Mini-amp for Output Tube Distortion

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
IMD almost creates dissonance because the mixing products rarely fall on musical intervals. This is where "power chords" come in handy.
This is one of the reasons why I like my guitar amps pretty clean most of the time. I enjoy sus2, sus4, minor 7th, major 7th, half-diminished, and other similar "colourful" chords, but these don't sound good at all if there is appreciable IMD.

The 6JW8 mini-amp is clean enough to work well for these chords if I turn the volume knob well down. But if I turn up, I'm quickly into "classic rock" territory, and anything more complex than a two-note power chord sounds really nasty.

I'm not at all a good rock guitarist. I get bored very quickly with power chords and pentatonic scales and very limited playing dynamics. But this amp might help me get a bit better at this sort of thing - it's the first time I've had an amp that can be fully overdriven without becoming too loud for me to tolerate!
I was getting similar "stretched" sine waves by playing with the screen voltage in a pentode gain stage.
Exactly! In his valve preamp book, Merlin Blencowe talks about pentode load-lines, and how altering them can change the distortion and timbre. Combining his information with what I see on an oscilloscope, it seems that either lowering the anode load resistor, or lowering the screen voltage, tends to produce these asymmetrically stretched signals with lots of second harmonic distortion. (The load-line moves "above the corner" of the pentode curves.)

Going the other way, either raising the anode load, or raising the screen voltage, tends to make the signal flat-top on both positive and negative half-cycles, and the timbre is definitely different. (The load-line moves "below the corner".)

Trying to figure out what this little mini-amp is actually doing is making it very obvious that there are at least two gaping holes in my test-bench instrumentation capabilities at the moment. I can't really do FFTs (except the crude one built into my 'scope), and I can only measure frequency responses by the very tedious, very slow, entirely manual process.

There is a third thing I wish I had, the ability to produce a tone-burst signal, so I can see what signal dynamics do to the amp (bias shifts, etc.)

I think I see some DIY audio instrumentation in my future!


-Gnobuddy
 
IMD almost creates dissonance because the mixing products rarely fall on musical intervals. This is where "power chords" come in handy. They are mathematically created such that the primary IMD components are less dissonant, and are usually limited to 2 or 3 notes.


Yes. Due to the fact that power chords usually are two notes a fourth or a fifth apart, which means frequency relationships in small fractions of 4/3 or 3/2, respectively, IMD's don't sound too disharmonic.
Best regards!
 
I believe you are missing the point by a country mile (or a country 1.60934 km, since we are in Canada.)

I must confess, I used the dB relationship to take your analysis one step further. I did not have the time to do it myself (me bad).

Trying to figure out what this little mini-amp is actually doing is making it very obvious that there are at least two gaping holes in my test-bench instrumentation capabilities at the moment. I can't really do FFTs (except the crude one built into my 'scope), and I can only measure frequency responses by the very tedious, very slow, entirely manual process.

There is a third thing I wish I had, the ability to produce a tone-burst signal, so I can see what signal dynamics do to the amp (bias shifts, etc.)

I think I see some DIY audio instrumentation in my future!
If you have a computer running XP, Visual Analyzer might be something to look at.


Visual Analyser
 
Thanks for the link!

Apparently Visual Analyzer runs under Wine on Linux, so that is certainly something for me to check out.

I have an ancient laptop and a valid copy of Win XP, but that laptop originally came with Vista. Vista was unusable, so I wanted to install XP. Even back then, it turned out to be a major chore to track down XP drivers for all the crucial laptop hardware. I don't know what my chances would be of getting XP reinstalled on that laptop today (it has Linux on it now.)


-Gnobuddy
 
Completely oposite from the original concept of the thread, with all the talk about abusing tubes earlier I was thinking of building a board that I could hook up to any output tube and drive it with or without malice involved. Mostly just a simple amp in concept, some gain, tone stack, send it to the inverter and drive the outputs the way you want.



7oRVxJB.png



I want to try the step up inverter boards that I have so use one to get the high voltage, another that has a positive and negative rail use for the Mosfet drive. I did not include the terminal locations, going to have terminals to send out to the grids from the inverter capacitors. Biased with a cathode resistor. If fixed bias is desired short the cathode resistor and hook up the negative supply on the Mosfets. Still taking the drive from the capacitors mind you, will clip at 0V as normal. But move the grids to the source of the Mosfets and add the positive supply you get Class AB2.


Or at least that is the thought. If anyone sees an issue now is a good time to bring it up.
 
  • Like
Reactions: 1 user

PRR

Member
Joined 2003
Paid Member
> I wish I had, the ability to produce a tone-burst signal, so I can see what signal dynamics do to the amp (bias shifts, etc.)

Tone generator, you should have.

Put a door-bell pushbutton in series. Tap the button and watch the amp overload.

If your finger gets sore: build a tremolo pedal but turn it to the extremes. Total off/on and a low duty cycle.
 
@Gnobuddy
Did you have a look at audacity - primaryly an open source multi-track editor.
But can be used as tone generator and spectrum analyzer of hole tracks (not realtime) as well.
Thanks for the suggestion! Yes, I have used Audacity to generate sine sweeps, and do FFTs of wav files. But, as you say, the limitation is that it isn't real-time.

I think it would be easier to tweak bias on the fly if I could watch a real-time FFT and listen to the sound too, both at the same time.

I should try the FFT function on my 'scope again...maybe it's good enough for this job.

-Gnobuddy
 
Tone generator, you should have.
Yes, a little DDS signal generator I built from a kit, as well as the usual impromptu sources (play back a sinewave WAV file, etc.)
Put a door-bell pushbutton in series. Tap the button and watch the amp overload.
I have a couple of old 4066 CMOS bilateral switch ICs sitting around somewhere. Also several different JFETs.

I could use either one to switch between two different amplitudes easily enough. But then I realized I should probably be switching at the signal zero-crossings to avoid very nasty transients. That, I couldn't figure out how to do (at least, not yet!)

Maybe just do a "fast fade" with a JFET / opamp? It doesn't have to be any quicker than the transient generated when guitar pick hits guitar strings. I'm not sure exactly how fast that is, though.

Not being a shredder, the fastest I can play is about 12 notes per second. Making a crude assumption the tip of the pick traces a sinewave, and the further crude assumption that it drags the string for a quarter-cycle before it breaks free, that suggests very roughly (1/48) second or 21 mS to drag the string from rest position till it breaks free of the pick and flies back. But how fast does the string fly back? I have no idea. Is it reasonable to assume a quarter-cycle of the strings vibrational frequency?

Hmm. Maybe Audacity to the rescue again? If I can record and import the signal of a picked string into Audacity, maybe I can work out the timing of the various parts of the attack?


-Gnobuddy
 
:)

I think a good solution may have finally dawned on me. I have a couple of guitar pedals that have built in loopers. Instead of a tone burst generator, why not make a loop of a repeatedly picked guitar note, and just play that back as the test signal? I'll have to try that out one of these days.

-Gnobuddy
 
It should be easy to make a simple circuit with one of those CMOS switches to simply short across the top resistor in a voltage divider that's wired in series with the audio generator. Even if switched in the middle of a cycle, the resulting transient should be smaller than the original signal as long as no DC offsets are involved.

The doorbell on your house un-screws. Just saying.

Chances are at least one of your guitar's has this function built in.

Have you ever seen a video Mr. Metal Shredder ripping into a screaming feedback laden lead rip then he makes the sound pulsate on and off? It's easy on a Les Paul, simply turn one pickup all the way up, and the other all the way down, strike a note or chord, then slam the pickup switch back and forth. Many other guitars with two volume pots can do the same thing....it's a little harder on a Strat or other guitar with a multi position switch since you can't just blindly slam it back and forth.

Then capture a piece of this pulse tone and loop it with whatever audio tools you have.
 
7oRVxJB.png


If anyone sees an issue now is a good time to bring it up.
I'm wondering if there is sufficient current to charge up the input capacitance of the IRF820 gates in the available time, and if the 470k bias resistors can discharge them sufficiently fast.

I looked at the IRF820 datasheet, and it says there can be up to 360 pF input capacitance. Also up to 24 nano-Coulombs of stored input charge at the gate. With 1 mA of drive from the LND150, I estimate that translates to a slew-rate of 2.5 V/uS at the gate (one-fifth the slew rate of the often despised TL072). Alternatively, it will take 24 uS for 1 mA to stuff 24 nC of charge into that gate.

So charging (positive swings to the gates) are a little iffy, but discharging through those 470k resistors worries me considerably more. 360 pF input capacitance and a 470k resistance translates to a cutoff frequency that's a bit less than 1 kHz. I think that's too slow even for a guitar amp.

I have little experience with power MOSFETs, so I may be wrong. Let's hope George chimes in. (Or you can steal his MOSFET driver schematic, which he's posted on diyAudio a few times.)

-Gnobuddy
 
In the normal common source sense "input capacitance" refers to the gate to source capacitance which is large.......but we don't care!

We are wiring the fet as a source follower. The source follows the gate, IE stays at the same relative potential offset. This effectively "bootstraps" the input capacitance and removes it from the picture. In the real world, no follower is perfect, but a high Gm fet comes close, so it is wise to avoid those 5,000 pF fets, but 350 something is a don't care.

Next up is the output capacitance (drain to source). Since we tie the drain to a well bypassed positive power supply, and the source can drive 2.5 amps with 3 ohms of ESR, again we don't care.

The reverse transfer capacitance, Crss IS important to us. This is gate to drain capacitance, and since the drain is bypassed to ground, Crss becomes the input capacitance of the stage. My (old) data sheet says 2.7 pF, on par with many tubes!

When searching for a suitable tube friendly fet, look at the capacitance VS voltage curves. Silicon life forms always have more capacitance at low voltages, so look for those that have a low and relatively flat Crss in the range of voltages where it will be used. This one goes well over 100 pF at 1 volt, but drops to about 3.5 pF at 15 volts then gently decreases to 1.5 pF at 500 volts. Feed this guy enough voltage where it always has more than 15 volts across it, and all will be well.
 
Try a small pF cap, or series cap and resistor pair from the output tube grid to ground.
I spent a few minutes tinkering with LTSpice last night, and came up with the attached one-knob tone control idea. The frequency response tilts gently downwards over part of the tone knob's control range, which might be just what this somewhat bright little amp wants.

As a bonus, there is a slight mid-scoop at about the right frequency (roughly 500-800 Hz) for flatter "tone" settings. This might be good when dialing in clean tones.

Insertion loss is low, too, at around 6 dB.


-Gnobuddy
 

Attachments

  • One_knob_tone_control_001.png
    One_knob_tone_control_001.png
    54.8 KB · Views: 144
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.