To be honest, I will admit that the "self-split" output stage concept makes me grit my teeth and look away.
That's because a "self-split" output stage is an impossible compromise between mutually incompatible requirements.
To get reasonably balanced phases, you need class A (as you pointed out), and a "long tail", i.e., a large-value common cathode resistor.
And to get any reasonable power out of the thing, you need a "short tail", i.e. the smallest possible value of common cathode resistor, and class AB operation. Except the imbalance is now so severe that you still won't get much power out of the thing, because one output device is barely working.
So it appears to me that "self-split" is a lose-lose situation, an engineering absurdity that should be quickly buried in the back yard, and forgotten forever!
But it seems to work for some people, so I guess it has a reason to continue existing in a few dark corners of the Internet.
Not every designer is obsessive enough to worry about potential component failure modes that may never actually occur.BTW it had a volume put being used as a grid reference
I have a couple of 5902 pentodes in the junk box, and a couple of some sort of fairly low-mu submini dual triodes.
The question is whether I will actually find a use for them before I die, or whether they will appreciate in value and turn into the retirement fund I desperately wish I had.
-Gnobuddy
I tend to agree about the self-split power stage. Basically, you end up with around the same power output as an SE power stage using just one of the same tubes. However, it does let you use a smaller PP output transformer and in guitar amps it's quite OK to not have a well balanced PI and PA. Basically, it works for small guitar amps, using cheap tubes like in this one, but that's about all.
Sent from my LG G2 with CM13 using Tapatalk
Sent from my LG G2 with CM13 using Tapatalk
And that is certainly a perfectly valid way to look at it.
Consider this, though: supposed you wanted to create a car with very little horsepower at the drive wheels. You could simply drain most of the fluid out of the automatic transmission, or put really weak springs in the clutch, so that there was tremendous slippage and loss of power in the transmission, and very little power made it to the wheels.
That's kinda what the self-split output stage does. (Except it doesn't actually self-destruct from the excess heat, as the slipping transmission would in short order.)
Or you could actually install a small, low-power engine in the car, tuned to run efficiently and well, and give you great fuel mileage.
That's like a number of designs that showed up in the hundred-buck amp challenge thread: basically, you use small, low-power output valves originally designed for TVs or radios, and run a lower than usual B+ voltage, in an otherwise conventional amp design. Kinda like the design Catslab just posted in this thread.
I will freely grant that, in the big picture, there is nothing actually bad about the self-split concept. But it grates on me, in the same way as the slipping transmission/big engine idea does. It's just the way my brain happens to be wired.
-Gnobuddy
But that's exactly what's being done here. The idea with these micro-amps is to replace the ubiquitous 6L6 or EL34 output tubes with low power alternatives. (Usually tubes intended for preamp use are used as output tubes) The idea being to let you "crank it to 11" to get the overdriven tone without making very much power/volume. The OP is using little 1w pentodes as outputs so he's already put the "smaller, more efficient engine" you're praising in there. Are you saying he should've looked for even smaller pentodes and driven them more efficiently to get the same output and save what, half a watt, of wall power? What would be the point? Efficiency is in no way a goal or benefit when making a micro-amp. Even 1 watt is usually too loud so anything you can do to knock that down even more is great.
Last edited:
Precisely. And he's not using "self-split", so those little pentodes will work (relatively) efficiently, and generate more or less the textbook power output they were designed for.
No, I was saying the opposite. I like the approach that Catslab took (a "proper" design, with a proper phase-splitter that actually works, using low-power output devices.)
What I don't like so much is a "self-split" design that uses a pair of 2W valves to generate 100 mW of heavily distorted audio. The slipping transmission approach.
As I said earlier, you have a perfectly valid point.
It's like the contemporary push to encourage people to eat insects (instead of beef, chicken, or what-have-you.) Insects consume far less natural resources to produce the same quantity of animal protein, compared to, say, beef. If we switched to making meal-worm cacciatore instead of chicken cacciatore, we'd benefit the planet a little bit, maybe spend less money too. It's perfectly logical.
But also personally unpleasant, to me. It may be completely illogical, but I won't be eating any insects any time soon. I'd much rather switch to being vegetarian, if it ever comes to that.
But how about this, I'll put building a prototype "self-split" microwatter on my "to do" list, just for you, poptart! No guarantees that I'll like it, though.
My current guitar amp design is more like Catslab's - it uses small NOS radio output pentodes, and puts out about 2-3W at full blast, enough for me to play along with a few acoustic guitars, and maybe a fiddle or a keyboard too.
Two watts is about right in that setting, but, as you say, is certainly much too loud at home. So I included an L-pad between amp and speaker that lets me overdrive the amp and dial down the speaker volume when I want that.
I like the flexibility of this approach, compared to making a 0.1 W amp that really can't be used anywhere except in your living room.
Speaking of 0.1 watt, a little math: Say you're using an efficient guitar speaker with a 100 dB@1W sensitivity. Suppose you want to keep the SPL to 60 dB @ 1 meter, which is about the same loudness as a normal conversation between two people.
That means you need to drive the speaker with a power of (-40 dBW), which translates to one-hundredth of a watt, or ten milliwatts.
I've experimented with dropping output power this low and even lower. I found I can even clearly hear ten microwatts (yes, millionths of a watt!) fed into a typical guitar speaker a few feet from my ears.
But, so far, I cannot get good guitar tone at these power levels, no matter what I do. I think that human psychology (loud sounds grab our attention more) and human physiology (Fletcher-Munson loudness contours) are both playing a part here.
-Gnobuddy
First of all, apologies to the OP for accidentally hijacking this thread.
So far all the reports I've seen from people who have built self-split power amps is that they sounded fine and have decent power output. The thing to note as that these all used a low value cathode resistor for the tail, similar in value to what would be used for a regular Class A PP power stage. This inevitably means very poor balance, but acceptable in this kind of a guitar amp application. Like they say, the proof of the pudding...
Also on the subject of sub-1W guitar amps, the Zvex Nanohead is a great example of one that has excellent tone (I've even seen ZZ Top using them). I do agree with though, you eventually can get down to a power level that's pretty much useless for guitar.
Sent from my LG G2 with CM13 using Tapatalk
So far all the reports I've seen from people who have built self-split power amps is that they sounded fine and have decent power output. The thing to note as that these all used a low value cathode resistor for the tail, similar in value to what would be used for a regular Class A PP power stage. This inevitably means very poor balance, but acceptable in this kind of a guitar amp application. Like they say, the proof of the pudding...
Also on the subject of sub-1W guitar amps, the Zvex Nanohead is a great example of one that has excellent tone (I've even seen ZZ Top using them). I do agree with though, you eventually can get down to a power level that's pretty much useless for guitar.
Sent from my LG G2 with CM13 using Tapatalk
Okay, Poptart and you have convinced me to try this concept out some time.
The old "paraphase" phase splitter also had inherently poor balance (once you factor in component tolerances and age-related effects), and some people seem to really like that sound. And I deliberately unbalanced my last cathodyne phase-splitter by about 1 dB, to let some 2nd harmonic through.
A question for Catslab: is it "cat's lab", or "cat slab"? Inquiring minds are dying to know!
-Gnobuddy
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.