Ethos - TLE Overdrive

[phasorman]

I also put the LED indicators for the channel select and the active/bypass mode on here.

Note that the diodes D1 and D2 are for op amp protection, *not* clipping, distortion, overdrive, or whatever you want to call it.

The input buffer provides varying amount of gain, depending on which channel is selected and depending on whether the unit is active. In bypass mode, R4A and R4 are disconnected, so the op amp provides unity gain, i.e., it's just a buffer.

In active mode, if the clean channel is selected, we get a low frequency gain of

(1+R5/[R4+R4A]) = 2.25

In active mode, if the overdrive channel is selected, R4A is bypassed, so we get a low frequency gain of 11.

[phasorman]

I'm guessing C50 is 0.022 uF and C40 is 0.022 uF, just based on common capacitor values. Anyone up for

Whoops, sorry, I left off a resistor values: R47 is 1K and R50 is 68K.

If the GAC switch is off, the op amp is just a unity buffer. If it's on, then the op amp provides a gain of 2, which I assume is to compensate for losses in the GAC filter, which is passive.

Perhaps someone could breadboard this, or whip it up in SPICE, or even do a hand analysis, and plot the frequency response?

Does this circuit look familiar to anyone?

[phasorman]

The op amp on the left gets its input from the GAC stage. The op amp on the right gets its input from the op amp on the left, when the unit is in active mode, or the input buffer op amp, when it's in buffer mode. So in bypass mode, the signal runs through two op amps. The output of the op amp on the right, in addition to driving the main output, also drive the speaker simulator, which is always running regardless of whether the unit is in active or bypass mode.

I'm pretty sure the capacitor and resistor in the feedback loop of the op amp on the left must have something to do with the TLE circuit -- it wouldn't make sense to have those parts there otherwise.

I lost track of what TLE wires went to what parts on the switch, so the TLE is a bit of a mystery to me right now.

[phasorman]

Here are the presence circuits for both the clean and the overdrive channels. They get their inputs from the tone stacks, and they drive the GAC circuit. Well, actually that's the case for the clean channel. For the OD channel, the signal passes through a distortion circuit between the tone stack and the GAC. The distortion is performed on the "gooped" part of the PCB which is still a mystery to me.

Ooops, I forgot to write where R44 gets its input. See the schematic with the clean channel tone stack (which I'm about to post.)

The capacitance measurement of C39 seemed pretty solid, but the measurement of C38 was a bit iffy.

Notice interesting T-configuration in the feedback loop of the presence op amp for the OD channel.

[phasorman]

It uses four op amps (two dual op amp chips). Alas, I couldn't get readings for a lot of the capacitors here. They just made my DMM freak out.

Does this look familiar to anyone? I am wondering if the designer adopted some familiar patterns from other speaker sims.

The second stage is a two-pole Sallen-Key high pass and the third stage is a two-pole Sallen-Keyu lowpass. The last stage looks like a two-pole Sallen-Key lowpass, but it's driven by a network of capacitors and resistors, so it's more complicated than that. Working that out analytically is possible but would be a bit of a pain. I haven't though about the first op amp yet.

[phasorman]

Two 10K resistors split the ground and positive supply voltage (I'm calling V_CC throughout) to create a "virtual ground" I'm calling V_REF. These are bypassed by giant 470 uF electrolytics on the back of the PCB.

[mmolteratx]

It uses four op amps (two dual op amp chips). Alas, I couldn't get readings for a lot of the capacitors here. They just made my DMM freak out.

Does this look familiar to anyone? I am wondering if the designer adopted some familiar patterns from other speaker sims.

The second stage is a two-pole Sallen-Key high pass and the third stage is a two-pole Sallen-Keyu lowpass. The last stage looks like a two-pole Sallen-Key lowpass, but it's driven by a network of capacitors and resistors, so it's more complicated than that. Working that out analytically is possible but would be a bit of a pain. I haven't though about the first op amp yet.

Not terrible. Just a 3-pole SK using one op amp. Math is a bit messy if you want to analyze it though.

http://www.edn.com/design/analog/436397 ... one-op-amp

EDIT: Nice work so far though. That's dedication.

[phasorman]

EDIT: Nice work so far though. That's dedication.

I'm teetering on a fine line between dedication and obsession.

[phasorman]

The vast majority of the circuitry in the Ethos Overdrive is devoted to various kinds of usual linear filters, aka equalizers, tone controls, etc., with quite a lot of user control options. The only distortion I can see is in the little tiny gooped section.

Interestingly, many of the "capacitors" in the tone stacks (both for channels) are actually made up of two capacitors in parallel. Some of the capacitors are marked, so one can guess what the other one is. In other cases, who knows. In any case, there are various reasons for wanting to use two caps in parallel (or resistors, or whatever), but I'm not sure what the particular reasoning is here.

This gets a bit complicated given all the switching options (especially for the overdrive channel, which has the "boost" switch.)

The tone stack is a bit unusual in that the bottom of the bass pot is grounded (actually V_REF, the "virtual ground") through a 200 ohm resistor, instead of hooked to the top of the mid pot as in the standard Fender/Marshall tonestack. I recall the bass being grounded "accidentally" in Vox amp designs, and which is said to contribute to some of the Vox sound.

Perhaps someone familiar with the Dumble amps that inspired the Ethos Overdrive could comment?

[Aaack, I wrote "it's" at the bottom of the page when I should have written "its."]

[mmolteratx]

The vast majority of the circuitry in the Ethos Overdrive is devoted to various kinds of usual linear filters, aka equalizers, tone controls, etc., with quite a lot of user control options. The only distortion I can see is in the little tiny gooped section.

Interestingly, many of the "capacitors" in the tone stacks (both for channels) are actually made up of two capacitors in parallel. Some of the capacitors are marked, so one can guess what the other one is. In other cases, who knows. In any case, there are various reasons for wanting to use two caps in parallel (or resistors, or whatever), but I'm not sure what the particular reasoning is here.

This gets a bit complicated given all the switching options (especially for the overdrive channel, which has the "boost" switch.)

The tone stack is a bit unusual in that the bottom of the bass pot is grounded (actually V_REF, the "virtual ground") through a 200 ohm resistor, instead of hooked to the top of the mid pot as in the standard Fender/Marshall tonestack. I recall the bass being grounded "accidentally" in Vox amp designs, and which is said to contribute to some of the Vox sound.

Perhaps someone familiar with the Dumble amps that inspired the Ethos Overdrive could comment?

[Aaack, I wrote "it's" at the bottom of the page when I should have written "its."]

It's a scaled version of the Dumble tone stack. I think the "Skyliner" but I'm not too familiar with the terminology for that amp. The real deal has a 500kA bass pot which is grounded through a 10k resistor, and 250kA treble and mid pots. Values seem to be scaled back by a factor of 50.

[phasorman]

The clean presence circuit in the Ethos Overdrive is a single-opamp noninverting highpass shelving filter.

I will presume that C39, which measured 4.5 nF, is really 4.7 nF, since that's the nearest common value.

Does the analysis below seem reasonable with what people would expect a "presence" control to do? It comes after the tone stack. (As an aside, in the OD channel, the OD presence circuit, which I'm not analyzing here, comes after the gooped distortion circuit, but the tone stack comes before the distortion.)

My understanding is that in most tube amps, the presence control changes the feedback from the audio output transformer back into the phase inverter. There's no such things in the Ethos Overdrive, so my guess is this circuit is trying to achieve a similar effect with a simple filter in the forward signal
chain.

Here we go:

1) R44 and R45 provide a voltage divider feeding the pin 5 (+ terminal) of the 8672A by VR11:

R45/(R44+R45) = 10K/(2.2K+10K) = 0.8197

2) For low frequencies, C39 is essentially open, so the op amp acts as a unity buffer. For high frequencies, C39 is essentially a short, so the op amp has a gain of anywhere from

1+R42/(R43+VR11) = (1+3.3K/220R) = 16 (presence control turned all the way to V_REF)

1+R42/(R43+VR11) = [1+3.3K/(220R+20K)] = 1.1632 (presence control turned all the way to NC)

3) Combining (1) and (2), the DC gain is 0.8197, and the "infinite frequency" gain ranges from:

16 * 0.8197 = 13.1151 (presence turned to V_REF)
to
1.1632 * 0.8197 = 0.9535 (presence turned to NC)

4) The upper knee ranges from:

1/(2*pi*(R43+VR11)*C39) = 1/(2*pi*(220R*4.7 nF)) = 153920 Hz (way beyond hearing range) (presence control turned all the way to V_REF)
to
1/(2*pi*(R43+VR11)*C39) = 1/(2*pi*((220R+20K)*4.7 nF)) = 1674.7 Hz (presence control turned all the way to NC)

5) The lower knee ranges from

1/(2*pi*(R42+R43+VR11)*C39) = 1/(2*pi*((3.3K+220R)*4.7 nF)) = 9620.1 Hz (presence control turned all the way to V_REF)
to
1/(2*pi*(R42+R43+VR11)*C39) = 1/(2*pi*((3.3K+220R+20K)*4.7 nF)) = 1439.7 Hz (presence control turned all the way to NC)

[phasorman]

The GAC circuit in the Ethos Overdrive is driven by the output of the op amps in the presence circuits. On the clean side, the output of the presence op amp is dumped right in; on the OD side, it first goes through a lowpass RC filter made with a 47K resistor and a 22 uF electrolytic, which in theory is also loaded down by the GAC circuit. I ran two simulations, one with that initial RC filter for the OD circuit, and one without; the frequency response of the GAC circuit is almost the same either way.

While the resistor values were read off of the parts, the capacitors C40 and C41 were read with a multimeter. The C40 reading of 0.22uF looked pretty solid, but the 0.02uF reading of C41 was a little iffy looking. I used 0.022uF for C41 in the results shown here; using 0.02 uF didn't change the result significantly.

The result is that GAC is a basically passive bandpass filter with a center frequency of 400 Hz. When it's activated, the following op amp is set up to provide a gain of two, I think to compensate for the "insertion loss" of the GAC filter.

When the GAC circuit is switched off, the GAC circuit technically loads down the op amps feeding it, but the amount of loading seems to be quite insignificant.

[Incidentally, I discovered some errors in my previous posts during this experiment... R49 is 39K, not 3.9Km, and in the handwritten GAC schematic I previously posted, the capacitor in parallel with R50 is C41, not C50.]

Here's the OD version LTspice schematic:

gac_from_od_channel_schem.png

Here's the clean version LTspice schematic:

gac_from_clean_channel_schem.png

Here's the OD version frequency response:

gac_from_od_channel_freqresp.tiff

Here's the clean version frequency response:

gac_from_clean_channel_freqresp.png

[matt239]

How does the "drive" work?
- Including on the clean channel?

I wouldn't think the sound comes just from eq.?

What makes the drive, or compression, or overtones?

Are there clipping diodes? Or overdriven FETs?
Or is it op-amps being overdriven?

[ambusch]

Long time lurker here. But wanted to contribute to this because I think its a pretty cool pedal. While completely unverified, using the schematics phasorman posted, I've created OD channel equivalent. The use of 4 dual channel opamps looks kinda like "the dude" pedal from J-Rocket.

Note in the GAC position I omitted R46 which is the 1Meg reference to virtual ground (VR) for the Clean/OD channel common. I also just show some of the tone stack parallel caps as a single cap for simplicity. (Using film caps in parallel with e-lytics is a common trick to lower ESR though I'm not sure the purpose in this schematic). Also I omitted the GAC switch connection between R??2 and VR because its basically just a unity buffer with a gain of 2. vs. a gain of 1. If it causes clipping in IC4A, need to make that change.

R55B I wasn't sure where is is but I assumed it's somewhere near R55 given the other portions of the schematics. I left it in there but unconnected. Probably part of the TLE switching (which I also omitted for now). In the schematics, there were several yellow box caps, so I labeled them as such, but in any case I tried to make all the known part numbers the same as phasorman's schematics.

Also note that because Dumbles use a lot of different topology in the tonestack with more options you can shake a stick at, I left much of that unconnected. If you decide to build this, it will have a LOT of off-board wiring to complete and get it functioning. But the schematic is enough to get at least an idea of the pretty awesome engineering in this pedal and now you don't have to click through several hand drawn schematics (which are great BTW) Thank you Phasorman!!!

I've put together a rough sketch of assumed "clipper" stages for both hard clipping and soft clipping because there are many many options here as well.

A cool idea, because the gain stages in tubes distort assymetrically but the first stage inverts, is to do assymetrical soft clipping but in the opposite ways for each stage. Insert your favorite clipping methods here (don't forget you can soften clipping with resistors in the diode path). Or just make your own clipping options till it sounds cool.

Due to the buffered bypass path, you could even do a clean blend (similar to some klon clones and reportedly similar to "the dude").

Anyway, let me know if there are any errors in the "known" portions. Either way it's a pretty cool amp-in-a-box using opamps with a compensation for guitar amps typical mid cut tone stack.

Hopefully someone enjoys this. I've got a rough board layout as well if anyone is interested, but wires are coming out of all sides of the board.