Minimum # of components for signal path of trem

[harmless]

I'm working on a tremolo that uses a microcontroller to generate the LFO with a vactrol to connect to the signal path.  This is my first circuit 'design' (if you can call such a simple thing a design) so I'm looking for recommendations about the signal path.  At one end, I see things like the tiny tremolo, that basically connects the input to the ldr and directly to the output.  At more of the opposite end I've seen a PIC based tremolo that has an op amp buffer at both the input and output (with a gain knob on the output).

I've tried to do research on when or when not to use buffers, but everything I've found seems to be about buffers vs true bypass (these things also talk about how expensive dpdt and 3pdt switches are which doesn't seem to be the case anymore).  Since the microcontroller can easily generate the indicator LED, I can get true bypass with a DPDT quite easily.  Given that the entirety of my signal path is basically a variable resistor, do I still need an input buffer to present a high enough impedance when the circuit is on?

If the signal path is just a resistor, basically, do I need blocking caps?  I won't have any additional DC voltage to leak, but I guess these could protect against badly behaved components elsewhere in the chain.

I've also seen a lot of circuits with 1M resistors to ground off of input and output - this is to guard against pops when switching, right?

If you've read this far, thanks.  I guess my basic question is: I'm leaning towards my signal path being just the vactrol plus blocking caps and resistors to ground at each end.  Is there any reason for me to do more than this?  I'm building this for me to use rather than sell, so while I'd prefer to have a quality product, I don't need to aim for commercial bullet-proof.

[.Mike]

I'm working on a tremolo that uses a microcontroller to generate the LFO with a vactrol to connect to the signal path.  This is my first circuit 'design' (if you can call such a simple thing a design) so I'm looking for recommendations about the signal path.  At one end, I see things like the tiny tremolo, that basically connects the input to the ldr and directly to the output.  At more of the opposite end I've seen a PIC based tremolo that has an op amp buffer at both the input and output (with a gain knob on the output).

I've tried to do research on when or when not to use buffers, but everything I've found seems to be about buffers vs true bypass (these things also talk about how expensive dpdt and 3pdt switches are which doesn't seem to be the case anymore).  Since the microcontroller can easily generate the indicator LED, I can get true bypass with a DPDT quite easily.  Given that the entirety of my signal path is basically a variable resistor, do I still need an input buffer to present a high enough impedance when the circuit is on?

If the signal path is just a resistor, basically, do I need blocking caps?  I won't have any additional DC voltage to leak, but I guess these could protect against badly behaved components elsewhere in the chain.

I've also seen a lot of circuits with 1M resistors to ground off of input and output - this is to guard against pops when switching, right?

If you've read this far, thanks.  I guess my basic question is: I'm leaning towards my signal path being just the vactrol plus blocking caps and resistors to ground at each end.  Is there any reason for me to do more than this?  I'm building this for me to use rather than sell, so while I'd prefer to have a quality product, I don't need to aim for commercial bullet-proof.

In my opinion, you need gain for a tremolo. Think about it. When at 100%, the volume is whatever your guitar puts out. At 0%, there is no volume. Quickly turning the signal from 100% to 0% to 100% to 0% etc. results in a perceived loss in volume. Additionally, having a resistor in series with your signal that varies from ~500-Ohms to 100K-Ohms is going to monkey with whatever is next in your chain.

I would suggest you build a few tremolos (or other effects) from known good designs before dipping your toes in the design water. I built for a year and a half before I hacked together my PIC Trem, and I was in waaaaaaaaaaaaaaaay over my head. Without help from a number of people, it would have been scrapped.

Good luck!

Mike

[harmless]

Thanks for the feedback.  The perceived loss of volume thing definitely makes sense (to stretch a goofy metaphor it seems like a similar phenomenon to PWM to produce analog values).

I say 'design' but what I really mean is find all the tremolo schematics I can, attempt to understand what they are doing and the tradeoffs they are making, and then make my own decisions based on that input (as opposed to starting with a blank sheet of paper and some equations).  I'm still breadboarding anyway and I have some opamps so I guess I should build each of these options and see what they sound like and how they behave.

[R.G.]

I've tried to do research on when or when not to use buffers,

Use buffers any time you
(1) need to drive any load which gets under about 1M resistive to ground when driven by a guitar. For a variable-resistance tremolo, this means all the time.
(2) care about preserving the treble content of a signal out of a passive, magnetic-pickup guitar.

Given that the entirety of my signal path is basically a variable resistor, do I still need an input buffer to present a high enough impedance when the circuit is on?

Yes. Otherwise, you will get varying treble too. Unless you want that, you need to buffer the guitar signal. This is aside from the tremolo being down-only and suffering the perceived loss of level.

If the signal path is just a resistor, basically, do I need blocking caps?  I won't have any additional DC voltage to leak, but I guess these could protect against badly behaved components elsewhere in the chain.

You know all you need to know about this, from that statement. All resistors? No need for caps, except for those other pesky things which might possibly be in the chain that you have no idea about now.

I've also seen a lot of circuits with 1M resistors to ground off of input and output - this is to guard against pops when switching, right?

Yep.

I'm leaning towards my signal path being just the vactrol plus blocking caps and resistors to ground at each end.  Is there any reason for me to do more than this?

You need the buffer unless you're willing to live with the perhaps-excessive loading.

[PRR]

The trem is NOT just the LDR, not even LDR+gain. The LDR must work-against a resistor.

Consider: Ideal case: perfect zero-impedance source, perfect infinite load. Put an LDR between. For any value of LDR from zero to almost-infinity, there is no loss, this no change-of-loss (tremolo). (If both LDR and load are infinity the calculation bombs; the reality is absurd.)

If load is less than infinite, then it *matters* how non-infinite it is. Looking around I see 5K into some (ill-behaved) pedals and 5Meg in some Ampegs.

LDR can quickly wobble 1K to 100K. In a simple case you work it against a 10K. If you idle the LDR at 10K, you have a steady loss of 2:1. If there's no reserve gain handy, you follow with a 2:1 booster, which also buffers the varying impedance of the LDR+R combo. In some cases impedances are benign and some other knob can turn-up a notch, the signal path is just LDR+R.... but to cover "all" (ha) cases you probably will be buffer/boosting.

[harmless]

Thank you both for the informative replies.  I feel like I have a better handle on this stuff now. I started breadboarding last night and I'm psyched to see how this sounds.

[egasimus]

IMHO just copy a circuit that works and replace the LFO with your own design. There isn't that much of a variety in LDR trems after all.

[harmless]

That's basically what I've ended up doing.  I borrowed the audio path from the PIC tremolo posted here ( http://www.diystompboxes.com/smfforum/index.php?topic=77471.0 ) which borrowed it from the tremulus lune ).  I haven't done any tweaking of values to account for my particular ldr/led yet, though the data sheet for the vtl5c9 I'm using lists a similar on resistance to what he measured (630R vs 500R).

I plugged it all in and it works, with a minor caveat.  In any waveform mode other than square, I get a high pitched whine in time with the lfo.  Like a wee wee wee wee type noise.  The square wave doesn't have that problem and doesn't seem to pop when it turns on (which is a common problem I found when searching for a solution).  Is this likely due to my breadboard being a huge mess of crisscrossing wires of altogether too long lengths?

I tried disconnecting the 2nd LED (the indicator not in the LDR/LED pair) and it had no effect.  I also didn't have the 2 caps and the diode to put into the power filter section (where Vref is generated), but that shouldn't matter, right, since those are there for varying supply issues (and backwards connections in the case of the diode).

Is this something that should be corrected by just building this into an enclosure (with shorter wiring, veroboard for components, etc)?

[Gurner]

In my opinion if anyone is going over to the dark side (micro), then a Vactrol isn't the best choice wrt trems  ......your micro would then be 'controlling' only one half of the necessary signal  divider chain.

What a lot of people seems to miss wrt trems, is that to effect a signal voltage (tremolo effect) needs a potential divider  chain  ...not an ldr on its own (your guitar volume knob is a potential divider, an LDR/Vactrol on its own isn't)  .....yep, to have a potential divider need two lots of series resistances......and a  vactrol, ldr etc provides only one part of the potential divider resistance needed .....you therefore need another fixed resistance in series with your vactrol for the other part ....result = instant signal loss at the junction, man cries  

However if you have your micro control a digital pot instead, things become a lot easier wrt tremolos ......the signal enters the top lug of the digital pot, the other 'outer lug' is your reference voltage (normally 1/2 VCC...ie a pseudo ground) & your digipot wiper is the signal output.....your micro now controls the whole potential divider chain ....from full signal to no signal, with no signal loss.[insert here the sound of a can of beer being cracked open]

Re your extraneous noises....hard to say, but joining analogue with digital (esp with low level audio signals) is fraught with challenges....my first recommendation would be to ensure that the PIC has its own dedicated supply & ground return signal lines back to the circuit power source....don't parallell the pic off your your supply lines feeding your signal chain components.

[.Mike]

The reason I used the vactrol in my little design is because I was taking a ready-made PIC and using it in an application that hadn't been done before. It worked pretty good, and ended up being the basis for the Tap Tempo Tremolo. Fun stuff. Hopefully you will release your design when it is done.

I didn't have any of the whining like you describe on my breadboard. The VCLFO PWM output is at 19.5KHz. I'm not sure what you are using, but is it possible that the PWM frequency is low enough that it could be leaking into the audio? Could you focus your debugging by hooking a pot up to the LED and turning it really fast to see if it is the problem is the turning on of the LED in general vs. the PIC turning it on?



Mike

[harmless]

I'm using an Arduino instead of a PIC. I think I'm reading things correctly when I see that the PWM frequency is approx 490hz.  That's pretty clearly in the audible range.  I'm going to look and see if I can change it.  If that doesn't work I'll try being more careful about keeping the power / gnd lines separated for the arduino and the signal path.

[earthtonesaudio]

Except for make up gain, you can get nearly ideal performance with two LDRs, arranged as a voltage divider, being driven by two separate LEDs (out of phase with each other).  If you're not dogmatic about true bypass you don't even need a mechanical switch for the signal path, to "disengage" the effect, set the LEDs so the LDRs pass the signal with approximately unity gain.

I'm not trying to detract from the good advice given so far.  In this case there is a non-trivial difference between a good design and the minimal design... But, for certain applications, if you know what you're doing, you may be able to get away with it.   

[Gurner]

Except for make up gain, you can get nearly ideal performance with two LDRs, arranged as a voltage divider, being driven by two separate LEDs (out of phase with each other).  If you're not dogmatic about true bypass you don't even need a mechanical switch for the signal path, to "disengage" the effect, set the LEDs so the LDRs pass the signal with approximately unity gain.

I'm not trying to detract from the good advice given so far.  In this case there is a non-trivial difference between a good design and the minimal design... But, for certain applications, if you know what you're doing, you may be able to get away with it.  

Agreed, two vactrols would work...but I seldom see that implementation (it sounds like a bit of grief, vactrols aren't cheap, an led/ldr implementation needs separation from ambient light - more so if two pair are going to be used....higher power consumption, trickier matching, then there's the coding!).

I was just passing comment as a lot of folks see an LDR/Vactrol & think it can work as a guitar pot - it can't ...at least not without something else to work  alongside it (a resistor, another ldr etc)

[harmless]

Woot.  I fixed the noise issue.  It was a wiring mistake.  I was running the signal path off of the 5v from the arduino's internal voltage regulator instead of the 9v from the power source.  I had one line on my breadboard intended for the 9v power and mistakenly connected it directly to the 5v rail on the other side.

I think I'm going to find a source for closed circuit stereo jacks and add expression pedal support before I put this thing in an enclosure.  My software still needs a tiny bit too in that the analog input for all the pots sometimes wiggles back and forth by +/- 1 which causes the tapped tempo to get overridden by the rate pot.

Then I need to do my first vero layout and try to build the thing.

When I'm done I'll definitely post the schematic and the code.

[earthtonesaudio]

@harmless, that sounds great!  Good luck getting the ADC wiggle sorted.  I seem to recall electricdruid had some small code snippet that put in hysteresis, maybe it's in the TAPLFO thread.


In case anyone is interested here is an example of the two-LDR setup I described earlier.  Link
A lot of the other caveats may just fall into place... metal enclosure provides light-tightness, microcontroller provides PWM signal, reference voltage may be readily available, etc.