Dissecting a Budget Tube Screamer Clone

[cushychicken]

Need a cheap Tube Screamer for your pedal chain? Looking for an affordable distortion pedal to start modding on? Got a guitar gearhead you're desperate to get a stocking stuffer for?

Look no further, friends!

http://cushychicken.github.io/dissecting-ts-clone/

[Jdansti]

Nice, but this should be in the "For Sale/Trade" forum.

[mth5044]

Nice, but this should be in the "For Sale/Trade" forum.

wuh? I can't read the article at the moment, but I don't see snything for sale.

[Hatredman]

Was about to ask the same question. He's not selling anything and the article in the link is informative.

[R.G.]

The linked page has the feel and odor of a press release about it. While it's not directly a sales offer, it is itself a link to a sales site.

The link to "Donner Electronics" is a bit ominous.

[amz-fx]

The tone control on the linked page seems to be drawn incorrectly. The ends of the tone pot should be connected to the opamp +/- inputs and not in the feedback loop.

You might want to fix the tone amplifier drawing.

Best regards, Jack

[ashcat_lt]

The tone control on the linked page seems to be drawn incorrectly. The ends of the tone pot should be connected to the opamp +/- inputs and not in the feedback loop.

You might want to fix the tone amplifier drawing.

Best regards, Jack

Yep, as is its a treble boost.  Are we sure that the error is in the article, or is it in the design?  "...educated guesses..."

The 250K gain pot means it's got just a bit more than half the maximum gain of a TS, and when the LEDs are engaged, they bring a 470K resistor in parallel with them so gain is even further reduced.

It is a common misconception that the diodes limit peak output.  In this kind of circuit the output absolutely can and absolutely will be larger than the Vf of the diodes.  People want to call it "soft clipping" but it's not actually clipping (til it hits the opamp rails) and it is actually a form of crossover distortion.

[R.G.]

It's actually 1+clipped because the signal goes into the + input.

[ashcat_lt]

It's actually 1+clipped because the signal goes into the + input.

It is a lot like mixing the gained-and-clipped signal with the unity input, but I don't think it is exact.  That model tends to not work with low gains and/or small inputs where it never crosses the diode "threshold".  Close enough for most things, though, and the point is the same.

Edit - actually, it's always wrong for any portion of the signal swing that doesn't cross the threshold, but the error gets smaller as gain is increased.  Max gain is 55, so adding 1 is less than 2% error.  Min gain is 2, though, and adding 1 is 50% wrong.

[R.G.]

The output of a noninverting opamp stage from the + input to output is always the input signal times 1+ Zf/Zi where Zi is the input impedance to ground at the inverting input.

For an opamp to be op-amping, the + and - inputs must be within a few millivolts of one another as long as the opamp's AC and DC characteristics can make this so. The "few millivolts" is the offset voltage plus the output signal divided by the open loop gain of the opamp at that frequency. It's just enough to drive the output to that signal voltage, in other words.

Zf is nonlinear in this case, but as long as Zi has a capacitor to block DC and there is a DC path from output to inverting input, the DC gain must be unity from the + input to the output. So imagine that we slowly change the DC bias voltage from, say 3Vdc to 6Vdc. The output follows this, as it must. On top of this, you get the AC contribution of Zf/Zi times the AC input. This can be large, since the AC gain is large, but the DC gain is always unity.

So the output signal always has one times the input signal impressed on it, plus Zf/Zi times the input signal, as that is the output signal that, when filtered through the feedback/input impedance divider will move the inverting input enough millivolts to make it follow the noninverting input, as it must.

With large AC gains, the unity gain added to the clipped signal will be relatively small, but it's always there.

I have this suspicion that we're in violent agreement, just using different terms. 

[cushychicken]

Hey RG - I got the same feedback from another forum about this sounding like a press release. In hindsight, I got a little too excited about the design-for-manufacture angle to be critical of the pedal. It's not without warts; there are a few through-holes that got damaged by sloppy PCB panel routing, and as far as I can tell, it's totally lacking in testpoints, so this thing probably only saw rudimentary testing (if any) before it left the factory. In other words, it's like buying most electronics made in China - kind of a crap shoot, but the gamble is fairly low stakes when the prices are so affordable.

In any case, I wasn't really going for a review, per se. I really wanted to showcase the fact that this thing is basically a Tube Screamer for much, much cheaper. I also wanted folks to know that, in case they wanted to start getting into pedal mods, there's an easily moddable pedal that's effectively identical to a sought-after vintage effect. It took me about 15 hours of work to do all of the photography and ohming out of the components, and when I found that Donner had an affiliate link, it seemed like a reasonable way to be compensated for the effort involved. Call it a hack of the affiliate system.

Also, just in case this is the same RG - I love your article on "Technology of the Tube Screamer". That inspired a lot of this work, and the effort I'd previously put in to simulate the TS808.

[PBE6]

I don't think the Tubescreamer and its clones produce crossover distortion. If the diodes were directly in the signal path they would certainly generate crossover distortion, as they would be under-conducting until the signal strength was somewhere in the vicinity of the nominal V(diode) value, but when they're in the feedback loop of a non-inverting opamp they behave quite differently.

The opamp tries to keep the + and - inputs at the same voltage at all times, which it does by sending current through the feedback network. Including diodes in the feedback loop ensures that the V(out) = V(in) + V(diode), where V(diode) is a variable that depends on the characteristics of the diode itself, Rf, Rg and the input voltage. At higher input levels and moderate to high gain levels, the diode conducts a lot, which lowers the gain significantly from what you'd expect based on the gain settings (although it never reaches unity gain), causing the output voltage to be close to V(in) + V(diode, nominal)  (which is about 0.6 V for a silicon diode). At low input levels and/or low gain settings, the diode conducts less and the value of V(diode) is small but the relationship still holds. As we approach the beginning of the waveform (where input levels approach 0 V) the limit of the output can be shown to be V(out) = V(in) + V(in)*(Rf/Rg). Again, the relationship still holds and since V(out) = V(in) + V(diode), by inspection we have V(diode) = V(in)*(Rf/Rg) for very small signal levels. Essentially, this shows that the shape of the V(diode) curve is linear at the origin.

Contrast this with crossover distortion, where the output of a simple system consisting of an input impedance, a pair of series diodes and a load impedance is given by V(out) = V(in) - I(diode)*Z(in) - V(diode). For very small signals at the  beginning of the waveform, all the the above terms approach 0 which means the output also approaches 0, giving rise to the "dead spot" characteristic of crossover distortion.

EDIT: Wow, it took way too long to do the math and write that. After reading RG's explanation above I feel inclined to add that my comment is based on an ideal circuit, I'll leave "offset" and the like to the experts!

[PBE6]

EDIT: Voltage divider approach makes more sense.

For the simple system above, Vout/Vin = Zout / (Zin + Rdiode + Zout).

The resistance of the diode skyrockets for small signal levels, swamping out the Zin and Zout terms in the denominator which means the limit of the slope near the origin is Zout / Rdiode = 0.

[slacker]

I don't think the Tubescreamer and its clones produce crossover distortion.

It's like crossover distortion in that the sides of the waves get distorted but not in the same way. As opposed to "hard clipping" where the peaks get distorted and sides stay the same. Obviously that's vastly over simplifying it.

[PBE6]

It certainly does result in a hash being added to the original signal, although it's different than crossover distortion which is especially nasty sounding.

[lethargytartare]

If you're going to analyze a clone like this, I'd think you would compare the clone to the original TS schematic, or compare the response curves your modeler produces for the clone versus the 808 you did earlier.  But in no way would I tell someone to rush out and get one of these to hack away at -- modding a board full of smds??  30 bucks would get you the parts to build a clone from scratch, learn more, and have more long-term flexibility, durability, etc..  And back on comparing schematics: it might have been interesting to compare this clone -- which, by name, sounds like it's a clone of a Blues Driver, not a TS --  to the Mooer Green Mile, which is clearly a TS clone.  All in all, it does end up reading a lot like a sales pitch cloaked in a diy musing...

[lethargytartare]

...when I found that Donner had an affiliate link, it seemed like a reasonable way to be compensated for the effort involved...

So really, this DOES belong in the classifieds.   So you'll get 7% of any sales that you direct to donner's site using the link above, right? 

I also wanted folks to know that, in case they wanted to start getting into pedal mods, there's an easily moddable pedal that's effectively identical to a sought-after vintage effect.

A board of smds is not a great way to start getting into modding pedals, and this would not be easily modded.

[cushychicken]

If you're going to analyze a clone like this, I'd think you would compare the clone to the original TS schematic, or compare the response curves your modeler produces for the clone versus the 808 you did earlier. 

The LTSpice model in the earlier post is a TS808 schematic. That's how LTSpice works - you draw the schematic and feed in some simulation parameters. Feel free to use that as a comparison point if you're interested, you'll see that they're quite similar. If you've never used LTSpice before, shoot me a PM and I'll set you up with some guides on how to set it up and use it.

But in no way would I tell someone to rush out and get one of these to hack away at -- modding a board full of smds?? 

Why not? It's not difficult with the proper equipment - solder tweezers are cheap, and readily available.

30 bucks would get you the parts to build a clone from scratch, learn more, and have more long-term flexibility, durability, etc.. 

Thirty bucks? Really? I'd love to see where you can get that; most of the Tube Screamer kits I've seen tend to run about $40-45. Shoot me a link if you've got one handy!

And back on comparing schematics: it might have been interesting to compare this clone -- which, by name, sounds like it's a clone of a Blues Driver, not a TS --  to the Mooer Green Mile, which is clearly a TS clone.  All in all, it does end up reading a lot like a sales pitch cloaked in a diy musing...

I wasn't able to find a Blues Driver schematic online, but electronically, the Donner is very clearly a Tube Screamer clone - same opamp, same clipping stage, same tone control stage. Now you've got me wondering the same thing about the Blues Driver...

[bloxstompboxes]

30 bucks would get you the parts to build a clone from scratch, learn more, and have more long-term flexibility, durability, etc.. 

Thirty bucks? Really? I'd love to see where you can get that; most of the Tube Screamer kits I've seen tend to run about $40-45. Shoot me a link if you've got one handy!

You could buy all your parts from Taydaelectronics and probably do a build for about $20.00 depending on shipping. There is nothing expensive in a tubescreamer. The most costly parts are the jacks, pots, stompswitch and enclosure. Even so, the jacks are around a buck each at tayda, pots are $.50 a piece, and an enclosure is around a fiver. Resistors and caps are 1 to maybe 10 cents a piece. Figure $3 for the switch. You could buy some copper clad there for like two bucks and etch your own board or use a tagboard layout.

That's it, job done.

[Cozybuilder]

If you want a complete kit, you could find it on E-Bay for under $20. Try searching DIY OD2 pedal