In what ways does the phase inverter tube affect tone?

[johnnyjellybean]

Just like the title asks. Can anyone explain?

[Smitty]

A truly fulsome answer would require an entire chapter in a good tube-amps-for-guitar-players type book, but here's a little something to think about. The three main areas of consideration are the type of inverting circuit, its operating point(s) and the type of tube. What follows is a response absent of electronic theory and is not an invitation to other amp builders to or electronic engineers to 'correct' me. I have purposefully oversimplified my explanation to make the discussion more meaningful to guitar players and not military radio technicians with an inferiority complex. No offense intended to well adjusted 'Sparkies'.

While there are many inverting circuits out there, let's compare two popular ones: the split load (cathodyne) and the long tail pair (schmitt). The split load phase inverter takes advantage of the circumstance that the signal voltage at the junction of the plate and plate load resistor is out of phase with the grid of a tube, while the signal at the cathode is in phase. So if you take two signal paths off the same tube, one at the plate and one at the cathode, you will have two identical, out-of-phase signals to drive your push pull output stage. The long tail pair (LTP) is a little more complicated but a simple answer for our purpose here today would be if you share a common cathode resistor that controls the current flow in a second tube, you can have two out-of-phase signals coming off the plates of two tubes (or more commonly a miniature dual triode like a 12AX7). Notice how I didn't say identical for the long tail pair. The commonly attributed tonal characteristics of the two phase inverters are such:



Split Load
Much more balanced signal so less harmonic distortion compared to a LTP
The load is split across a single tube, so it has a lot less output than an LTP. It is commonly used to drive a pair of 6V6s. If you are driving a big output section it's typical to follow a split load with separate driving tubes.
It overdrives a lot like a regular preamp gain stage, which can be fairly euphonic in the early stages of overdrive [There's your "So what?"]
A big part of the sound of a Tweed Deluxe is the split load phase inverter.
Long Tail Pair
Compared to a split load the signal is less balanced from side to side, so you get some even order harmonic distortion...more or less of which is controlled by the operating point.
Because the signal is coming off the plates of two tubes, it is capable of significant amplification as well as phase splitting duties. This circuit can drive a quad of 6L6s into clipping if operated properly.
One side overdrives a lot like a regular preamp gain stage but the other side does not. While tone is highly subjective, many find the overdrive characteristics of a LTP somewhat dissonant. [Your other "So what?"]
Lastly, the architecture lends itself to the use of negative feedback, which can really change the tone. The 5F6A Bassman is a good example.
Operating Points

As is the case with preamp tube gain stages, the higher the voltage, as measured from the plate to the cathode, the greater the 'headroom' of the stage. Many will comment that a tube with a higher operating point, in terms of voltage, will sound glassier, yet brittle in overdrive. Lower plate voltages elicit terms like warmer and darker with earlier breakup. This is something that can be easily tuned in an amplifier. It works on both the split load and the long tail pair.

Type of tube

This part is fraught with complexity, because you can't just substitute tubes in a phase inverter without changing operating points, so let's assume we are also changing the circuit to account for this. [Easy there, Sparky]

12AX7
Greater amplification so it can push the output stage into clipping sooner, but higher impedance and lower power, so it can't push the output stage much past that. The clipping tends to be sharper at the entrance to and recovery from due to this. Also a little less bass response. [Note: Typically operated with less bias voltage so it will tend to overdrive sooner at its grid.]

12AT7
In between a 12AX7 and a 12AU7. The author's favorite tube for an LTP. Try one in a Tweed Bassman or JTM 45, the sub actually works quite well.

12AU7
Not much amplification (about a fifth of a 12AX7) so you have to start with a well-amplified signal at the input, but a lot more power. The power of the tube allows it to drive the output tube just a tiny bit into something called grid drive, or class 2 operation so when the output stage is clipping it's not quite as sharp at the entrance into and recover from clipping. Your ears hear this as 'smooth'. Also because of the lower impedance, better bass response.

The type of phase inverter, the tube and its operation has a tremendous influence on the overall tone of an amplifier. It is well worth exploring and understanding for a professional guitar player or avid enthusiast seeking to find his tone.

I hope this is the explanation you were looking for and not the answer to "If I change the PI tube in my Peavy Classic 30 from a Sovtek to a JJ, will I like that better?

Let the flames begin.

[hasserl]

Good explanation Smitty, not sure it was what the OP was looking for or not, but good stuff.

Note, a 12AT7 as a LTPI in place of an 12AX7 should have reduced plate load resistor values. The reduced impedance will allow the tube to drive the power tubes farther into non-linear operation. The SF Fender 47k value is actually better than the BF 82k/110k values for this reason. A well set up SF w/ 47k PI load resistors can deliver big fat power tube distortion. It may sound less gainy at low drive levels, because the lower gain will not push the power tubes into distortion as early. But when you drive it hard they deliver better.

[johnnyjellybean]

More or less exactly what I was looking for. I would also be wondering that if different manufacturers of the same type of tube, let's say a 12ax7a, impart different tonal footprints as a preamp tube, then there should there also be noticeable differences swapping them as inverter tubes no?

[hasserl]

More or less exactly what I was looking for. I would also be wondering that if different manufacturers of the same type of tube, let's say a 12ax7a, impart different tonal footprints as a preamp tube, would there be as noticeable a difference swapping them as inverter tubes as when there are changed in the other pre positions.

There can be, it depends on the amp design. Older non-MV type amps with fewer gain stages that derive much of their characteristics in the phase inverter and power amp, the differences between tubes is more dramatic. In newer higher gain amps that develop most of their characteristics in the preamp, the brand of phase inverter tube has much less effect, IMO of course, others may differ.

[johnnyjellybean]

There can be, it depends on the amp design. Older non-MV type amps with fewer gain stages that derive much of their characteristics in the phase inverter and power amp, the differences between tubes is more dramatic. In newer higher gain amps that develop most of their characteristics in the preamp, the brand of phase inverter tube has much less effect, IMO of course, others may differ.

In this case I'm talking about newer high gain amps. Any of the ones in my sig for example.

[Smitty]

Good explanation Smitty, not sure it was what the OP was looking for or not, but good stuff.

Note, a 12AT7 as a LTPI in place of an 12AX7 should have reduced plate load resistor values. The reduced impedance will allow the tube to drive the power tubes farther into non-linear operation. The SF Fender 47k value is actually better than the BF 82k/110k values for this reason. A well set up SF w/ 47k PI load resistors can deliver big fat power tube distortion. It may sound less gainy at low drive levels, because the lower gain will not push the power tubes into distortion as early. But when you drive it hard they deliver better.

Yes. By changing the load to a value closer to two times the internal resistance of the tube, it will deliver more power, but it will also deliver less voltage. More amps less volts. I tried the 47K and found that in order to come up with the signal volts at the plate resistors to get the output tubes to clip I needed to start with more signal at the grid and then bias the phase inverter tube a lot colder so I could get all that signal into the phase inverter tube before its grid started clipping. I found I needed to double the bias voltage as well as the input signal to come out to the same place in terms of phase inverter output voltage as the 100K/82K setup, before I could even begin to take advantage of the power.

The problem there is I ended up needing more Vbb (B+)to get that job done right. Not an easy thing to find at that point in the power supply.

Let's say you did that, though. You still need to solve the problem of the coupling caps discharging at zero grid volts on the output tubes. Well, once you solve THAT problem everything sounds so good that you just go right back 100K/82K because it gets you your 35 to 40 volts real easy.

The thing is unless you have a class AB2 driver stage in front of your output stage you're not going to get much above zero volts anyway. It's more about the shape of the wave as it approaches zero and recovers from it. I know you know this stuff, though, because I see you talking about it with guys like Ray Ivers and Gil Ayan on the boards where guys who actually study how a tube works hang out.

[RedMan]

I would say that there are quite a few other types of PI's not usually explored in guitar amps and they really should be. It's a bigger part of the overdrive character of the amp than the power tubes themselves.

[donnyjaguar]

Like Redman says. A transformer-based phase splitter sound very musical to my ears - and you can usually delete one tube stage too thus simplifying design.

[hasserl]

Like Redman says. A transformer-based phase splitter sound very musical to my ears - and you can usually delete one tube stage too thus simplifying design.

Interesting you mention this, as I've been in a couple of other threads recently regarding Fender's Musicmaster Bass amp, which uses a transformer for a PI. My Musicmaster Bass is one of my very favorite amps. Just a slight mod to the Tone control and a speaker change and dang if that little amp doesn't put a smile on my face every time I fire it up. Other than being a little low on power and the cab a little lacking in resonance, it's almost a perfect guitar amp.

[Ronsonic]

I love modding the Musicmaster Bass amps. Replace that hideous front end with a clone of a tweed Harvard or something like and they just sing. Very rewarding little things.

Sometime I'll have to do a full build based on one of those mods and see just how good that sort of amp can get. I'm thinking utter brilliance is attainable.

[zzmoore]

Good explanation Smitty, not sure it was what the OP was looking for or not, but good stuff.

Note, a 12AT7 as a LTPI in place of an 12AX7 should have reduced plate load resistor values. The reduced impedance will allow the tube to drive the power tubes farther into non-linear operation. The SF Fender 47k value is actually better than the BF 82k/110k values for this reason. A well set up SF w/ 47k PI load resistors can deliver big fat power tube distortion. It may sound less gainy at low drive levels, because the lower gain will not push the power tubes into distortion as early. But when you drive it hard they deliver better.

Pardon me if I am asking the obvious... This refers to a "un-molested" silver face amp, and not to a SF that has been BF'd. Am I understanding you correctly?
Thank You

[RedMan]

Split loads and paraphase are my current faves. I just finished a 100 watt 4x6L6 with a split load that sounds fantastic. I've used transformer coupled too. Great if you like a lot of overtones. You have to be careful how you drive it. I've used the primary for the cathode resistor with great results.

[ChickenLover]

Note, a 12AT7 as a LTPI in place of an 12AX7 should have reduced plate load resistor values. The reduced impedance will allow the tube to drive the power tubes farther into non-linear operation. The SF Fender 47k value is actually better than the BF 82k/110k values for this reason. A well set up SF w/ 47k PI load resistors can deliver big fat power tube distortion. It may sound less gainy at low drive levels, because the lower gain will not push the power tubes into distortion as early. But when you drive it hard they deliver better.
I just built (or rather...rebuilt...for the third time :D) an amp with a cathode-biased 2xEL34/6L6, no NFB power section and decided to try a SF inspired PI (47k plates, 470R cathode, 22k tail, 470k PI grid loads) and it sounds great when cranked. This amp in previous lives always had a regular 82k/100k 12AX7 PI and never distorted very nicely. Now it does and it's a smooth, graceful transition from clean to dirty all the way up the dial.

I've always wondered why high gain amps almost always use the Marshall PI values. We've got a TON of signal available in a high-gain preamp...why use the 'hottest' PI?

[hasserl]

Pardon me if I am asking the obvious... This refers to a "un-molested" silver face amp, and not to a SF that has been BF'd. Am I understanding you correctly?
Thank You

I'm talking about a well set up SF amp, so that may mean some slight modifications from the stock Fender circuit (depends on the circuit, there were quite a few). But I'm a believer in the 47k plate loads in the PI that were used in SF amps. I think Fender got it right with the SF amps there. I think it's a mistake for people to just change these to the BF values. My opinion only, and it's obviously a minority position.

I just built (or rather...rebuilt...for the third time :D) an amp with a cathode-biased 2xEL34/6L6, no NFB power section and decided to try a SF inspired PI (47k plates, 470R cathode, 22k tail, 470k PI grid loads) and it sounds great when cranked. This amp in previous lives always had a regular 82k/100k 12AX7 PI and never distorted very nicely. Now it does and it's a smooth, graceful transition from clean to dirty all the way up the dial.

I've always wondered why high gain amps almost always use the Marshall PI values. We've got a TON of signal available in a high-gain preamp...why use the 'hottest' PI?

There you go! That's been my experience also.

[dehughes]

Wow. Great thread, and answers the questions I had...

One other question: Would a Split Load PI work well driving a cathode-biased 5881 pair? I'm contemplating running either an EF86 into a 12AX7 LTPI into the 5881s, OR, a JTM-45 preamp into a split load into the 5881s.

What are your thoughts?

[KeithC]

Does the SF 47k resistor value have anything to do with boomy bass?
I have been told to change the .1 caps coming off those resistors to a lower value..either .047 or even .022

This particular amp already had the .47 resitors changed to the BF values. Seems that is the first thing people do to "BF" a SF amp!