How to increase Harmonic distortion for clean sounds

[echuta13]

Not being scientific by any stretch but what about running multiple low-gain stages to enrich the harmonic complexity for a clean channel? Maybe running stages parallel with different voicings? What about voltages? Higher voltages give higher headroom and lowering them tends to "brown" the sound. Just thinking out loud...

[Tahoebrian5]

That's I think where I'm heading with this although the more gain stages I think will be more complex and maybe lose some clarity.. Will be a fun experiment though. I believe brown sound is only referring to overdriven clipped sounds, at least that is my impression but I think I get what you are saying. I've thought about parallel stages but I think I may try that for a dirty channel.

[Mick Bailey]

I believe brown sound is only referring to overdriven clipped sounds, at least that is my impression

I can't ever remember hearing 'brown sound' before EVH stated using it to describe his brown-out Variac sound.

[JoeM]

Maybe you should consider using FET rather than a tube if you want more 2nd harmonic. Just a thought, check out this link.

A closer look at the Fetzer Valve

I have a Yamaha JX series amp (all SS) that uses a FET for the first gain stage and it has a wonderful clean sound. Easily rivals any of my tube amps.

[Alan0354]

Tahoebrian5,
I will introduce you to a way of looking at the triode gain stage which allows you to think about the distortion level.
You put an input signal voltage on the grid.
You take an output signal voltage from the anode

What you have inbetween is a signal voltage generetor of Vg1k x mu. For a 12AX7 that is the grid 1 to cathhode voltage (Vg1k) x 100.
The output of this signal voltage generator feeds 2 resistances in series connecting to AC 0V forming a voltage divider
The first resistance is the internal tube ra, the second resistance is the anode load resistor Ra.
The output is take from the centre of these 2 resistors.
As the current in the tube changes with signal (swings up and down either side of the bias current) then the internal ra value changes. The Vg1k x mu generator signal does not.
It is that change in ra with signal which produces your distortion.

If you want to keep the harmonic Distortion content up then using a tube with high ra compared to the Ra value is what you want. The high ra is the most likely to give you high ra change with signal.
That means you do NOT want to use something other than a 12AX7.
The 12AX7 has an ra of 62K at usual operating current and is usually used with a Ra load resistor of 100K. That is why it gives a reasonable amount of distortion.
ASIDE: When you see 12AX7 used in HiFI circuits it is operated with Ra of 470K so that the delta ra vs Ra is small hence small distortion. (delta means "change of")

For more distortion:
Increase the delta ra vs Ra by
1) reduce the operating current by increasing the cathode bias resistor, that increases ra and therefore also increases the change in ra with signal
2) reduce the Anode Load resistor Ra, try say 82K 68K or even 47K instead of the 100K, That makes the ra change more significant compared to Ra and so you get more distortion.

Hope this make sense to you.

Check out the page labelled 345 here:
http://tdsl.duncanamps.com/pdf/vm345.pdf

This is the data for a 12AX7A / ECC83 Tube.

That shows you the change in Ri with operating current, Ri is what Merlin calls ra and so I used ra as well (sometimes this is also called rp).
For maximum distortion you want to operate on that part of teh Ri curve which shows maximum slope, that is at the low current end.

Cheers,
Ian

This is very good information. You prompted me to review the theory. I have something that seems to be opposite from your assertion. I just want to present this and let you comment on this. Below is the drawing I drew:




1) Amplification factor u=gm x rp. And voltage gain = gm x (rp//RL)

2) In Fig.1, I drew the equivalent model of the tube where grid is like an open circuit. The current source Ip=gm Vg. As you can see, rp is parallel with RL. So RL needs to be as large as possible to let rp dominates in order to introduce more distortion.


This is my observation:

3) From Fig.2, I look at one curve for a given Vg. As shown rp=vp/ip. You can see slope at point B in Fig.2 is steeper than at point A. Therefore rp is higher at point A than point B. This means for a given cathode resistor, the lower the Vp, the higher the rp. That's the reason Fender Deluxe reverb using +170V at plate to get more of a bluesy sound rather than crystal clean sound.

4) I plot out from the plate curve using one particular load line and the waveform looks like in the bottom graph. The top is flatter than the bottom of the sine wave. This shows even harmonics.

[Alan0354]

Maybe you should consider using FET rather than a tube if you want more 2nd harmonic. Just a thought, check out this link.

A closer look at the Fetzer Valve

I have a Yamaha JX series amp (all SS) that uses a FET for the first gain stage and it has a wonderful clean sound. Easily rivals any of my tube amps.

But FETs have drain curves more resemble pentodes, not triodes. This means it has higher drain resistance than plate resistance than triode.

[JoeM]

Alan, I don't have the technical understanding many others have here. But my thinking was that a FET had more 2nd harmonic (maybe depending on how it's used). I've just been interested in something other than the typical 12AX7 pre-amp. Run-Off groove has some interesting designs all using FETs.

[Mick Bailey]

But FETs have drain curves more resemble pentodes, not triodes. This means it has higher drain resistance than plate resistance than triode.

But a unity-gain FET front-end on a tube amp sounds wonderful (to me) and improves the clean sound of almost any amp, but this is probably more to do with impedance matching. Sounds like taking a blanket off the amp.

[eschertron]

Thank you GT, that was actually quite enlightening and a different way of thinking about the circuit in general. So it sounds like my assumption that there are other triodes out there with more inherent THD , due to changing internal Ra, is not true. I had come across the idea reading thru random build threads and lost it. I guess you can't believe everything you read online! So lower B+ and lower anode resistors will get me more THD, thank you!

Let me add one bit to Gingertube's exposition: In my mind, where you choose the bias point (hot, cold, centered) has a great deal of effect on the second harmonic distortion that you get from a tube. All else being equal, changing the cathode resistor in a preamp stage can vary the output waveform immensely, with audible results. Take GT's math and run with it - understand the engineering principles and they will serve you well!

[eschertron]

Any free simulation for tubes? Does LT spice simulate preamp tubes and it's distortion?

I can't speak for LTSpice. I downloaded it once but never got around to learning it. I do a quick and dirty equation to calculate the second harmonic distortion in a stage by comparing the amplitude of the positive-going vs negative-going halves of a hypothetical input wave. The equation as given to me (sorry can't remember from where) 2HD = [difference in amplitude between pos half and neg half]/[ 2 * total amplitude]

"Eyeballing" the spacing between the load lines gives an even quicker and dirtier estimate. But you already know that!

[NateS]

But a unity-gain FET front-end on a tube amp sounds wonderful (to me) and improves the clean sound of almost any amp, but this is probably more to do with impedance matching. Sounds like taking a blanket off the amp.

That's impedance matching. I'll reiterate take a look at:

Mu Doubler

For all the 'clean' 2nd harmonic you wish to dial in (and then some).

[Alan0354]

But a unity-gain FET front-end on a tube amp sounds wonderful (to me) and improves the clean sound of almost any amp, but this is probably more to do with impedance matching. Sounds like taking a blanket off the amp.

Ha ha, we are talking about adding distortion, nobody is talking about sounding better!!!

Could it be that's the reason Matchless use a tetrode or a pentodes as the first preamp stage on one or more of their amp? FET drain curve really looks like pentodes plate curve at the linear portion.

[Gingertube]

I think that it is easier to viualize if you use an AC equivalent circuit with a voltage generator of Vg1k x mu and a series connected rp + RL (using your terms)
BUT
If we look at the AC equivalent circuit you used, using a current source, the same thing can be seen. the parallel combination of RL and rp will change as rp changes. Thus it is still the delta rp which is the source of the distortion and it is still the delta rp size compared with RL that determines the amount of distortion.

The statement "RL needs to be as large as possible to let rp dominate" is not quite right. The distortion source is not rp itself but the delta rp due to the change in current with signal swing. Delta rp will always be smaller than RL. to get maximum change in the parallel combination of rp and RL you want RL as small as possible.

For guys who read the posts above I've used Alan's naming conventions here. To line this up with previous posts "rp" is what I called "ra" and "RL" is what I called "Ra". I actually think the convention Alan uses is clearer but whan we started linking in the article from Merlin I use the convention he uses.

Getting this out at 8.00 a.m. in the morning, forecast max temp here in Adelaide today is 46 degrees C (115 degrees F) and as this is the 5th day in a row with temps over 40 degrees C (104F) I'm expecting brain fade to set in later in the day.

Cheers,
Ian

[Alan0354]

I think that it is easier to viualize if you use an AC equivalent circuit with a voltage generator of Vg1k x mu and a series connected rp + RL (using your terms)
BUT
If we look at the AC equivalent circuit you used, using a current source, the same thing can be seen. the parallel combination of RL and rp will change as rp changes. Thus it is still the delta rp which is the source of the distortion and it is still the delta rp size compared with RL that determines the amount of distortion.

The statement "RL needs to be as large as possible to let rp dominate" is not quite right. The distortion source is not rp itself but the delta rp due to the change in current with signal swing. Delta rp will always be smaller than RL. to get maximum change in the parallel combination of rp and RL you want RL as small as possible.

Thanks for the reply. I think the major difference between my interpretation is in my model, rp is parallel with RL, where your interpretation is rp is series with RL. That's where the difference in result comes from. I have been search the RDH4 where I got the equation in my notes. This is my interpretation:

In chapter 2.3, page 25 of RDH4, it called Mutual Characteristics=gm, Dynamic Characteristics gmd ( from chapter 12.2 page 487). I have to back deriving the formula why I justify that rp is parallel with RL instead of series. Excuse me in my hand writing but I cannot write latex formulas in this forum, so I have to scan my writing:



The formula of dynamic characteristics is given in P25 of RDH4. I back derive the gain of the stage = gmdRL=gm(rp//RL).

Also this is how bipolar and FET equivalent model look like also. The difference is the collector output resistance is so much higher than the usual RL that it never enter into the picture. Other people can double check my assertion.

If the model is rp//RL, then you need RL to be high so rp dominates. But if rp is series with RL as you indicated, RL has to be kept low to let rp dominates. That's where the difference and depends on which model is correct.

That's the best I can present my case, we need someone that is strong in the theory to break the tie!!! I am not saying I am right, this is just my best effort for now. Your post is a very good post, I learn something in playing with the plate current to control distortion.

[Alan0354]

Let me add one bit to Gingertube's exposition: In my mind, where you choose the bias point (hot, cold, centered) has a great deal of effect on the second harmonic distortion that you get from a tube. All else being equal, changing the cathode resistor in a preamp stage can vary the output waveform immensely, with audible results. Take GT's math and run with it - understand the engineering principles and they will serve you well!

Ian is absolutely correct to point out that you get more distortion when you lower the plate current by adjusting the cathode resistor. As you can see in my post #20, the plate resistance rp is higher and change more rapidly at point B. That's something I learn from his post.

This also holds true if you lower the zero signal plate voltage and get the same effect.