (intentionally) tails off - this should help to tame the LM386.

cheers

Col

Edited April 20, 2009 by col

[Hank McSpank]

I think I've just stumbled upon the source of that AGC circuit zfrittz6 posted up a page or so ago...

(full URL here - http://www.qrp.pops.net/Idaho.asp )

Therefore it was a little off to suggest it was - ahem - 'lifted' from the Aussiemart compressor thread (& who's to say Aussiemart didn't 'lift' their's from the above site in the first place!).

I reckon this cct can be adapted to work with most poweramp chips ...even though it seems a little crude, I'll give it a trial with my TDA7052 soon. I'm even tempted to straighten out the pins of my 'terminated' LM386, just to see how well this cct performs with it, (this design - along with it's many derivatives - appears to be very ubiquitous.... so there must be *something* in it!)...but perhaps deriving the FET control voltage from my PIC instead of the output from the chip.

Edited April 21, 2009 by Hank McSpank

[zfrittz6]

simulacion

Q1 makes potentiometer electronic. D1 and D3 are in continuous polarization and C7 decoupling AC.

This circuit has very low distorsion.

Simulation showing the distortion that occurs with any FET

This circuit works well with any Poweramp for the few components you have, but is more distorted than the previous

You see the difference?

Saludos

Edited April 21, 2009 by zfrittz6

[Hank McSpank]

simulacion

Q1 makes potentiometer electronic. D1 and D3 are in continuous polarization and C7 decoupling AC.

This circuit has very low distorsion.

This circuit works well with any Poweramp for the few components you have, but is more distorted than the previous

That app Circuitmaker.....looks easy on the eye - I'd not heard of that it before.

I really ought to do more simulation, but truthfully, I haven't the time to learn it (though I realise, if I spend time learning it - it'll save me time!) - certainly the the spice sims seem to sump a lot of time to get up to speed, so I'm just getting out the breadboard & going for it. I've just been Googling for Circuitmaker - I see that the app has gone end of life, though there are plenty versions still abounding on the net (eg student ones). I quickly downloaded/installed the student one....for the life of me I couldn't find an LM386 (I wanted to simulate that simple AGC circuit) - with it being such a common/popular chip, I'm puzzled why it's not in the library?

The first circuit may have less distortion, but it's component count is getting a little chunky. I'm wondering here if the type of distortion the simpler AGC circuit introduces will even be audible (which is all we care about). When I've been putting together circuits with JFETs using a sig gen sine wave as an input, there has been some slight changing of the waverform as seen on my scope (I'm loathe to use the word distortion - even though that's what it is - becuase what I saw was mild distortion)...but no impact on the actual guitar 'sustained' output. Therefore perhaps what's important here is the type of distortion (eg clipping = very bad, gross waveforn modification = very bad, mild non square wave related distion - ok?)

edit: I've just seen your 'edited' post with sim of the simpler AGC circuit - that looks quite bad...I'm not seeing any such squaring wrt the sine waves on my JFET AGC circuits. What I see is more a scooping out ofthe hollows - I'll try & get a USB Scope screen scrape next time I'm working on the circuit.

Edited April 21, 2009 by Hank McSpank

[zfrittz6]

Sorry for the English translator of google.

If you can not find for Simulation lm386 nothing happens, I see how it simulates.

If you look good in the waveform above the circuit has no deformation in the wave, but with only FETS, there is a cut of the wave, that is the distortion at low frequencies is larger and if you can hear , because a FET is a distortion, but if we use a transistor as a regulator and OP as a wave attenuator is not cut.

Saludos

[Fresh Fizz]

I saw the following schematic in the Dutch Elektor may 2009 (Power in the pocket - eenvoudige PWM-versterker by Ton Giesberts)

Is not included in the english release of may, it might well appear a month later.

schematic

components

print layout

Self-oscillating PWM amp

1W / 8 Ohm, 1.7W / 4 Ohm

Power supply: 6...9 Volt

L1-C5: low pass filter to filter out oscillation frequency

D1-D2-R4-R5: Dead time setting to prevent both MOSFETs to be opened at the same time.

The oscillation frequency varies with applied voltage

According to Elektor's info:

9V - 660 kHz - quiescent current 44 mA

6V - 510 kHz - quiescent current 10 mA (4 x AA penlites!)

5V - 450 kHz - quiescent current 6 mA, but according to Elektor too little voltage for the MOSFET's gates.

A modification to reduce the power consumption at rest would make the amp more suited for the sustainer.

The power consumption could easily be tested with a small resistor in series with the 9V battery. Measure the voltage across the resistor and then apply Ohm's law to obtain the current. The resistor should be big enough to give a result on your multimeter, but not so big that it influences the measurement too much! (And don't forget to connect your sustainer driver while testing!) Just apply Ohm's law, for a desired output of 250 mVolts at the expected 44 mA you need a 5.6 Ohm resistor.

It's a bit like biasing your output valves in a balanced AB amp.

The oscillation frequency depends on R3 and C4 (and R2 and the output impedance of whatever is connected at the input I reckon).

To lower the oscillation frequency and quiescent current increase C4 and run the test again.

Cheers,

Fresh Fizz

[col]

I saw the following schematic in the Dutch Elektor may 2009 (Power in the pocket - eenvoudige PWM-versterker by Ton Giesberts)

Is not included in the english release of may, it might well appear a month later.

schematic

components

print layout

Self-oscillating PWM amp

1W / 8 Ohm, 1.7W / 4 Ohm

Power supply: 6...9 Volt

cool

A modification to reduce the power consumption at rest would make the amp more suited for the sustainer.

So how would you go about that ?

cheers

Col

[col]

Sorry for the English translator of google.

If you can not find for Simulation lm386 nothing happens, I see how it simulates.

If you look good in the waveform above the circuit has no deformation in the wave, but with only FETS, there is a cut of the wave, that is the distortion at low frequencies is larger and if you can hear , because a FET is a distortion, but if we use a transistor as a regulator and OP as a wave attenuator is not cut.

Saludos

That transistor regulator is very interesting.

Having said all that, I have a simple circuit with (excluding input and power buffers) uses 1 op-amp, 1 jfet, 1 diode and some caps and resistors.

In feedforward config, it has very little distortion up to the point where the input gain amp clips. At which time, the peak output is only about 15 % of the output with normal sustained string level inputs. There is also an increase in distortion (not clipping though) at very low frequencies that is slightly worse at higher input levels.

In feedback config, it can achieve low distortion and a reasonably level output amplitude with inputs from about 150mV to about 450mV

below 150mA, the output reduces gradually, and above, the distortion increases, soft clipping the tops of the wave.

The distortion at 82Hz is a *little* worse, but nothing like in zfrittz pics. The attack and release both take about 2.5 wave cycles at 82Hz (so very quick)

I favour the feed-forward version, but as with any simple circuit (and with the feedback version), the difficulty is in setting the thing up to work with a particular driver, pickup and guitar. There are going to be short burst of slightly audible distortion as the string decays to the point where the pickup doesn't output enough to drive the AGC into distortion, but these can be minimised by carefully tweaking all the settings.

I am definately going to investigate Zfrittzs circuit with the extra 'regulator' transistor though - that might give even output over a broader input range and even lower distortion levels.

cheers

Col

[col]

Here's the feedback version showing 82Hz a 450mV input.

The clipping in the first wave cycle dempnstrates that the AGC only takes 2-3 cycles to kill the peak and settle to a steady output. You can see in the last couple of cycles that there is some distortion (look at the shape of the -ve peaks). This only happens at very low frequencies, and it more prominent at higher input levels.

(off the page to the is a unity gain buffer and to the right is an LM386 power stage)

The light blue is ground, the green is virtual ground (4.5v)

Note the divider(22k & 18k) that sets the voltage at which the jfet starts to become sensitive to the control voltage, this allows the circuit to fully use the linear part of the jfets resistive range (I think ?)

The 10k and 100k resistors under the diode control attack and decay (according to the 'fast peak limiter' article, the large one should be at least 10x the size of the small one).

I suppose that with a scope, you could use a trimmer and really squeeze the max out of whichever actual jfet you use in a real life circuit.

Heres the same with 150mV input, you can see that there is virtually no distortion.

Now with 650mV to show how the distortion at higher input levels manifests itself

The feedforward design is more elegant IMO because by the time higher inputs are causing distortion, the output levels are much lower, so the distortion is barely audible.

With the feedback version, the difficulty will be to set it up so that it is sensitive enough to low levels, but at the same time, full sustain input levels are still low enough not to cause unacceptable distortion.

Increasing that window is doable, but requires the addition of extra components.

eg. a dual op-amp with one half improving the diode rectifier - no other extras there, and, for the feedforward version, the other half of the op-amp providing separate control over the gain to the control voltage (no extras there either as you can then use a follower for the input instead of a gain stage).

cheers

Col

Edited April 21, 2009 by col

[Hank McSpank]

[Here are some 'real world' USB Scope screen scrapes which shows the type of distortion I'm seeing controlling a JFET.

To explain where my JFET sits, firstly look at this simplistic diagram for reference...

I'm actually using 2 x Rin ....as opposed to the more normal 'one' seen above. My JFET is placed 'across' (in parallel) with one of the Rins - when the JFET is gated on, it bypasses (shorts out), one of these Rins - therefore leaving less overall combined resistance for 'Rin' ...this increases the opamp's gain, & vice versa. Nice & simple.

Firstly, this is JFET fully on...therefore crereating the maximum Opamp gain (just over 4V peak to peak.....I'm only using a single 5V rail so it's getting close to the clipping point), there a tiny amount of distortion but too little to be of a problem (certainly for sustainer circuits - not hi end audio!)...

Next the JFET 'pinched off' therefore the opamp is at its lowest gain setting (in this scenario it ouputs about 1.2V), again, a tiny amount of distortion, but to paraphrase Paul Daniels "not a lot"...

Finally, when the JFET is at it's 'mid bias point (half way between cutoff & full on), this is where the opamp output distortion is most pronounced (it's quite marked, in that it's starting to 'triangularize' the sine wave - but I still don't think it'll be a problem to use this as a driver output signal)....

Anyway, just thought it worth illustrating the type of JFET induced distortion I'm seeing when it's being used in the feedback loop of an opamp.

Edited April 22, 2009 by Hank McSpank

[Fresh Fizz]

I think I've just stumbled upon the source of that AGC circuit zfrittz6 posted up a page or so ago...

(full URL here - http://www.qrp.pops.net/Idaho.asp )

Therefore it was a little off to suggest it was - ahem - 'lifted' from the Aussiemart compressor thread (& who's to say Aussiemart didn't 'lift' their's from the above site in the first place!).

I reckon this cct can be adapted to work with most poweramp chips ...even though it seems a little crude, I'll give it a trial with my TDA7052 soon. I'm even tempted to straighten out the pins of my 'terminated' LM386, just to see how well this cct performs with it, (this design - along with it's many derivatives - appears to be very ubiquitous.... so there must be *something* in it!)...but perhaps deriving the FET control voltage from my PIC instead of the output from the chip.

I still find it a strange design. (I posted the version I knew with a cap at pin 5 of LM386, not the aussimart) To me it seems that only one of the can work.

The signal that goes from pin 5 into the MOSFET gate is not DC decoupled. I understand that a positive voltage is needed to properly bias the MOSFET. But when the battery decharges and voltage drops the bias setup is changed. Maybe it's been done with a purpose and I don't get it.

There are not that many more components required to mix AC from pinout 5 and DC from the battery.

AC: cap to resistor (the 4.7k one) to variable resistor (10k)

DC: + of battery to resistor (new one #1) to zener diode to ground.

junction (new resistor #1 - zenerdiode) to resistor (new one #2) to variable resistor (10k)

With an extra resistor it would even work better, but then I have to draw a picture. Now it's already not that clear.

Cheers

Fizz