EHX Hot Tubes - Analysis?

[fryingpan]

Hi,

I'd like somebody to help me analyse how the EHX Hot Tubes works, it's an easy pedal but I have a few headscratchers.

This is the schematic:



So, we have a pulldown resistor and a coupling cap straight away. The two 220K from V+ to ground serve to bring the opamp's input to ½V+, but what's the purpose of that 1uF polarised cap to ground? And the shorted 120K resistor? What's the point?

Then, we have a 180pF cap to ground, which should pass highs (so in practice it's a low-pass), what resistance do I couple it with to determine the cutoff frequency?

Now, disregarding the direct out, the first opamp seems to be in buffer configuration (the cap couples the direct out output, so we can disregard it too), then we have an inverting opamp whose gain is around 3...? (75K/20K) and the cap ends up working a low pass (in the high kHz?). What's the point of the other 20K resistor from the + terminal to the first opamp's feedback loop? Isn't the gain a bit low?

Next come the inverters. I don't really know how inverters work, so what's the purpose of the feedback resistors and the resistors to ground? (I have the suspect that the resistors to ground act as pull-down resistors). After two inverting stages there is a tone section that is similar to a Big Muff tone section, and it appears to be a low pass with a ~230Hz cutoff frequency and a highpass over 1300Hz, and you just mix the two with the tone knob, the insertion loss is compensated by another inverter. (On the bypass branch it's basically the same, without the tone section, so it's also a boost somehow). I'm not sure about the purpose of the large 4.7uF caps and the 47K to ground.

By what I see, it shouldn't be an issue to power the unit at, say, 15 or maybe 18V to have greater headroom by the CMOS chip (apparently the 4049 chip should withstand up to 20V). The Hot Tubes can also be a bit dark sometimes so it could be beneficial to add pre-emphasis/de-emphasis I suppose, and that shouldn't be hard to do (smaller coupling caps and a corresponding bass boost close to the output).

Did I get it right?

[Plexi]

The 4.7uf caps are coupling, and the 47k resistor to ground is to tame the output.

[fryingpan]

Of course, the inverter's output will present a DC offset. Why tame the output though? Wouldn't the volume pot suffice?

[fryingpan]

And did I get the filtering or have I made a few mistakes?

[Scruffie]

The 47ks are pull downs to prevent switch pop with the tone bypass switch.

The 120k is a schematic error, it shouldn't be shorted, that's just a reference chain, splitting one resistor in to two lets you stabilize the reference without having a big ol' cap on the opamp input.

The 20k/4u7 to the non-inverting input provide a reference voltage for the next opamp provided by the preceding opamp.

Inverting CMOS self reference to half supply but otherwise work like an inverting opamp albeit with different characteristics and inherent limitations.

Giving CMOS more or less voltages does more than affect headroom, it changes the gain, clipping onset, bandwidth... you might like it, you might not but generally distortions that use CMOS go for a lower voltage for the clipping characteristics that provides.

[fryingpan]

The 120k is a schematic error, it shouldn't be shorted, that's just a reference chain, splitting one resistor in to two lets you stabilize the reference without having a big ol' cap on the opamp input.

How so? And is the reference voltage not ½V+? That is, the is the voltage divider now composed of 220K (reference) 120K + 220K? Or does (should, due to the mistake?) that only mean the 220K to ground resistor is divided into 120K + 100K?

[Scruffie]

I had a quick look, the schematic is wrong, there's just a standard 220k voltage divider with a 120k reference resistor, not a bias chain.

[fryingpan]

I still don't understand what you mean as a reference resistor (sorry!).

[Scruffie]

It should be two 220k resistors, one to 9V, one to ground with a cap to ground from the middle and a 120k from the middle to the non-inverting input.

[fryingpan]

Oh, now I get it, basically what appears to be a short is an "overlapping, not connecting" trace (in the schematic anyway). All right so what's the purpose of the cap and resistor? It would appear that there is no voltage fall over the 120K resistor (the cap at the beginning is an open circuit for DC), the other connection to ground is a 180pF cap so another open circuit, the opamp input is high impedance. I understand the purpose of the small cap (I suppose to filter RF, although it would be better form to pair it with an appropriate resistor), not of the other (the big one basically).

[idy]

The question I've always tripped on for these invertor designs is how were the resistor pairs ("feedback" and "to ground") determined for the two stages. The first has 150K "fb" and 1.8M to ground. The second has 200K and 2.2M. Different gain?

[fryingpan]

Oh, now I get it, basically what appears to be a short is an "overlapping, not connecting" trace (in the schematic anyway). All right so what's the purpose of the cap and resistor? It would appear that there is no voltage fall over the 120K resistor (the cap at the beginning is an open circuit for DC), the other connection to ground is a 180pF cap so another open circuit, the opamp input is high impedance. I understand the purpose of the small cap (I suppose to filter RF, although it would be better form to pair it with an appropriate resistor), not of the other (the big one basically).

Maybe I've got it now. The 1uF cap is so large that basically it is a short circuit to ground for any AC. So basically the input impedance of the pedal is 120K, as it's in parallel with a 1M pull-down resistor and the opamp's input impedance which is usually very high (so the overall input impedance is a bit low really for a guitar signal). Whereas it's going to be an open circuit for DC and thus the reference voltage is set.

[PRR]

[PRR]

> 150K "fb" and 1.8M to ground. The second has 200K and 2.2M.

The to-ground resistor has almost no effect on audio gain. It does affect DC level.

As Scruffie said, CMOS with NFB tends to bias to half-supply. But P and N MOSFETs are not exact complements. Also these stages are, at high level, serious distorters. And we know that in music, asymmetrical distortion is often interesting. The 2Meg+200K pair throws about 10% off-"center" into the DC. I do not know if that makes it more-centered or less-centered (and it can depend a lot on who and when the CMOS was made). If you are truly curious, build one, then try 470K-10Meg to ground or to V+, try all possibilities.

Classic CMOS was rated 18V, not "20V". In pure logic you may get away with 20V for years. In Linear Mode the dissipation rises much faster than the voltage and you will probably melt by 15V. Anyway why do you want huge voltages? Distortion is about banging your car against the guardrails. It is easier to dent your car in a narrow tunnel than a wide expressway.

[fryingpan]

Just thinking at some modifications in order to use the basic circuit as a starting point for a simple preamp (so more headroom, better input and output impedance, better bandwidth, etc.). Think Beta Bass but simpler and less "heavy" sounding.

[Plexi]

Idk why, but I've always see this design as a Red Llama pushed by some kind of op-amp od. And switchable Big Miff tone stack.

[fryingpan]

> 150K "fb" and 1.8M to ground. The second has 200K and 2.2M.

The to-ground resistor has almost no effect on audio gain. It does affect DC level.

As Scruffie said, CMOS with NFB tends to bias to half-supply. But P and N MOSFETs are not exact complements. Also these stages are, at high level, serious distorters. And we know that in music, asymmetrical distortion is often interesting. The 2Meg+200K pair throws about 10% off-"center" into the DC. I do not know if that makes it more-centered or less-centered (and it can depend a lot on who and when the CMOS was made). If you are truly curious, build one, then try 470K-10Meg to ground or to V+, try all possibilities.

I gather that CMOS inverters behave like (imperfect) inverting opamps, and gain is determined by Rfb/Rin. I don't see an input resistor on the second inverter, and the input resistor on the first inverter is basically the 2M linear overdrive pot plus the 8.2k resistor. So basically at Rin > Rfb we have overall attenuation (plausible) and at minimum Rin we have ~20x gain (~26dB) Does the output impedance of the first inverter determine the gain of the second one? (Apparently CMOS inverters have high output impedance when used linearly, and 1M input impedance is recommended for the following stage, so can I assume an output impedance around 10K (as if it were a guitar signal) and so another 20x gain on the second stage? And what's the point of the shunt resistors to ground? Isn't voltage stabilised at around ½Vdd (or whatever) by the feedback? Is the ground resistor aimed at replicating the virtual short circuit between the + and - terminals of an opamp (with the + terminal shorted to ground)?

[PRR]

CMOS inverters used "linear" are really poor "op-amps". The gain is not high. The gain depends way too much on supply voltage. Setting a high resistor ratio may not get a lot of gain. Also the self-hiss level is high.

Unless you specifically wish to explore CMOS distortion, I'd be looking at some less idiosyncratic preamp. 

[fryingpan]

Well, you don't have to employ CMOS inverters exclusively. The Beta Lead/Bass amps for example amplify the guitar signal with opamps initially and use CMOSs as tone shapers basically, you could do something similar with FETs or BJTs too. I'm exploring a design based on FETs for the first gain stage, either FETs or opamps for pre/de-emphasis and CMOS chips for tone shaping (and either a passive tone stack or active filters for EQ). If I ever get round to it.

[reddesert]

Idk why, but I've always see this design as a Red Llama pushed by some kind of op-amp od. And switchable Big Miff tone stack.

Both the Red Llama and the Hot Tubes are derivatives of Craig Anderton's Tube Sound Fuzz, which apparently came in both op-amp-driven and regular flavors: http://www.diystompboxes.com/smfforum/index.php?topic=67420.msg537052#msg537052 although the regular flavor is more common.

To fryingpan's question, I don't think the voltage is stabilized by the feedback - these are inverters so they could drive the output to zero or Vdd easily. There is a blocking cap on the inputs of the first two inverters so the DC level is not set by the previous stage. I think the shunt resistor of 1.8M or 2.2M pulls the input to ground in the absence of signal, preventing oscillation or noise. If there is signal, the impedance of the signal and of the feedback resistor is lower than the shunt resistor, so it overcomes the shunt to ground.