CD4049UBE Hot Harmonics Weirdness

[Nick C.]

I just boxed up a new build. Switchable 2 mode like ROG Double D, but 1/2 modded Tube Sound Fuzz and 1/2 modded Hot Harmonics.
Here's the problem, by itself the pedal works great, but with any buffered pedal before it Hot Harmonics mode sounds awful with lots of high pitched noise. The version of the curcuit is the simple one
I bumped the 150k to 470k and dropped the 1m5 and 2m2 biasing resistors. Aren't cmos' self biasing?

Anyway I would like to put it on my board after an OD to slam the TSF and I don't have a true by-pass OD. Any ideas on what is going on or a simple fix? Could it be the missing 1m5 or 2m2? How about some series resistance at beginning?

[samhay]

^How about some series resistance at beginning?

That's what I would try first. You can experiment using a pot wired as a variable resistor in series with the input.

[FuzzFanatic71]

I don't know how to fix the problem you describe except to tell you to build a true bypass O.D to put before it. You can never build too many decent overdrive pedals.

[Mark Hammer]

There is no real lowpass filtering in there to tame to high end.  Stick a 330pf-470pf in parallel with the 150k feedback resistor.  Hopefully, that should tame things a little.

[Nick C.]

That's what I would try first. You can experiment using a pot wired as a variable resistor in series with the input.

Yea, I'm going to try this tonight, thanks.

You can never build too many decent overdrive pedals.

This was going to be my last one Also, I like to put wah (which is buffered) before OD.

There is no real lowpass filtering in there to tame to high end.  Stick a 330pf-470pf in parallel with the 150k feedback resistor.  Hopefully, that should tame things a little.

I do have a 100p there and a LP tone pot at the end and there's still too much to tame.

The thing is that it sounds great when driven by just the guitar, but almost squealing when driven by the high current low impedance of a buffer.

[LiLFX]

Try putting the AMZ pickup simulator between your buffered pedal and the input of this.

[Mark Hammer]

I do have a 100p there and a LP tone pot at the end and there's still too much to tame.

100pf is still kinda low.  BY my calculations that is still just a 6db/oct rolloff starting around 10khz.  Seriously, increase the cap value.

[anotherjim]

Nick says he has 470k instead of 150k, so 100pF across it will start to roll off at about 3.3Khz.

Input impedance could well be an issue. With a passive guitar input, the guitars source impedance is a major part of the input impedance, so a buffered low impedance source instead is a completely different game.

Raising the 150k to 470k might not have helped either. Let's assume the original design had 150k for a reason. A CMOS inverter in linear mode only has an open loop gain around 8 to 10, if feedback resistance is too high compared to input impedance, it cannot boost the signal at the output enough to balance the feedback path current with the input path current.

The input resistors to ground are not there for bias as such, they force an offset from the 1/2Vcc bias you get via the feedback resistor. 2 reasons I can think of...
1: The offset makes clipping asymmetrical - which adds even order harmonics.
2: The offset reduces average current consumption. When the inverters are biased to 1/2Vcc, they suffer "Shoot Thru", which is a current path through the complementary pair from Vcc to Ground because they are both turned on in this state. The offset reduces the current by turning one of the pair towards off.
I think those resistors are important!

Adding, say, 4.7k resistor in series with the input might be all that's needed. As suggested by LilFX, a pickup simulator ought to give performance more like a direct guitar input.

[Mark Hammer]

Nick says he has 470k instead of 150k, so 100pF across it will start to roll off at about 3.3Khz.

Input impedance could well be an issue. With a passive guitar input, the guitars source impedance is a major part of the input impedance, so a buffered low impedance source instead is a completely different game.

Raising the 150k to 470k might not have helped either. Let's assume the original design had 150k for a reason. A CMOS inverter in linear mode only has an open loop gain around 8 to 10, if feedback resistance is too high compared to input impedance, it cannot boost the signal at the output enough to balance the feedback path current with the input path current.

The input resistors to ground are not there for bias as such, they force an offset from the 1/2Vcc bias you get via the feedback resistor. 2 reasons I can think of...
1: The offset makes clipping asymmetrical - which adds even order harmonics.
2: The offset reduces average current consumption. When the inverters are biased to 1/2Vcc, they suffer "Shoot Thru", which is a current path through the complementary pair from Vcc to Ground because they are both turned on in this state. The offset reduces the current by turning one of the pair towards off.
I think those resistors are important!

Adding, say, 4.7k resistor in series with the input might be all that's needed. As suggested by LilFX, a pickup simulator ought to give performance more like a direct guitar input.

I guess that's what happens when you try and fit forum browsing into a full-time day job!   

You're right, of course, the rolloff is lower than I first thought.

[snap]

What is: "1/2 TSF & 1/2 HH" ? Is there a full schematic of it? (including indication of which inverter on the chip is which inverter in the schematic drawing?)

[Bill Mountain]

I think it is most definitely the lack of input resistance on the first stage.  The buffered signal has more highs and lows and lower source impedance which will slam the shit out of the front end of the first stage.  With that much gain and little to no filtering you're bound to squeal and fart out.

Add a filtering, a buffer, and series resistance to the circuit and you'll have consistency across many rigs configurations.

[Nick C.]

Input impedance could well be an issue. With a passive guitar input, the guitars source impedance is a major part of the input impedance, so a buffered low impedance source instead is a completely different game.

I think it is most definitely the lack of input resistance on the first stage.  The buffered signal has more highs and lows and lower source impedance which will slam the shit out of the front end of the first stage.  With that much gain and little to no filtering you're bound to squeal and fart out.

Yep and yep!

After reading what Jack Orman said in the AMZ pickup simulator page, "To simulate the response of a guitar pickup, it is often recommended that some resistance should be added in series with the input of the effect. Typically 10k to 15k..." I decided to try the simple series resistor at the input and low and behold about 10k calmed things down nicely. I remember reading that without an input resistor that the gain of the CMOS inverter is derived from source impedance of the previous stage and that these things are very nonlinear, so yea the gain must have been very high and non-linear causing all kinds of weirdness.

The unfortunate byproduct of the 10k at the input is that my gain is down and I've lost the saturation that I liked to begin with, so I'll try to make it up at stage 2 or 3.

Speaking of stage 3, anyone understand how that is working? 20k/220k=.09, so it's reducing gain? Does it add anything? Maybe harmonics?

Thanks for the help from all. The CMOS sure is a strange beast, but it can make some sweet sounds.

[anotherjim]

The 3rd stage has me baffled. It is attenuating but also inverting too, so it looses "true polarity" with a total of 3 inverting stages. I think there is a version of this type of effect with a "Fetzer" jfet input stage which would make it immune to different input impedances and give correct polarity -  but it's name escapes me.

If there's room in the box, you could fit a switch that shorts out your additional input resistor?

[duck_arse]

Nick C - do you have any unused inverters in that ic? if so, you must tie their input pins high or low, doesn't matter which. they may be a contributor to bad behaviour if left unattended.

[Mark Hammer]

Duck's right about that.

As I've mentioned here more times than people would prefer to have heard, I think the best characteristics of invertor-based overdrives come out when one doesn't oblige them to provide ALL of the gain, but keeps their gain modest, and hits them with a harder signal.  Which is why I like to precede them with an op-amp, or similar, gain stage that also includes some tone shaping of its own, to "prepare" the signal pushing the invertors.  Here's a needlessly complex example of that: http://gaussmarkov.net/wordpress/circuits/forty-niner/

[anotherjim]

It's an interesting test to input a triangle wave into an inverter. Watch it on a 'scope as you raise the input level and see the soft clipping make it look more like a sine wave. It will appear to resist becoming a square wave.
So, yes, as Mark just said, cmos inverters are not gain machines, but rather useful wave shapers.

[Nick C.]

Happy to report, all is well. Behaving well with or without a pedal infront of it. I had to rebuild it a few times to get the right gain and tone. Wish I had tried it with a pedal when I was breadboarding it.

I liked the simplicity of this version of Hot Harmonics over the other versions which have an op amp. Why add an op amp when I have all these perfectly good extra amplifiers in one package

[Nick C.]

Here's the final schem

[snap]

I`d be afraid of even more weirdness, building according to that schematic. (including switchpop)

[Mark Hammer]

Simple solution is to insert a 1M resistor between each of the sub-circuit outputs and C7, and use the toggle to bridge the resistor to whatever half you want to use.

I will also note that the two outputs are 180 degrees out of phase with each other, because of the 3-vs-2 invertor structure.  In a great many instances that won't matter in any audible way.  But be aware of it for those odd exceptions.