MUFF WIGGLER Forum Index
 FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 
WIGGLING 'LITE' IN GUEST MODE

I need help: drawing a wave multiplier circuit w/ rectifiers
MUFF WIGGLER Forum Index -> Music Tech DIY  
Author I need help: drawing a wave multiplier circuit w/ rectifiers
Blackened Justice
Okay, so some time ago I had an idea for a module (please don't tell me it's not original, I realize it may not be), please see the original topic: https://www.muffwiggler.com/forum/viewtopic.php?t=7847&highlight=recti fiers+modules

So, I need help in actually moving this project forward. I have absolutely no experience in designing circuits. I have made a patch for the Nord Modular G2 demo, that's pretty much complete in design, I'll attach it in my post. The thing is, I really have no idea how I would go from translating that into a readable schematic, with component values and the like.

I'll describe it as best as I can, with words, for those who don't have the G2 demo. Okay, I still haven't decided on the final number of stages, but the module is basically several stages of VC offset summing, rectifying, fixed offset summing and amplifying. Assuming audio signals in the +-5V range (10 V peak to peak), and a +-12V PSU:

Basically you get an input signal, mix it with a CV input (LFO) and rectify it. Then you add a -5V offset to center the wave at 0v, and then amplify it with a 2x gain to retain the 10 V peak to peak. Then repeat again, with another CV input. I have been considering 8 stages of this, as more seems overkill.

Ideally, the module would have both a dedicated CV in and audio output for each stage, a couple of master audio inputs with attenuators, to facilitate feedback loops. It would be DC-coupled, in order to process LF signals too.

If anyone is willing to help me, I'd sincerely appreciate it a lot, I'll try and design the PCB afterwards, and if anyone is interested, I might do a small commercial run.

Best regards,
João Paquim
Blackened Justice
Anyone out there?
neandrewthal
I've never designed a circuit either, but here's my go at it. Hopefully it will at least be a starting point. Maybe someone who knows better can critique it. Oh yeah, the -5v should be -2.5v. You could derive that with some s[/img]ort of voltage divider circuit.

daverj
neandrewthal wrote:
Oh yeah, the -5v should be -2.5v. You could derive that with some sort of voltage divider circuit.


Or leave it -5, since that's an easy voltage regulator to get, and use 200K resistors to -5 instead of 100K to -2.5
jjclark
As I mentioned in your other thread, you can AC couple the sections instead of offsetting. This will simplify the circuitry. And I suggest that you use "precision rectifier" circuits (basically putting the diodes in the feedback of the opamps) otherwise your rectified output will not go down to zero. Diodes will have a 1/2volt (for germanium) to 0.7volt (for silicon) drop across them. Placing the diode in the feedback loop cuts the effective voltage drop by a factor equal to the opamp gain.
Just do a google search for precision opamp (fullwave) rectifiers to see the circuit.
neandrewthal
daverj wrote:
neandrewthal wrote:
Oh yeah, the -5v should be -2.5v. You could derive that with some sort of voltage divider circuit.


Or leave it -5, since that's an easy voltage regulator to get, and use 200K resistors to -5 instead of 100K to -2.5


Oh, I didn't know you could have some inputs amplified and some not with the same op-amp. Then you could insert the CV in there via a 200k as well, making 3 of the 5 opamps in that part of the circuit redundant.
daverj
Blackened Justice wrote:
I have been considering 8 stages of this, as more seems overkill.


You might want to make it with fewer stages. Something to consider is the amount of amplification this will have, and what that will do to noise.

If there is, for example, 10mv of noise on the original signal, which is only 1/1000th of the original 10v p/p signal, so not very noticeable, by the time you get out of stage 8 that 10mv becomes 2.5 volts, or 1/4 of the total signal.

That's a lot of hiss.
neandrewthal
daverj wrote:
Blackened Justice wrote:
I have been considering 8 stages of this, as more seems overkill.


You might want to make it with fewer stages. Something to consider is the amount of amplification this will have, and what that will do to noise.

If there is, for example, 10mv of noise on the original signal, which is only 1/1000th of the original 10v p/p signal, so not very noticeable, by the time you get out of stage 8 that 10mv becomes 2.5 volts, or 1/4 of the total signal.

That's a lot of hiss.


Yeah, and with 8 stages I could see things getting so harmonically complex that the signal itself sounds more and more like noise.
daverj
Not to mention that it would push many frequencies up beyond the range of what you can hear. A 60Hz triangle going in would be over 15KHz at the 8th stage.
Blackened Justice
You're probably right. Remember, all the testing I've done has been done with the Nord G2 demo, so real-life issues such as noise really aren't a problem there. What should I consider, then? 4 stages? Maybe a dual set of 4 stages multipliers behind a single panel?

As for AC coupling, that would be cool, but the offset amount could be of some interest to have controlled by a knob.

I'll have to do some more testing, I wish I had an oscilloscope, but for now, I'm going to have to trust my ears.

Please keep the suggestions coming, everyone. It's been really useful.
DGTom
If it were me, I'd build single, or, maybe dual stages, & just cascade them when you need to; its modular!
Blackened Justice
What about a 4 stage, or whatever, but with individual inputs for each stage, where each stage's output would be normalled to the next stage's input? It would allow for completely modular patching and still have them cascaded by default.
Blackened Justice
neandrewthal wrote:
daverj wrote:
neandrewthal wrote:
Oh yeah, the -5v should be -2.5v. You could derive that with some sort of voltage divider circuit.


Or leave it -5, since that's an easy voltage regulator to get, and use 200K resistors to -5 instead of 100K to -2.5


Oh, I didn't know you could have some inputs amplified and some not with the same op-amp. Then you could insert the CV in there via a 200k as well, making 3 of the 5 opamps in that part of the circuit redundant.


Okay, I've been reading up on op-amps, and now have a better idea of what you did on the circuit. However, I'm a bit confused by your comment here, could you post this simplified version you speak of? Maybe even using AC coupling instead of summing -5V DC, to simplify it further, as was suggested.

Cheers,
João Paquim
neandrewthal
Here is my circuit simplified by Dave's suggestion:





(the cut off resistor at the top is 100k)

I don't see how AC coupling would simplify my circuit any further, but I think what jjclark was getting at is to take this circuit:




but also run your lfo into each opamp via a 10k and 20k resistor as shown, then run it through a capcitor, lather, rinse, repeat. I have to admit I don't understand it, but if it works that way with higher performance, I'd say that's the way to go.
Blackened Justice
Thank you very much, you've been extremely helpful. AC coupling would simplify it by removing the -5V regulators and the associated 200k resistors, and just putting a capacitor in series with the pair of diodes.

I've got some questions about circuit design, in general:
1) How do you choose the resistor values when dealing with opamp circuits? I know the gain of an opamp is defined by the reason between the resistance in the feedback and the resistance at the input (when dealing with inverting amplifier configurations), but how do you actually specify said resistors at 100k and not 10k or any other value?

2) Why is there a 1k resistor on the output? Shouldn't the output inpedance be the lowest possible?

3) Before each stage of rectification, do you really have to invert and reinvert the voltages, couldn't you like invert it once and put that inverted output through the first diode, then put the original signal in parallel through the other diode? Would that make a too low impedance input? If so, then I could just as easily use an op-amp at the input in a non-inverting amplifier configuration, right? Or has it got something to do with the initial mixing of the input and CV voltages?

Thank you very much,
João Paquim
daverj
1 -op amps can use a wide range of resistor values. They can effect the input impedance, the bandwidth, the noise, the maximum voltage swing, etc... In op amps connected to external inputs the input impedance is often the deciding factor. 100K is the most common. You then select the feedback resistor based on the input resistor.

2 - in the early days the 1K was added to prevent the opamp from getting killed when you short the output to ground each time you patch a cable. Modern op amps can survive shorts. But shorting an op amp can cause a high current spike of 50-80ma in your power supply, which can get back into other modules. With the resistor there it limits that current. It also reduces problems if you happen to connect two outputs together.

3 - In his circuit the first op amp in each pair is used as a mixer. The second one is used as an inverter. The first op amp in the second and third pair also amplify the signals and level shift them.

re: AC coupling - while you can AC couple the stages, the drawback is you lose the ability to handle slow CV voltages. Also, because the duty cycle will change based on the shapes of the input signals, the cap will cause DC shifts as the duty cycles change. Adding a 5 volt regulator and resistors will only add maybe $1 to the cost, and is worth it.

You can also eliminate the 5 volt regulator and use a larger resistor to -12 volts. The advantage of the 5 volt regulator is that it isolates that voltage from the main power supply rails, so does not add power supply noise coming from other modules into your signal.
Blackened Justice
Thank you very much for your answers smile I really appreciate it.

Regarding answer 3), why not mix the signals passively, and avoid one of the op-amps?
daverj
Quote:
Regarding answer 3), why not mix the signals passively, and avoid one of the op-amps?


For circuits like this you would normally use dual or quad op amps, so no extra chips are needed to have the mixer.

The diodes are already a form of passive mixer. Trying to add another passive mixer in front of that will result in attenuation of the signal because of the resistance after the diode. That will increase the noise from extra amplification to compensate for the loss.
Blackened Justice
Ok, thank you very much for everything. I'm gonna get some components, and try to breadboard a prototype ;D
I have some LM741 opamps, will those do fine for this type of thing?

Cheers
daverj
They're slow. They're also bipolar, which means that using this circuit they'll drift with temperature (extra resistors would be needed to prevent that).

Better off using TL072 chips.
Blackened Justice
I have some more questions about the circuit:

1. If I want to take "tapped" outputs from the circuit, where would I take the signal from? Would I have to just repeat the ouput part of the circuit? Or mix the -5 and the CV using seperate opamps in series, and take the output from the -5 one?

2. In the output section, couldn't I mix using a non-inverting configuration, and thus avoid the last op-amp? It made sense in the original design, for both regular and inverted outputs, I think it can be avoided.

3. I might want to normalize the CV ins to +-5V sources. Can somebody point me in the direction of a simple attenuverter, or polarizer, if you will?

4. For all the +-5V I might have in the circuit, would I be better off using voltage regulators to get them, or just use a pair of resistors in a voltage divider configuration?

5. Should I use silicon, germanium or Schottky diodes?

6. How can I estimate current draw for this?
daverj
1 - The circuit as shown is a 2 stage unit. So there is only one "tap" point in the middle. Take it from the output of the third op amp

2 - You can take the final output from the 5th op amp and eliminate the last one.

3 - did you do a search of this forum? I would think there's a couple of threads with schematics for that

4 - Using a regulator for the -5 is best. You could use larger resistors and take it straight from the -12, but you'll mix in any noise and fluctuations that are in the power rails. A regulator eliminates that.

5 - I would use Schottky. Any of them would work. It might be best to add a large value resistor between where the diodes meet and -5 (or -12) to keep the diodes on at all times (maybe 100k).

The resistor value going between -5 and the input of the 3rd and 5th op amp would be very slightly different value for the different type diodes, though not by much, to compensate for the different voltage drops.

6 - just build it and measure it. The current draw is going to be very small.
Blackened Justice
daverj wrote:
1 - The circuit as shown is a 2 stage unit. So there is only one "tap" point in the middle. Take it from the output of the third op amp


But then I'd be getting the desired signal mixed with the CV In 2, which I may not want.
neandrewthal
Aww, shucks. I didn't even think of that d'oh!

You'll probably have to just duplicate the output section for each tap, but get rid of the second amp if you want.
daverj
This is what makes designing circuits so much fun. hyper
Blackened Justice
Can't I basically divide the mixing part in 2 sections, so that I mix the rectified signal with the -5 in an op-amp and mix the rectified centered signal with the CV 2 in a second op-amp? So that the tap would just be a parallel branch after the first op-amp.
neandrewthal
Blackened Justice wrote:
Can't I basically divide the mixing part in 2 sections, so that I mix the rectified signal with the -5 in an op-amp and mix the rectified centered signal with the CV 2 in a second op-amp? So that the tap would just be a parallel branch after the first op-amp.


I was almost going to suggest that, but then I realized that the LFO signal you insert into the second op-amp would only be half wave rectified.
Blackened Justice
No, I mean actually adding another opamp dedicated to the mixing of the CV in, before the parallel branch to the diode.
daverj
Sure you can. At that point why not just build several single stage units , each with 2 inputs. Then normalize the output of the first unit into one of the inputs of the second unit (and second to third, etc...).

That way they can be used individually, or be cascaded in a chain if you don't use one of the inputs.
Blackened Justice
Blackened Justice wrote:
What about a 4 stage, or whatever, but with individual inputs for each stage, where each stage's output would be normalled to the next stage's input? It would allow for completely modular patching and still have them cascaded by default.


I suggested exactly what you did a few posts ago ;D /\
|

Anyway, I'm gonna breadboard it when I get the -5v regulators, and then I'll write back to comment on it.

On a more pratical concern, 2 ins and 1 out per stage leaves us at 3 jacks per stage. I'll probably put a polarizer on the CV ins and a regular attenuator on the Signal ins. So that leaves us at 2 pots per stage.

I was trying to design a polarizer myself, and came to a relatively simple design: put your input into a 1:1 inverting amplifier configuration, and put the output of that opamp through an attenuator. Then put the original signal through another attenuator. Then connect the output of both attenuators together. Of course, this would depend on the existence of dual pots, a single knob to control 2 resistors. Does this exist?
My other solution was to route the inverted and the original signal into both of the potentiometers "inputs" and then take the output from the pointer. Would this work? I know that even with the potentiometer at the extremes, there would still be a bit of the other signal present, but couldn't this be overriden by using a very large value pot?

I ask because I looked at the schematics for the Fonik attenuverting mixer, and it seems overly complicated.
daverj
Blackened Justice wrote:
My other solution was to route the inverted and the original signal into both of the potentiometers "inputs" and then take the output from the pointer. Would this work? I know that even with the potentiometer at the extremes, there would still be a bit of the other signal present, but couldn't this be overriden by using a very large value pot?


That will work fine as long as the signals are both buffered. You don't want to connect both the pot and the input to the inverting amp directly to the input jack. That could cause variations in the gain or oscillations. Run the jack to both a non-inverting amp and to an inverting amp and then those two amps to the two ends of the pot.

There's no worry about bleed since the signal at one end of the pot is exactly the inverse of the other end so the bleed is canceled out. You would only get bleed if using this method to mix two unrelated signals.
Blackened Justice
But won't I get slightly less than full voltage swing?

Why can't I connect the pot to the input jack? What do you mean variations in the gain?
daverj
Many modules have series resistors on their outputs. These range from zero (no resistor) to maybe 1K. If you connect one end of the pot to the jack and the input of the inverting amp, and the other end of the pot to the output of the inverting amp then the pot itself becomes part of a feedback circuit with the amplifier.

Tthe gain of that circuit is unknown since the series resistor on the output of whatever module gets plugged in is unknown.

With a buffer amp between the jack and the pot the output resistor of the module plugged in becomes negligible. And the inverted signal can't feedback backwards through that buffer so the pot is not part of a feedback circuit.

As for your other question, if the pot allows a small amount of the opposite signal to mix in when the pot is at an extreme, yes it slightly reduces the levels. But that is a constant and is easily compensated for by using a slightly smaller resistor on the input of the amp that the center wiper of the pot is connected to.

Or, better yet, it can be compensated by increasing the gain of the inverting and non-inverting amps. In fact, you could increase the gain of both to maybe 2x gain anyways. That gives the circuit more range, so it can go from about 2x positive to 2x negative gain.

You might also want to toss in a non-polar cap and a switch to connect it on at least one of the units, so you can AC couple a signal into that unit (maybe do it on the first unit in the chain).
Blackened Justice
Thanks for your description, it makes more sense now. Well, but I guess I'll use Fonik's design, even if I don't quite understand it, because it uses a single opamp per channel.

An AC/DC coupling switch seems cool, even if any possible offset the AC input signal has could be corrected by the internal offset summers.
daverj
Quote:
An AC/DC coupling switch seems cool, even if any possible offset the AC input signal has could be corrected by the internal offset summers.


An offset pot will remove DC, but won't remove low frequencies (sub sonic), which a cap will. In fact, it could be a 3 position toggle switch and the 3rd position could be a different frequency high pass filter by simply using a smaller cap.
Blackened Justice
Just an update: I've breadboarded the circuit (a single stage, using 3 opamps), and it works great! I put a guitar through it, and it sounded pretty much like an octave fuzz. With simpler waveforms, it added some pretty tasty and crunchy harmonics. It sounds really cool being animated with a low frequency CV, and is great for drones when using audio-rate CVs, with a harmonic relationship between the In and CV frequencies. I'll try another stage, and then I'll design a PCB for it. I'm using TL074s, so the design repeats itself after 4 stages (4*3 opamps=3 TL074s).
Blackened Justice
Okay, I'm at the process of designing a PCB, and I have to make some decisionsto make:
Should I use polarizers or just regular attenuators? Polarizers would add a pair of opamps and some resistors per stage, and I don't know if it's justified.
daverj
There's no point at all to having a polarizer on the first stage. You'll get exactly the same result with positive or negative signals going in.

On further stages it might be interesting in some cases (and make no difference in others) since it would add or subtract from the previous stage. It would have the most effect when the signals are similar or related.
MUFF WIGGLER Forum Index -> Music Tech DIY  
Page 1 of 2
Powered by phpBB © phpBB Group