the simple limiter (new effect)

[darwin_deathcat]

Hi all, here's the current schematic of a simple peak limiter I've been designing. I originally posted the first (Bad!) schematic in another thread, and got some great advice on how to move forward with this design. Although this schematic is much much better than the first one I made, I'm sure there may be some issues I have not caught. Can anyone see anything especially glaringly obvioulsy wrong? This is my first time designing a limiter, so any advice would be great. I have not yet breadboarded the thing because I want to get some of your feedback first (to hopefully save me from too much trouble on the breadboard!). Thanks in advance!

[Taylor]

This is just what I've been looking for - I want a limiter to keep the output of a feedback looper tamed. Hopefully you'll get some feedback from people who can read circuits better than I, but I'll probably put this thing together on perf soon, so we can see how it works in the wild.

Thanks for this!

[brett]

Hi
someone should probably SPICE model this.
I'm guessing that you'll want to reduce the 520k feedback resistor.  At present, the maximum amount of change in freedback resistance is from 470k down to about 250k (LDR=0k, so 520k parallels the 470k).  If the 520k resistor were 47k, you would get a 10 x change in feedback and gain, which is about right.

I'm also guessing here, but won't you want some resistance between C1 and C3 (look at the BMP gain stages for examples).  Then the circuit becomes simple to understand in terms of gain (like an inverting op amp, where the gain is the input resistance divided by the feedback resistance).

Any experts know a bit more?

[Cliff Schecht]

I'm pretty dang sure that all three transistors are mis-biased. Also, Q3 is serving no purpose right now (at least that I'm seeing). Signal comes in through the base and is output nowhere. Since the collector has no resistance, it will sit at 9 V (as will your signal itself, which again serves no purpose). Q1 can't share base resistors like you're trying to do here, neither transistor is getting the correct base current. You want to bias the transistor base for Q1 at about 5.1 V, which is 1/2 VCC plus the diode drop from the base to the emitter (your signal can never swing below this). Biasing at this point will give you max headroom. So with your 10k resistor on Q1, you're at about 200 uA of collector current. If you give Q1 separate biasing resistors (i.e. a resistor from VCC to base and base to ground), then the VCC to base resistor will be at about 195 k (we'll choose 220k). The resistor from base to ground is then about 280k, which you can choose at 330k (these values aren't super-critical, so it's good to use a common value). Your potentiometer trick on Q2 won't work very well either, you really want a steady DC current AND voltage to feed the base of that transistor. Instead, you could put a pot from the output to ground (with a DC blocking cap) and adjust signal levels here. Add another DC blocking cap and then feed Q2 with the appropriate resistor values from VCC to base and base to ground.

I'd recommend you get a simulation program and run some circuit simulations, it can teach you a lot about transistor design and bias.

[brett]

Hi
re: I'm pretty dang sure that all three transistors are mis-biased.

I agree that it probably needs a resistor of abouit 470k from the collector of Q1 to its base to provide some base current and turn Q1 "on".  Similarly for Q2.

(But I wonder why you need two transistors here?  Won't one modern transistor have enough gain to drive the LED? (ie Rc = 1k, Re=0, hFE=400).  If so, the limiter pot would naturally go on the base of the transistor.)

R4 should be 10k (or thereabouts, assuming it is a common emitter with high open-loop gain (10k/150 = 70), but less with the LDR and fixed resistor.  Why not use a BMP gain stage and replace the feedback resistor with an LDR?  (and obviously delete the clipping diodes and high-taming feedback cap)
cheers

[Gus]

Looks like the Q3 is a section of a BMP with the collector resistor copied wrong 10 should be 10K.

Even if you get the gain Q3 section biased right you will still have issues.

Q3 work out the changing input resistance of the circuit.  Look at the gain  reduction part (AC collector to base section) and DC bias feedback path R6.

  What is Q1s input resistance does it change with input level?  Q1 and Q3 both "load" what is before this circuit.

R2 is both a bias and gain control as drawn.  So as you turn up the gain the Q2 collector voltage goes more to ground and more LED current flows.  You need to control the LED DC current if you want it just barely on and then you can AC couple more or less current to it too vary the light output

You kind of have the idea of how a compressor like this works.  The details count.

  Issues in this case the input resistance, DC operating points of the transistors,  DC and AC current control in the LED.  Q1 might work fine as drawn this circuit being a compressor used as an effect? however I would use a EF before the circuit to buffer the input R.  Set Q2s DC operating points separate from the AC input a stable bias network and adjust the LED DC current with no signal input make a stable DC bias network and AC couple the signal.

I don't believe in sims with simple circuits like this, pen and paper and building can help understandinng better with circuits like this .


This is a simple 3 transistor compressor at 9VDC  Q1 might be a OK design with the circuit if a buffer is before it and often you want a offset before the sidechain starts to work.  You do not need both sides of the input waveform.   Add a EF buffer and a gain setting resistor in series with the base of Q3(like brett posted).

[CynicalMan]

I don't believe in sims with simple circuits like this, pen and paper and building can help understanding better with circuits like this .

Also, LED-LDR pairs are hard to simulate.

[Cliff Schecht]

I don't believe in sims with simple circuits like this, pen and paper and building can help understanding better with circuits like this .

Also, LED-LDR pairs are hard to simulate.

For this circuit, yes I completely agree. I meant he should run some simulations on some simple transistor circuits and maybe even something like the Orange Squeezer, but even then the trick becomes whether one can accurately simulate the pick attack and naturally decay of a guitar.

[darwin_deathcat]

Hi all, thanks for your great comments! This is just what I was hoping to get by posting it up here. Let me try to summarize/synthesize everyone's comments and see if I get it right (plus I'll add some of my own thoughts).

Okay, I'll start with the Q3 part: Yes, this is just a snippet, found in various "muff" circuits (this one comes from Jack Orman's muffer circuits, but is similar to the BMP stages). Thanks to brett for finding the mislabled R4, it definatley SHOULD be 10k. The feedback resistor network is all I've changed to the basic "muff" topology. As per my calculations, the total AC feedback reistance is ~250k when the LDR is 0 (no light shining on it) and 450k when it is 10000K (bright light shining on it). Will this not have the effect that I am thinking? That is, as the LDR's resistance grows, the total gain of the Q3 circuit snipit reduces? Another approach would be to wire the LDR up so that it works as a volume attenuator. (see schematic 2 below). In this case, varing the value of R5 would change the response for volume reduction (I've arbitrarliy set it to 520k for the moment). I put this stage before the Q3 stage, and use Q3 as a simple output booster section.

Next I'll tackle Q2. Just to be clear, Q2 is arranged so as to be a switch, not an amplifier. AFAICT, if one disregards R2, the rest of this part of the circuit should be okay the way I have it. Q2 will be switched on when the AC signal at it's base gets above ~0.5 volts, at which point current flows from collector to emmiter, in this case from the 9volt battery to the red LED. Using a transistor in this case should also work to "ramp up" the brightness of the LED since there is a "zone" around that 0.5 volt "threshold" where the transistor turns on but still has some resitance from collector to emmiter. So I was thinking that it would turn on the LED when the base current was somewhere a little less than 0.5v, but that the LED would not reach maximum brightness until the base current got somewhere over, say, 0.6v. Is that incorrect? The R2 part is what I'm not completely sure about. I want to be able to control the point int he input signal at which the LED starts to get turned on. So I figured that I'd need a boosting stage (Q1) to get the signal higher than necessary to turn on the Q2 switch, then use a pot to control the strength of that signal (shunting some of it to ground), and thereby the point in the "peak" where Q2 starts to turn on. Looking at it this way, does R2 serve that purpose?  

As for Q1, I think I'll simply replace it with another "muff" stage. As is it is, it's a total clone of the BazzFuss circuit as posted by BeavisAudioResearch on his site. I wanted to keep parts down, but I think that doing so in this stage will cause more harm than good. To explain the Q1 stage, my thinking was thus: Signal comes in from input, and is directed to both Q1 and Q3. Q1 amps up the signal so that it's in the correct range to activate the Q2 switch. None of the signal from this part of the circuit gets through to the output, it simply is the "forward" looking part of the circuit which activates and controls the "limiting" part of the circuit. The signal path that is fed thru Q3 IS the signal path that we are interested in for this. Q3 is set up so that the gain decreases as the signal gets above the threshold of the Q2 switch, thereby enacting the "limiting" effect. It seems to me that the main issue brought up with the topology as I have it now is that there may be "cross talk" between the Q2 and Q1 stages. Any ideas on how to isolate them better? I've added some resitance to the legs of the splitter, and given each leg it's own capacitor. Oh, and I've added a 1meg resitro to ground at the input to set the input impedance (whihc was definately lacking before!).

  The idea to have an input buffer is a good one. I should mention that my application for this circuit is to follow a seperate preamp (a booster, distortion, etc), so that it should be getting input levels at line level or above. To keep the circuit "smaller" and "simpler" I think that I won't, for now, add a buffer, unless folks think it is ABSOLUTELY necesary.

Since a picture is worth a thousand words, the following to schematics show the changes as I see them. The first maintains the gain reduction idea, and the second introduces the volume reduction idea. Please take a look and let me know what you think.

Oh, I don't have the capability or know-how to do simulations... If someone is interested enough in this to do a simulation, that would be really great! Once I get a reasonable-seeming circuit, I will breadboard it, and see how it actually works...

[darwin_deathcat]

Just bumping this up to the top. Anyone got anymore advice lookig at the last to schematics?