Tube Guitar Amplifier Overdrive Webpage

[Malcolm Irving]

How about figs. 1.19 and 1.20 in Merlin's article:
http://www.valvewizard.co.uk/Common_Gain_Stage.pdf
Looks like smooth clipping to me!
No doubt FETs and other 'sand-based products' can do the same if you deliberately engineer it!

[robrob]

Thanks JM, I hate it too

On the soft clipping graphic the '+' line wasn't supposed to represent the +V rail voltage, just point out that it's an AC signal with + and - voltage lobes.

I will supplement the clipping graphics with actual DSO captures.

What I meant by "trap" is a negative grid voltage can trap the free electrons between the cathode and grid but I will clean up the wording.

My website is definitely not geared toward electrical engineers and professional amp designers, my audience is made up of musicians and amp building amateurs.

I knew I would take a beating in this thread from the world's most knowledgeable tube amp authorities but I'm here to learn so I can attempt to help others learn too.

[Mike Sulzer]

@JMF

"One particular irk is that your explanations are *current* based (running out of electrons or accumulating too many), while clipping is muchbetter explained by *voltage* swing hitting limits (+V or ground) which it can not go beyond."

I think there is a bit more to it than that for a tube output stage. You do not get more than some number (maybe 225ma for a 6L6 without drawing grid current) of ma through a tube, and you will have to drop at least 50 volts to do that and get up onto the flat part of the curve. So what is the significance of this? First, even when in the linear range, dynamic output impedance of a tube amp can be high. I have measured four times nominal impedance (but I am sure that it varies), and so frequency and transient response are affected when connected to a guitar speaker. But if saturation is hard current limited, then this impedance goes even higher in saturation (but gets really complicated if grid current can be drawn or operating point shifted). This is very different from the usual voltage limited solid state amp where you tend to get clamped close to power supply voltages when saturated.

[uneumann]

An interesting point is also that introducing shifting DC offsets, shifting clipping (a)symmetry and shifting harmonic distortion pattern - controlled by dynamic envelope of the input signal - will in right combinations of "stacked up" clipping offer solutions that work decently in "both worlds" (medium and high gain). It is no coincidence that many practical guitar amps display such characteristics. They also naturally make the circuit react in more "touch sensitive" and "organic" manner than circuits devoid of such dynamic alteration of clipping characteristics, where harmonic pattern of distortion does not change but harmonic frequencies merely either rise or decay in amplitude depending on magnitude of overdrive.

That's a great summary statement.

[loudthud]

Hard Clipping: The most significant difference between solid state and tube amps is the amount of gain thrown away by feedback. When you throw away lots of gain you get hard clipping. If you made a tube power amp with 60 to 80 dB of excess gain, it will clip very hard. (Good luck on that, it will probably oscillate). Most guitar power amps have about 20dB of excess gain.

Touch Sensitivity: Another factor is the effect of the power supply. Tube amps use unregulated supplys. As the supplys sag, they affect the circuits they serve. Solid state amps typically use regulated supplys in the preamp and stiff supplys for the power amp. This is why tube amps are much more touch sensitive than solid state amps.

Distortion: Without feedback, JFETs actually distort more than triodes. Look up the work of Danyuk. He adds Source degeneration to linearize JFETs to more closely match the distortion of triodes. The simplistic architecture of tube amps with little or no feedback means they create even order distortion that increases with signal level. Opamps are relatively distortion free and stay so up until clipping occurs. Solid state circuits frequently employ diodes to create more tube like distortion.

[teemuk]

For opening words I would like to accentuate one very good point made in this thread, which is very well fitting...

I would certainly hesitate to write anything with definite general conclusions: someone has a counter example.

Because that could basically mean "definite general conclusions" like:

Touch Sensitivity: Another factor is the effect of the power supply. Tube amps use unregulated supplys. As the supplys sag, they affect the circuits they serve. Solid state amps typically use regulated supplys in the preamp and stiff supplys for the power amp. This is why tube amps are much more touch sensitive than solid state amps.

Well without going into details what effects all the aformentioned has it's basically about "fixed" vs. "dynamic" clipping thresholds. As said, replacing fixed clipping thresholds with dynamically varying ones is not rocket science. Implementation is actually cunningly easy in most cases. Several examples of this can be found, many of them from commercial products that have been sold to us for many, many years already.

e.g.

- Roland Blues Cube amps implement a little "power amp" in the preamp circuit that draws a lot of current under load and causes sagging of preamp voltage supply. Result is that FET gain stages of the preamp clip to dynamically varying supply rail limits. The effect is therefore exactly same as in generic tube preamps with sagging power supplies.

- Some Line 6 (SS), Quilter amps (SS), VHT and Maven Peal amps (tube) employ regulated supplys. But the regulator is controlled by an "envelope" presentation of the signal, making regulator's output voltage dynamically varying. Effect is once again dynamic reduction of clipping threshold to supply voltage limit.

- Fixed reference voltage of a clipping diode can be replaced with a dynamically shifting voltage reference. e.g. change ground reference to reference to output of similar aforementioned voltage regulator. Now you get diode clipping with dynamically varying threshold. In practice this scheme is not much different from ordinary gain stages clipping to supply rail limits.

Effects of sag are well known and researched. That means a fair bit of effort to mimic them in systems where voltage sag has inherently lower amplitude have been implented. Competent designers do not overlook important characteristics of operation, they will try to develop means to implement them to design somehow.

[teemuk]

Double post. Delete function inoperational.

[Usable Thought]

As a former reporter/writer/editor/know-nothing, I think there is often great value in a layperson (albeit a fairly experienced DIY'er who is continuing to educate himself) putting together the kind of web site that Rob has already created - many laypersons find it of value. And I admire that he's driven by his ethical/editorial standards to come here & post knowing he will get some strong critical feedback from people who know far more than he does. He's made it clear that's what he's here for.

And JMF, my bet is, that Rob is fully prepared to rewrite as need be - and that for him that IS the point. He's obviously willing to do what he needs to, so as to get something that at the very least is "not wrong" (and maybe even "right") from the POV of technical experts while still serving the needs of his audience.

AND - as a former writer/editor who in my time has helped many professionals write their first books for popular audiences - I have one suggestion - which Rob, you can take or leave. What I'm hearing from the experts here is that there are still many myths floating around about not only tube vs. SS capabilities and not just about the current state of SS design - but also the history of SS design.

To me as a writer, it would be a pretty juicy opportunity to not only cover the original topic of overdrive, but in the process - whether fitted in as a sidebar or woven into the main piece somehow - to dispel the fog of these myths & usher in some interesting content that is apparently less well-known. The two topics seem a natural fit. It is always fun to expose mythology for what it is. E.g. I have greatly enjoyed David Collins's video survey of guitar cap myth vs. reality; and likewise even though most of it is over my head, I enjoy learning (slowly) from Teemu's very well written book on SS.

I see a lot of books out there in the marketplace on valve amplifiers. I see fewer on solid state. Yet to me it's all interesting. I think the more good information intended for DIY'ers and audio/guitar enthusiasts, the better. Sure, sometimes valve hobbyists can seem like cork-sniffing status-oriented fools . . . but there are quite a few who really do want to know more about the reality and not the myth.

Anyway as an onlooker I just want to encourage robrob as well as all those here giving him such useful corrections & ideas. I'll go back to lurking with great enjoyment.

[teemuk]

...also the current state of SS design.

Good point. General impression seems to be that it's stuck on 1960's and is limited to crude applications like a plain, simple diode clippers. In practice we must not overlook several decades of time to both research the characteristics of tube amps, as well as develop all kinds of circuitry to mimics those characteristics. e.g. Earliest "tube emulation" schemes are about as old as earliest SS guitar amps so it would be ridiculous to think that remarkable development couldn't happen in period of about 50 years. That's half a century folks. Things do move forward.

[pdf64]

How about figs. 1.19 and 1.20 in Merlin's article:
http://www.valvewizard.co.uk/Common_Gain_Stage.pdf
Looks like smooth clipping to me!

Yes, I've puzzled over that too, as it doesn't seem to match my findings.
I wonder what frequency the signal is? Maybe a high frequency and the 100k grid stopper are acting together to make the soft clipping more apparent?
Whatever, I suspect that even so, as the signal level was increased, the clipping would become harder.

Now that you mentioned it, I'm not sure Rob's site was meant to be read by engineers/designers, Ron can correct me if I am wrong, but it's more for musicians that are curious about how their tube amps work...

Perhaps it would be helpful to come up with a concise, MEF approved text that would satisfy the inevitable 'why do tubes sound different?' question from regular musos, without any glaring technical inaccuracies?

[teemuk]

I wonder what frequency the signal is? Maybe a high frequency and the 100k grid stopper are acting together to make the soft clipping more apparent?

Perhaps. An interesting little detail of grid clipping is that higher source impedances make it harder, while lower source impedances make it softer. It's actually pretty evident when you think of the circuit as voltage divider consisting of source Z and grid Z, and where grid current draw significantly decreases grid Z. If source Z is lowered the resistive divider ratio remains lower, along with its effects to limit signal amplitude (clipping).

The high resistance before the grid merely eases load of the driver stage. When grid impedance decreases the series resistor at the grid introduces a limit below which the load impedance of the driver stage can not decrease even if the grid Z decreases significantly. Many generic phase inverter circuits handle driving highish load impedances much more gracefully than they handle driving very low ones so one gains more graceful power tube driver circuit behaviour with the drawback of grid clipping actually becoming "harder".

So, if you want soft grid clipping don't fit in a high-resistance grid stopper, fit in a driver stage with low output Z and high current amplification capability. This also minimizes DC bias shifting at grids and resulting effects to crossover distortion. One good reason why a design like Ampeg SVT power amp has somewhat soft clipping characteristics in comparison to many other generic tube power amps; the driver stage is such that it prevents hard grid clamping. Another reason is clamping the input signal amplitude with a solid-state diode below certain level, just a bit above the input sensitivity (and overdriving point) of the power amp. Third reason is the screen circuit design that results to less screen voltage modulation than what happens in generic guitar power amps.

Then again, most guitar power amps are designed very differently than SVT. Perhaps for a reason.

[ThermionicScott]

I knew I would take a beating in this thread from the world's most knowledgeable tube amp authorities but I'm here to learn so I can attempt to help others learn too.

Kudos for being a good sport about it, Rob.

Teemuk is perhaps too humble to toot his own horn here, but his book on solid-state guitar amplifiers is a great read, full of myth-busting and fun historical facts.

[tedmich]

As a former reporter/writer/editor/know-nothing, I'll go back to lurking with great enjoyment.

+1 on the well expressed opinion!

(is that a tiny Axolotl resting on someones hand in your avatar??)

[Usable Thought]

+1 on the well expressed opinion! (is that a tiny Axolotl resting on someones hand in your avatar??)

Thanks!

As for the icon, it would be very cool if it were an axolotl. It's actually an eft, which is also cool - bright orange skin if the photo were in color - in my wife's hand. I don't know why I picked it as an avatar except we like efts a lot.

[Justin Thomas]

Efts are awesome! I used to catch them all the time as a kid and keep them as pets. They start in water, nmove on land looking like salamanders (except with wobbly skin), and then go back to the water with gills and a color change. They're like backwards retrograde amphibians...
https://en.m.wikipedia.org/wiki/Red_eft

Justin