Following up on my interest in a simple phaser that doesn't require special parts, which began here (http://www.diystompboxes.com/smfforum/index.php?topic=88682.0), I've been playing with some ideas. I tested out Tim Escobedo's Q+D VCF (http://folkurban.com:80/Site/QDVCF-714.html), and found it to be really nice for how simple it is. Of particular interest is his CV scheme. I haven't ever seen that done elsewhere and I totally don't understand why it works. I might start another thread to discuss that. But, put simply, it does work, and that's all we need to know to put it to use.
The idea here is much like my earlier CMOS idea, and also much like ROG's Phozer (http://www.runoffgroove.com/phozer.html). What's unique about this circuit is that it gives you a very phaser-like sound with very few parts, and most notably, no optocouplers, LDRs, FETs, etc. No rare parts, nothing to match, no trims to set up to tune it in. Just a quad opamp and some passives.
The LFO is the one Rick used in his Little Angel (http://www.diystompboxes.com/smfforum/index.php?topic=86297.0). I don't know whether that's an original idea of his or a stock LFO, but I like that it gets pretty sine-ish with just a single opamp. Triangle might be nicer, but I'm aiming for ultimate simplicity here. The input goes to a buffer, then splits to 2 stages based on the Q+D VCF - these create 2 peaking filters with a slightly lowpass response. That's analogous to a 4-stage phaser with negative feedback. The resonance in the lower-tuned filter is cut a little (1k in series with the 4n7 cap pair) as it got a little woofy with full resonance. These stages are tuned to different frequencies and controlled by the LFO.
(http://musicpcb.com/wp-content/gallery/circuits/brazen-prophet.jpg)
I will be doing a bigger version with a switch hitter amp on the end and more controls. This will allow for a resonance control which can change the filters from peaking to notching. I'm also going to try out a CMOS inverter version. If it works, we can get a fake 12-stage phaser with just 2 chips!
Note: you do need a glowing blue resistor for the negative feedback loop around the lower right opamp, otherwise the mojo will die.
Wonderful stuff Taylor, posting a layout? Very inventive!
Strategy
Thanks Strategy. Like most things, it's a jumble of other folks ideas. :) But I think it's pretty promising.
I will definitely be posting a home-etchable layout once I tweak it some more - I just came up with the idea while trying to sleep last night, then built it and tweaked it this evening, so it could benefit from some tweaking before I call it "done".
PM coming at you.
Elegant, Taylor, elegant.
I think any hairy-arsed numpty can make a novel circuit by throwing half a dozen handfuls of components at it, but it takes a true ninja to make something cool with hardly nuthin'.
Little Angel, Tiny Tremolooooooo, Noise Ensemble, now this - we've been quite spoiled recently.
Quote from: markeebee on February 26, 2011, 05:23:35 AM
I think any hairy-arsed numpty can make a novel circuit by throwing half a dozen handfuls of components at it.....
I can't, by the way, so props to the hairy-arsed too.
nice taylor... cant wait to hear this neat little circuit...so the quad would be an tl074 i guess!.
ive got a glowing blue resistor..i shine an led on it... :icon_wink:
Neat and simple - I like it a lot. The LFO by the way is very much stolen from the Phase 45 LFO, and as you say, really simple but effective. I've used it in a few different circuits. I'm wondering if the input buffer is needed, or could be omitted and the opamp used for another differently tuned filter stage - the Phoser lacks an input buffer, as does the Nurse Quacky, which also has a similar filter. BTW, I breadboarded a single filter like those a couple of months ago together with the S/H circuit stolen from the Maestro FSH, in an attempt to make a super simple S/H filter, but I was dogged by bleed from the noise generator, maybe I should revist it, as the actual effect sounded really cool (apart from the bleed).
I'll have a vero at it, but first a noob question...where do you get the 4.5V?! the voltage divider I see has a 100k and a 47k, with the online calculators it outputs 2.8V from a 9V source...I guess there should be two 100k there?!
^^^
Yeah the P45 LFO has equal value resistors in the divider, I tweaked them for the Angel as it's running into a PT2399 running at 5v. Maybe Taylor would like to tweak it back to 100k/100k for the Prophet?
Also which is the orientation for the 10ยต output caps!? negative towards out!?
The schematic is drawn in simplified form - not shown are the details like vref divider, power supply filtering, bypass switching, LED, etc. which is the usual style for showing new concepts.
Quote from: frequencycentral on February 26, 2011, 07:49:05 AM
Neat and simple - I like it a lot. The LFO by the way is very much stolen from the Phase 45 LFO, and as you say, really simple but effective. I've used it in a few different circuits. I'm wondering if the input buffer is needed, or could be omitted and the opamp used for another differently tuned filter stage - the Phoser lacks an input buffer, as does the Nurse Quacky, which also has a similar filter. BTW, I breadboarded a single filter like those a couple of months ago together with the S/H circuit stolen from the Maestro FSH, in an attempt to make a super simple S/H filter, but I was dogged by bleed from the noise generator, maybe I should revist it, as the actual effect sounded really cool (apart from the bleed).
Yeah, in my real life version I actually don't have a buffer, but I have it after another pedal just because it overdrives easily and I needed a volume knob in front of it. As I tweak it I'll see if the buffer really is needed or not - I'd certainly be happy to free up an opamp for another filter!
Renegadrian, thanks for your willingness to do a vero layout, but you may want to hang back just a little bit as some changes will likely be made to this in the coming days.
There was a published design from a couple of decades ago using a similar setup. It happened to use an unavailable VCF IC, so it never got much exposure. It did a phaser the obvious way, two cascaded notch filters, like this. I can't remember exactly where it was. Possibly a synth circuit.
A phase is, after all, a moveable set of filter nulls and perhaps peaks. How you get the notches is immaterial to the sound. The bridged-T is an OK way to do this.
PAIA stuff from 1968 - 197? used the trick of using the forward resistance of a diode as a tuning method in a Twin T. I think the "Synthespin" did this, and some of the PAIA synth modules did as well. It works well for signals across the diodes down in the 10mV-100mV range, but gets rapidly more distorted as the signal voltage across the diode approaches any significant fraction of the diode forward drop. You can stack several diodes for bigger resistance range and bigger signal tolerance.
The same circuits should work with a bipolar used in its variable resistance region (similar to the use in Dr.Q, Pulsar, and Seamoon Funk Machine) or LDR (bigger signals, low distortion), JFETs as variable resistances, and probably switched-capacitor "resistances".
It isn't a fake phaser to the extent that it makes notch filters. It's a real phaser, as real as the phase shift method of getting moveable notches. It's another point in the continuum of tradeoffs of result/$, precision, and complexity.
Quote from: R.G. on February 26, 2011, 02:51:52 PM
There was a published design from a couple of decades ago using a similar setup. It happened to use an unavailable VCF IC, so it never got much exposure. It did a phaser the obvious way, two cascaded notch filters, like this. I can't remember exactly where it was. Possibly a synth circuit.
A phase is, after all, a moveable set of filter nulls and perhaps peaks. How you get the notches is immaterial to the sound. The bridged-T is an OK way to do this.
PAIA stuff from 1968 - 197? used the trick of using the forward resistance of a diode as a tuning method in a Twin T. I think the "Synthespin" did this, and some of the PAIA synth modules did as well. It works well for signals across the diodes down in the 10mV-100mV range, but gets rapidly more distorted as the signal voltage across the diode approaches any significant fraction of the diode forward drop. You can stack several diodes for bigger resistance range and bigger signal tolerance.
It isn't a fake phaser to the extent that it makes notch filters. It's a real phaser, as real as the phase shift method of getting moveable notches. It's another point in the continuum of tradeoffs of result/$, precision, and complexity.
Ah, very good info, especially regarding the diode trick. This circuit is certainly prone to distortion (but of course so are some other popular phasers). I will definitely play around with more diodes to see if I can get some more headroom. Also, a bi-phase kind of thing should be easy by inverting the diodes in one filter, as per Tim's original Q+D schematic. :icon_idea:
And thanks for the encouragement regarding the "realness" - it was my guess that a peak is a peak, a notch a notch, regardless of how it was obtained. I just put "fake" in there to forestall comments that I'm a dummy for not knowing what a phaser is. :)
I've done some more testing of this and it's pretty cool. One great thing is that it will be really easy to add various mods to it because it's such an "immediate" circuit concept. I want to try to add a control for resonance - I know how to do it using a dual pot, but hope to find a trick to do it with a single pot.
Quote from: Taylor on February 25, 2011, 11:01:56 PM
..but I'm aiming for ultimate simplicity here.
Why?
Btw. if you really do, then IMO you can dump whole input buffer, seems redundant..
T.
Simple circuits are cool because:
-more people will build them
-they provide a good stepping stone for beginners
-designing simple circuits that work well is difficult, so offers material for learning more about electronics, flexing our thinking ability, expanding lateral creativity/problem solving
-I'm often too strapped for time to build and debug my more complicated ideas - consequently I've never built the CMOS 12-stager linked in the first post. Whereas I could whip this one up very quickly
-they can be made in small boxes, not that I care about that personally (http://www.diystompboxes.com/smfforum/index.php?topic=74465.0). ;)
It's mainly the first thing. I'm contributing an idea to the community, so the idea is to make it something that more people will be into. I've already contributed some complex and obscure ideas that nobody has ever built, and while there's plenty of enjoyment in that, it would be cool to see somebody actually build these things. :)
I just breadboarded this (it's currently making whooshing sounds on the amp behind me :)), and I have a few tweaks to suggest. Firstly, the input buffer is redundant, I assume you put it there anticipating future need, but for now it's not really necessary. Also, I'd increase the mixing resistors to 10k each, just so your op amps don't have to drive 2k loads. Thirdly, I'd mess with the LFO bias a bit. For me, the stock setting had the bias way too low. The phaser sounded like a really bassy pulse, instead of the nice sine-y sound we want. I found that a 47k resistor in parallel with the 10u cap on the LFO's output helped a lot with that. That's even a place for a possible mod: a 1M-ish log pot in parallel with that cap that would shift the sweep range up or down.
The major issue to me seems to be the distortion. I don't have time now, but tonight hopefully I can take a look at the filters in LTSpice to see if you can reduce the gain without affecting the filtering.
Anyway, a very cool circuit. I hope you don't mind if a tag along in the design. ;)
I've actually got most of those changes planned/implemented. On my build, I have a CV pot that adds to the LFO so you can shift the sweep through different ranges - without that (as I've drawn it) it's definitely not too usable.
The schematic I posted was kind of just to get the idea across and I wouldn't recommend anybody build it just yet. I will have a tweaked version very soon. I don't mind at all you or anybody else suggesting mods or changes, in fact that's what I hope for! But it might a good idea to wait for my finalized schematic just so that the development is linear instead of parallel paths, which would confuse matters considerably.
Regarding distortion, Tim had a trim pot v. divider on the front end of his Q+D VCF - I'm just turning the guitar volume down. As RG noted above, adding more LEDs should give more headroom, and a fixed v. divider on the front end would do the trick, at the cost of noise.
Quote from: Thomeeque on February 26, 2011, 04:23:23 PM
Quote from: Taylor on February 25, 2011, 11:01:56 PM
..but I'm aiming for ultimate simplicity here.
Why?
simple circuits to do things that previously required complicated circuits are Art.
Quote from: Hides-His-Eyes on February 26, 2011, 05:12:39 PM
Quote from: Thomeeque on February 26, 2011, 04:23:23 PM
Quote from: Taylor on February 25, 2011, 11:01:56 PM
..(Triangle might be nicer,) but I'm aiming for ultimate simplicity here.
Why?
simple circuits to do things that previously required complicated circuits are Art.
On the other hand ultimately simple phaser with unpleasant sweep would be waste of time, that's my point.
If you think this thing is a waste of your time, then your time would probably be better spent checking out circuits that interest you. ;)
More constructively, I have worked out a more triangle-ish version of the same single opamp LFO.
Quote from: Taylor on February 26, 2011, 04:33:19 PM
Simple circuits are cool because:
-more people will build them
-they provide a good stepping stone for beginners
-designing simple circuits that work well is difficult, so offers material for learning more about electronics, flexing our thinking ability, expanding lateral creativity/problem solving
+1
Quote from: Taylor on February 26, 2011, 05:32:07 PM
If you think this thing is a waste of your time, then your time would probably be better spent checking out circuits that interest you. ;)
More constructively, I have worked out a more triangle-ish version of the same single opamp LFO.
Well, I've tried to be constructive, you just don't get it..
Sorry Tomas, I really don't want to get into a thing about it - life is too short to be upset over effects pedals! I appreciate your suggestion to remove the buffer - in fact as I noted above I had already built it without that, so I agree about that.
I take your point about the triangle LFO, and in fact I already had the same idea, which you know from what you quoted. The simple answer is that, with this kind of circuit, if there needs to be a tradeoff between simplicity and perfection, I will go for simplicity. The reason is that people have already made endless regular old phasers. There is no point in me trying to design a regular phaser with the same number of parts, because I am simply not a talented electrical engineer in the professional sense.
If I wasn't going to make the tradeoff for simplicity, then there would be no point in designing this circuit at all, or really almost any new non-digital effects since the 70s. They basically already got it "right" so if that's our only design consideration, I know I have nothing to add.
Your philosophy about effects design seems to differ from mine, and that's cool. There's room for both. :)
Come on guys, don't get irritated. I think Tomas was just saying not to be so obsessed with simplicity that you miss out on a better sounding design for the want of a few components. He wasn't having a go at you Taylor.
Props for this and the other new designs Taylor, I'm enjoying following the development of them. It's easier for a hack like myself to understand what's going on when things are broken down into relatively simple circuit snippets.
Ok, I see I was too slow posting, things are already under control :icon_biggrin:
Quote from: Taylor on February 26, 2011, 05:43:40 PM
..in fact as I noted above..
I have missed that actually, sorry..
Quote from: Skruffyhound on February 26, 2011, 05:50:43 PM
I think Tomas was just saying not to be so obsessed with simplicity that you miss out on a better sounding design for the want of a few components. He wasn't having a go at you Taylor.
Exactly, thanks - my advice would be to start with proper functional circuit and try to simplify it step-by-step later checking what's the price for each given simplification, not the other way.
And especially with sweep signal I'd be really careful.
T.
Quote from: Thomeeque on February 26, 2011, 06:10:53 PM
Exactly, thanks - my advice would be to start with proper functional circuit and try to simplify it step-by-step later checking what's the price for each given simplification, not the other way.
I politely disagree; I think that's a way to get a stripped down version of an existing sound, not an altogether new sound that comes from approaching a problem from scratch in a new way.
Blah blah blah. When do we get sound samples? :)
Quote from: Hides-His-Eyes on February 26, 2011, 06:45:06 PM
Quote from: Thomeeque on February 26, 2011, 06:10:53 PM
Exactly, thanks - my advice would be to start with proper functional circuit and try to simplify it step-by-step later checking what's the price for each given simplification, not the other way.
I politely disagree; I think that's a way to get a stripped down version of an existing sound, not an altogether new sound that comes from approaching a problem from scratch in a new way.
Sorry, there should have been "to start with
designing proper functional circuit and try to simplify it step-by-step later"..
This looks very cool and very easy to expand out - I'm assuming one could add more and more stages? I'm also interested in the idea of adding an expression pedal input to it - I've gotta use that Moog EP-2 for something! But this looks great and would probably be good for a newbie like me to build and expand upon at later stages.
I'd be interested to here some audio clips, though. :P
Quote from: auden100 on February 26, 2011, 07:48:17 PM
Blah blah blah. When do we get sound samples? :)
I don't like posting +1 posts, but I'll make an exception this time.
+1 ;)
breadboard it!... :icon_wink:
Quote from: deadastronaut on February 27, 2011, 11:36:34 AM
breadboard it!... :icon_wink:
Well, I did go to the post office today to pick up my Futurlec order, including a breadboard and quite a few other things, but it currently holds the beginning of a Tiny trem and I think I'll get that sorted out first. I'll let this one settle down first, but a sound sample would be nice :)
Quote from: Jazznoise on February 27, 2011, 10:43:28 AM
This looks very cool and very easy to expand out - I'm assuming one could add more and more stages? I'm also interested in the idea of adding an expression pedal input to it - I've gotta use that Moog EP-2 for something! But this looks great and would probably be good for a newbie like me to build and expand upon at later stages.
I'd be interested to here some audio clips, though. :P
Yes, more stages should be fairly easy. An expression pedal for manual sweep or LFO speed is simple too.
Sound clips and updated schem soon.
Quote from: Perrow on February 27, 2011, 01:49:16 PM
Quote from: deadastronaut on February 27, 2011, 11:36:34 AM
breadboard it!... :icon_wink:
Well, I did go to the post office today to pick up my Futurlec order, including a breadboard and quite a few other things, but it currently holds the beginning of a Tiny trem and I think I'll get that sorted out first. I'll let this one settle down first, but a sound sample would be nice :)
yeah know what you mean..ive got 3 full breads at the moment.. you can never have too many boards...but there's always never enough!... ::)
edit:
aha..clips soon... :icon_wink:
I see no clean mix here, is that unnecessary in this kind of circuit?
Right. In a regular phaser, the effect is obtained by creating a frequency-dependent delay, then combining the delayed signal with the clean signal to create cancellations which result in one or more notch filters.
In this circuit, we bypass all that and go straight to the filters so the blend is not needed. In this case they are peaking rather than notch filters, but I'll do a slightly bigger version that will allow for peaking or notching.
Wow this is cool !!
I'm all over this one, Taylor. :)
Ok, here's the updated version. I have not yet built this, but I tweaked the LFO a bit in simulation to get it more triangular, and also to make it capable of going both faster and slower than before. I'll never pass up an opportunity to make anything into a pseudo-ring modulator.
Besides the LFO, I've added more diodes to hopefully allow more headroom, and added a switch to switch between traditional sound and bi-phase, where the 2 filters are sweeping in different directions.
Of course, this leaves us with an unused opamp, and I've got all kinds of ideas for utilizing that. Also, a depth pot would be easy to add, and a resonance control for the filters would be very useful but requires a dual gang pot. But before we add those goodies I'd like to test out the basic circuit. If anybody has time to breadboard it I'd be interested to know what you think.
(http://musicpcb.com/wp-content/gallery/circuits/brazen-prophet-rev2.jpg)
v1 vero is ready. I have to correct it to v2 specs as soon as someone says it's ok.
btw, cv pot is like a starve control!?
No, cv pot is like a manual sweep control or LFO bias. It could be a trimpot as in the Phase 90, but I find i useful as an external control.
Ooo... I'm keeping my eye on this one. Sounds like it might be pretty cool. :icon_razz:
Ok, I built the above circuit, and it does work, with a caveat:
Although the additional LEDs increased headroom, they also changed the frequency range too much for my liking. It seems limited rather than expanded. I used blue LEDs rather than the spec'd red, which have a larger forward voltage. I would now recommend 2 red LEDs per LED chain.
I also added the resonance control. Notice the 2k resistor between the lower opamp's output and the 4n7 cap. If you replace that with a pot (10k-20k would probably be ideal) you have a resonance control for that filter, analogous to a feedback control on a standard phaser. The same can be done between the upper opamp's output and the 1n cap currently connected to it. You can use a dual gang pot to control both filters at the same time.
Now, what to do with the unused opamp:
1. make a real triangle integrator LFO (I know a certain someone who likes that idea).
2. add another filter stage to emulate a 6-stage phaser.
3. I think if we put the whole audio path in a negative feedback loop, we get notches instead of peaks. We don't see this opamp-nesting done in stompboxes much, but I don't think it's a crazy thing to do.
Edit:
Yep, number 3 does work.
Good work Taylor, I'm looking forward to breadboarding this.
The "biphase" thing is not really working the way I'd like. the 2 sweeps need to be offset somehow for them to be sweeping through useful ranges at the same time. I tried a single LED for the LED chain going to 9v, with a series trimpot. Tweaking this got it more in range, but not to my satisfaction.
The trick might be something like this: connect the standard CV pot directly to the lower filter, but in the upper filter, connect it within the right side LED chain. Then have the "biphase" trimpot connected within the other LED chain. That way the CV pot controls both filters when in traditional mode, but it's switched out of the upper filter in biphase mode, and the trimpot takes over LFO bias duty only for the upper filter. This should help focus the frequency centers for the opposing filters to a range that sounds best.
Quote from: Taylor on March 02, 2011, 09:13:04 PM
3. I think if we put the whole audio path in a negative feedback loop, we get notches instead of peaks. We don't see this opamp-nesting done in stompboxes much, but I don't think it's a crazy thing to do.
Edit:
Yep, number 3 does work.
On my modular synth I can put a signal through a lowpass filter, then mix (in the correct ratios) the output of the lowpass filter with the original signal, ensuring first that the output of the filter is inverted with respect to the original signal. My lowpass filter is now effectively acting as a highpass filter, courtesy of phase cancellation.
Yes, originally I had planned to do it as a switch hitter amp for the clean signal, which would allow to pan from inverting to non-inverting mix, with the idea that it would do as you say due to cancellations. For some reason I started to doubt that this would work and decided to do the thing with nesting the filters in the NFB loop, but actually the switch hitter is probably a solid idea. Thanks for confirming that that would work.
Overall I'm pretty happy with this idea. It's easily expanded in a few different ways. Right now the LFO works very well in medium to high speeds, but not very good at slow speeds. An integrator triangle generator would be better for those who like slow phasers; the inverting trick would be good to get actual notches, which will sound more authentically phaser; and using the last opamp for an additional phaser would be great for those who like more extreme sounds. I've yet to try the CMOS version I was thinking up, but if it works I'll do a version which is analogous to a 12-stage phaser, with proper triangle LFO, and true notches, all with just 2 chips and passives. ;D
reading what you wrote, it seems that v1 works better...that is verified, isn't it!? Maybe I shall try it without the buffer at the input...
Well... the first one, as drawn will not be too usable because it doesn't incorporate the CV pot, the distortion is a little too much because it only has a single LED per filter, and the LFO is a little too limited. I would recommend building according to the rev2 schematic, except for now I'd leave out the biphase switch and only use 2 LEDs per filter. It sounds quite good like this.
However, I expect there to be some tweaks and a few different versions with different goals over the next week or so. This is an idea in development. So it might make more sense to just breadboard or perf for right now if you want to check it out. If you're looking more for a finalized effect to go right on your pedalboard, probably a good idea to hang on a bit until we get somewhere that we can call "done".
right on, I'll vero the "final version" when done.
Although Leds and transistor are much more consistent than jfets, wouldnt you just need diode matching for this circuit anyway?
I think that led matching would make the notches equally spaced trough the sweep of the LFO and both the same depth.
This would start to sound less like two low pass filter and more like a real phaser.
Im not really sure about this, as i didnt breadboard the circuit yet. Just a tough.
Also, at slow speeds: why doesnt sound good? It doesnt have enough depth? Or is it just the oscillator not having enough "linear ramping"?
No diode matching is necessary. You're not going to get consistent depth through the sweep with a circuit like this, regardless of matching. The idea here is "quick and dirty" so there will be some tradeoffs - that said, I think it sounds quite nice.
At slow speeds, the sweep both looses depth and, as you say, doesn't have linear ramping - it becomes somewhere between a sine and pulse wave.
For those interested in potentially building this, what sort of configuration would you be most into? I have thought of a few different versions:
1) single chip, has a better LFO for better slow sounds
2) single chip, allows for peaks or notch filters (the current one only does peaks - notches are the more "classic" phaser sound)
3) single chip, same as the current one but with an extra filter for more "wet" or extreme sound
4) two chips, has all of the above
I can make schematics showing all of them, but I figure it might confuse people to have to decide between 4 different schematics/layouts. Give me some feedback on what most interests you among these. Here is a little poll (http://poll.pollcode.com/fPBs) for you to choose what interests you most.
I voted for option 4 (cool to vote for something you actually care about), but I do think there should also be a 'starter' version - in keeping with the 'super simple' concept, probably option 2.
I'll probably do the full featured one (option 4) and another one, because as you say I really like the idea of a single chip phaser.
yep. both options sound good to me.....simple is nice!...especially for breadboarding and learning!. nice one taylor, look forward to hearing it/them..
Quote from: Taylor on March 10, 2011, 02:41:53 AM
..I really like the idea of a single chip phaser.
Well, you could use quad op-amp chip, but it would be probably dirty cheating, right? :icon_mrgreen: T.
I'm already using a quad. If you're going to troll, please pay closer attention. :icon_lol:
Quote from: Taylor on March 10, 2011, 05:26:44 AM
I'm already using a quad. If you're going to troll, please pay closer attention. :icon_lol:
O-oh, sorry, that was really stupid :icon_redface:
OK, when I'm at it ;) may I yet troll about sound samples, are there any already, please?
Thanks, T.
My suspiscion is that this one isn't going to give you that 'well fed' feeling that a multi stage phaser can nourish - from what I can gather, this implementation is playing the differing phase shift of two filters off against one another ....innovative yes, pleasing...hmmm. I look forward to the sound clips!
On a more positive note, I can save you one cap in your build - the output DC level wrt both those final opamps is sitting at the same level.......and anyways, there's 2k sitting in between them which caters for any slight DC imbalances - the two caps on their output pins can be binned ....one cap is suffice on the final output.
Quote from: Taylor on March 10, 2011, 02:04:33 AM
1) single chip, has a better LFO for better slow sounds
2) single chip, allows for peaks or notch filters (the current one only does peaks - notches are the more "classic" phaser sound)
3) single chip, same as the current one but with an extra filter for more "wet" or extreme sound
4) two chips, has all of the above
I want to vote, but first I have to understand better! I'd say one quad, but with the most options it can give - so it is #3!?
Does it require a lot more components compared to #2 or #1?!
I'd like to see a circuit simple to build but with a good range of sounds, from subtle to mayhem! :icon_twisted:
To say, Little Angel meets the above requirements for me. Simple but good sounding.
Quote from: Gurner on March 10, 2011, 07:16:02 AMfrom what I can gather, this implementation is playing the differing phase shift of two filters off against one another ....innovative yes, pleasing...hmmm. I look forward to the sound clips!
Well, no, there's really nothing to do with phasing happening here. It's simply 2 filters, tuned to different frequency ranges, swept by the same LFO. I don't claim it to be particularly innovative, but I do think it's pretty sweet for such a small, no-fuss circuit.
Thanks for the suggestion on the output caps - you're quite right. :)
I do think this is innovative - don't do yourself down. I've played around with single stages of that kinda filter and they sound cool. The peaks/notches on a 3 stage version will certainly be sonically very interesting. I'll certainly be building the complex version of this - really looking forward to it reaching fruition.
Quote from: Taylor on March 10, 2011, 03:28:54 PM
Quote from: Gurner on March 10, 2011, 07:16:02 AMfrom what I can gather, this implementation is playing the differing phase shift of two filters off against one another ....innovative yes, pleasing...hmmm. I look forward to the sound clips!
Well, no, there's really nothing to do with phasing happening here.
There must be else you wouldn't get the phasing effect! (I'm meaning phasing here to be audio signals being summed with differing phases)
There'll be a different phase response appertaining to each filter, isn't it a case of by applying the CV, the phase slope will alter with the LFO - & since the signal output of these different 'phase slopes' are being summed you're gonna have phase notches?
No, notice I called this a "fake" phaser. If you read the first post I think it will be clear. These are not allpass filters, they are bandpass filters. By putting them in an opamp's NFB loop, we get notches, but it has nthing to do with the 2 filters' phases being different or anything to do with summing - the reason there are two is not to get any cancellation between them, but to get 2 peaks or notches instead of one, as you'd have in a 4-stage phaser.
Yeah Gurner's right there is phasing going on. Here's a simulator plot of one stage
(http://www.eskimo.plus.com/fxstuff/taylorsphaser.jpg)
The trace with the peak is amplitude and the other trace is phase with the scale on the right. As the peak sweeps up and down the frequency range the phase plot roughly follows it with the phase inverting at around the frequency of the resonant peak. The shape of the phase plot changes somewhat as the resonant peak moves through out the frequency range but in any position different frequencies have different phases.
Mix the outputs of two of them together and a given frequency from each will have a different phase, this will change as the filters sweep, so you've got phasing. Especially in the range between the two peaks you get cancellation giving a notch.
EDIT: this wasn't in response to you Taylor, you posted while I was typing.
Quote from: Taylor on March 10, 2011, 04:31:20 PM
No, notice I called this a "fake" phaser. If you read the first post I think it will be clear. These are not allpass filters, they are bandpass filters. By putting them in an opamp's NFB loop, we get notches, but it has nthing to do with the 2 filters' phases being different or anything to do with summing - the reason there are two is not to get any cancellation between them, but to get 2 peaks or notches instead of one, as you'd have in a 4-stage phaser.
I'd already read your first post (before making my prior posts)....I didn't even mention an allpass filter as clearly that's not what's going on here - I think you might be misinterpreting my reference to phase shift as thinking I mean 'traditional' (in the stompbox sense of the word) phasers....I don't, you may call it a fake phaser, but you're essentially playing two different phased signals off against each other (ie summing signals with phases that aren't 100% in sync), which will yield an ahem 'phased' sound.
Those 'peaks' have come about by using two separate (animated) filters - & when filtering you get a phase shift. So, you may be using peak filters ...but IMHO the outcome is you'll get differing phase shifts at different frquencies when putting a signal through each filters, thereby creating moving notches when they're summed.
Hmm, well in all respects I'm always willing to concede that I could be wrong. :) I know you know more about EE than I, I just often assume that people are misunderstanding me as that is often the case. (Poor communication skills on my part.)
I plan to put this through real life frequency analysis once I build a final version, so we can see exactly what's going on here. If you're right, then my design is much more clever than I'd thought! :P :D
In general, all filters shift phase. Most filters give a frequency-dependent amplitude. The exception is the all pass filter, which is probably why it's the one people often call a "phase shifter."
In fact, the only reason you hear notches in a "real phaser" is because of the additional process of mixing in the dry signal. Without the dry signal, you get vibrato with no tonal change.
The only thing I would suspect this parallel-bandpass topology incapable of... is vibrato.
Taylor, one really interesting thing to do with this kind of filter is to sweep the two filters with LFOs that are not the same.
You might want to read the stuff at GEOFEX about vocal tract filtering.
Quote from: R.G. on March 11, 2011, 10:48:10 AM
Taylor, one really interesting thing to do with this kind of filter is to sweep the two filters with LFOs that are not the same.
You might want to read the stuff at GEOFEX about vocal tract filtering.
Right, I was hoping to do that by inverting the diode orientation on one of the filters, to get one going down while the other goes up. This is cheap because it can use the same LFO with no extra opamp needed for inversion, but I haven't gotten it to sound very good yet. On the verge of some "yoyoyoyoyoyeeee" type sounds ala the Ludwig Phase II, but still kind of lacking due to the compromise of the design.
What might be nice would be to try a quadrature oscillator, and use the sin and cos outputs for the different filters. That gets the job done with only 2 opamps while having them staggered instead of just in anti-phase. Totally independent LFOs would be fun too, as you say.
The best part about this idea, to me, is that it's really "immediate" so it's possible to implement all kind of interesting ideas like the formant filtering you're talking about, changing the peak spacing is easy, etc. I would expect there to be tons of versions of this as time goes on. I kind of wish I had picked a less pompous name for this one. :icon_lol: It was a take-off of "brazen head" (http://en.wikipedia.org/wiki/Brazen_Head), ancient experiments into speech synthesis thought to be prophets.
Taylor, do you have a current/working version of this to post? I wouldn't mind having a go at this on the breadboard. :)
pompous or not...lets hear it...especially the yooooooowwwweeeeee...great description btw.... :icon_mrgreen:
Quote from: Taylor on March 11, 2011, 01:57:27 PM
Right, I was hoping to do that by inverting the diode orientation on one of the filters, to get one going down while the other goes up. This is cheap because it can use the same LFO with no extra opamp needed for inversion, but I haven't gotten it to sound very good yet.
The problem with that is that the diode-resistance setup needs to have the diode sitting at nearly its active forward drop at DC to be off, and only moves about 0.2V to fully on. You can't just invert the diode on one side without also trying to invert the DC bias on the diode. If you do, you get a big dead spot where neither one is doing much. Also, you have to have a CV which can both source and sink current, probably at the same time.
I think you might be able to get something like this to work if you used each diode in series in the middle of two resistors, one to V+ and one to ground, bypassed the center to ground with a BFC and fed the control voltage to the middle point. That forces you to put the resistor ends of the diodes to the T filter points, perhaps with a couple more BFCs, and worry about the interactions of the resistances. It would be a tight fit for whether this would be more or less parts and PCB space than adding another opamp. An opamp, on average, is four pins, and needs another four pins of resistors to do the inversion. It's one of those "what are the costs" questions I referred to in "PCB Layout for Musical Effects".
As a meta-discussion on LFOs, I've found that LFOs are quite picky. You have to know and correspond to the necessary input range of the thing being modulated, in both AC and DC terms. The variety of things being modulated make this a design task each time, not usually something intuitive.
QuoteWhat might be nice would be to try a quadrature oscillator, and use the sin and cos outputs for the different filters. That gets the job done with only 2 opamps while having them staggered instead of just in anti-phase. Totally independent LFOs would be fun too, as you say.
I built something similar with a dual OTA for each filter and a PWM'd control voltage from a PIC back when I was messing with the Ludwig. It was fun. An 8-pin PIC can supply two PWM'd outputs and have up to four pot or switch inputs on the other pins. Easy enough to get quadrature without the analog messing.
A three-phase integrator LFO locked at 60 degrees is interesting too, and can be done with CMOS inverters.
http://www.diystompboxes.com/smfforum/index.php?topic=85658.0
I made a layout a while back for the Q&D VCF. I built mine into a busted WAH pedal and also added in a simple LFO with Triangle/Square modes. It sounds really nice and I actually have used it a number of times in recordings. I'm thinking an envelope trigger on this would work really well, tapping into that CV source. Maybe an all in one pedal with mode selector (LFO, Phaser, Envelope trigger) is in order here :icon_biggrin:
Quote from: culturejam on March 14, 2011, 08:28:13 AM
Taylor, do you have a current/working version of this to post? I wouldn't mind having a go at this on the breadboard. :)
I was hoping to get some time to draw up some more final versions, but I might not be able to this week. If you look at my last schematic, rev2, that will work pretty well with these changes:
-make each LED chain 2 LEDs instead of 3. Try red or blue (red lower forward voltage=more distortion, maybe better sweep range)
-forget the trad/biphase switch for now. Connect upper LED chain to ground just like the lower one
Some things to play around with:
-the 1n/4n7 caps that make the "T" in each filter tune the base frequency.
-add the resonance control I mentioned in reply #44. Makes it much more versatile.
i know this is really old, but i love this idea! does anyone have sound clips? schematics? updates?
i especially like the idea of each notch being swept by a separate lfo.
I don't think I developed the idea much beyond whatever I posted in this thread, but it looks like the image links are dead, so I'll try to track down the schematics on one of my old computers.
OK, here's what I found. Disclaimer: this was a back-of-a-napkin sketch from a time when I knew way less about electronics. I would love to see others develop this idea more - for example, I'd definitely put an inverting opamp mixer at the end, among other things.
(https://i.imgur.com/xMTDlo5.jpg)
that's great! thanks for finding that