After building some plectrum acoustic instruments and experimenting with guitar amp sound, I am observing that the combination of speaker, cabinet, and electrical damping factor seem to be a design corollary to the soundboard of a stringed instrument.
Changing speakers IMO is a first order effect on sound, obviously in balance to the response of the circuit to being overdriven.
It seems to be more complex than frequency response, which can be modeled by tone controls.
It's hard to describe but it seems to be more about how the system responds dynamically to playing, in an acoustic sense as opposed to a mechanical sense, if that makes sense.
Damping and cone stiffness relate to design parameters like soundboard bracing and thickness. Power input and cone excursion relate to player energy and mechanical advantage in the bridge of an acoustic instrument.
Both make a complex system with many local optima in terms of frequency response and linearity, far more complex than a simple bass/treble rolloff situation.
OK, YMMV but I see these as the control parameters...
Last edited:
So what is your question?
Best paper-match?
Optimum power?
Or best g-amp tone?
6KCT is an interesting choice. It approaches a pentode-like mis-match and "poor damping".
Hammond 125A promises ratios for, on 8 Ohm load, 3K to over 100K; for 4r load, down to 1.5K with 6K as a mid-range option. Just do that. To be wild, use two 1P6T switches so "ALL" ratios are possible (including some that won't work; I doubt the tube or OT can be harmed).
Last edited:
A very good paper on choosing operating points and load is given by Paul Joppa in the the 1998 Valve magazine issues 3, 4, & 5. found here on the Bottlehead.com forum or under the heading Valve Archive if link doesn't work.
Work out your desired load for one tube and then double it for push pull.
Frequency response and time response are perfectly tied to each other via the Fourier transform - but only if certain conditions are met. If I understand the details correctly, those conditions are fairly well met as long as the speaker isn't driven too hard at low frequency, and the entire cone moves rigidly in one piece, like a piston.
But, with guitar speakers, it is quite routine for those conditions to not be met, at least to some degree. For instance, it's normal for a guitar speaker to be driven at frequencies high enough for the cone to experience break-up modes, i.e., the cone is not moving as one rigid piece. Instead, different parts of it are flapping about more or less independently, each emitting a different sound wave.
At these frequencies where cone break up is occurring, the measured frequency response at one microphone position is no longer able to accurately describe the motion of the entire cone.
The other condition for the Fourier transform to work perfectly - the system has to be linear - may also not hold true if the speaker is being driven hard (loud). Under these conditions guitar speakers may start to run out of (cone) excursion, at which point the speaker's spider and surround start to squash signal peaks and create nonlinear (mostly 3rd order) harmonic distortion.
If there is enough of this nonlinear distortion, once again, the Fourier transform will not be able to do an accurate job of mapping the frequency response to cone motion.
So I am open to the possibility that some of the subtleties of a guitar speaker that's working hard, may not be 100% captured by the frequency response alone.
I do think that the** frequency response still captures the major portion of what the speaker is going to sound like, though. Which is why "speaker emulator" filter devices do capture at least some of the sound of a guitar speaker.
** As we all know, guitar speakers beam treble like crazy, so there isn't a single frequency reponse: move the microphone a little off-axis, and the high frequency response will be very different. So perhaps it's optimistic to speak of "the" frequency response. Rather we should talk about the frequency response at a specific location wrt the speaker, for example, 1 metre away, and 30 degrees off axis.
-Gnobuddy
You know it is funny but until this thread I have never come across that simple explanation.
All the books and websites I've read have started from the assumption a transformer was already selected. Several sources said to use the plate-to-plate impedance from the data sheet. Kuehnel's book gives a very detailed mathematical analysis of the push-pull output, and many other sources give an abbreviated mathematical treatment. But all of them assume the primary impedance of the transformer is already known.
There are many sources which discuss composite characteristics and how to draw them or how to derive them mathematically. But then they too assume the primary impedance is already known.
I understood that each tube sees half or 1/4th the total impedance, depending on which class the tubes were operating in, but no source said that this was then used to select the transformer. They all looked at it from the reverse angle, that the transformer was already known.
When dealing with SE outputs or small signal preamp stages there are plenty of explanations of how to select the output load. Various rules of thumb based on specifications or graphically just looking at the plate characteristics.
Which is why I came here. If there was some simple method, surely it would have been in one of those sources. Surely someone would have said what you did, or give a step by step method as Gnobuddy did in this thread. Yet none did.
I am mystified why no book or website gives a simple explanation of how to design the push-pull the way so many sources do so for single ended outputs.
Last edited:
Don't EE's typically look at the kind of load to be driven when setting out to design the amp to drive it?
Years ago I happened on a web page where the host argued (contrary to the vogue of choosing a tube because everybody said it was best) that the output transformer was easily the most complex single component in the amplifier circuit and it made good sense to start with the OT and then build the rest of design to work in accordance with its characteristics. It was over my head and I left the page looking for that best tube , and by the time I got to a point where I realized I'd like to explore his method the page was gone, (or at least I wasn't able to find it).
Thinking about it another way, off the shelf transformer impedances and ratings can't be changed to suit your whim. Unless you're going custom wound to the needs of your chosen circuit you kind of have to start with what's available.
Incidentally, another valuable thing I learned from Paul and something that fits in well with his method of choosing op point is the way to scale plate impedance given on the data sheet to that at a different operating point you're interested in . . . . . which is that rp increases as current decreases by the cube root of the current ratio.
Yup, I came across numerous sources which specifically said to start at the load when designing, then work backwards. When dealing with a SE design they all gave a Step 1, Step 2, Step 3 instructional which included how to derive the load impedance. Twice Rp, 5x Rp, draw a load line, or some other process.
But as soon as you look at push-pull, not one single source said to derive the load as you would for single ended, then double it. Not one source gave a similar instructional with Step 1, Step 2. They all wanted to show how smart they were by proving mathematically or using Thevenin equivalents that the impedance seen by the tubes is 1/2 or 1/4 the total plate-plate impedance.
With (transformerless) solid-state designs, the load is an excellent starting point, and really narrows down the amplifier specs.
If you want to deliver, say, 30 watts RMS into an 8 ohm speaker using a class AB, you immediately know your peak output voltage (21.9 V), your peak output current (2.74 A), the expected power dissipation in each output device (~5W at 75% efficiency), the supply voltage (~48 V allowing for saturation voltages), and the average power supply current draw (~1 A) needed to make it work.
So, for a transformerless amplifier, this simple starting point is enough to tell you most of what you need to know about the output stage, and the power supply.
But all practical valve amps have output transformers - so there is now one more major variable in the mix. By varying B+ and transformer step-down ratio and valve type, very different amplifier circuits will be able to drive the same power into the same load.
As an example, I designed and built a 2 watt push-pull 6AK6 amp using around 225 volts B+ and a 22.5k transformer primary impedance (anode to anode).
If you believe the datasheet for a 50C5, you could deliver the same 2 watts into the same speaker using one 50C5, a 2.5k transformer primary impedance, and 110 volts on the anode. Half the B+, barely one-tenth of the transformer primary load impedance!
Conclusion: for valve designs we really don't get all that far just from the output power and load impedance. There is still a huge range of possibilities to pick from.
This sounds like it came from the world of valve "Hi-Fi", bordering on audiophool, where $300 output transformers are not nearly good enough, and $500 output transformers with polished solid brass end-bells are a point of pride. Even though the endbells, brass or otherwise, have zero effect on the performance of an audio transformer, other than to keep your fingers out of the windings. 🙂
Even in the DIY guitar amp universe there is (inevitably) much argument about output transformers. But it's quite clear what most of the major valve guitar amp manufacturers do, and have done for decades: use the cheapest, smallest, output transformer that will not burn up in use!
As far as I'm concerned, an electric guitar OT has to provide a full power -3 dB bandwidth from at least 100 Hz to at least 5 KHz, and anything beyond that is nice, but largely unnecessary. (And will be filtered out by the guitar speaker, in any case.)
The proof of the pudding is that quite a few successful Australian DIY valve guitar amps have been built using $10 line-matching transformers for the OT. (See Roly Roper's writeup for the theory behind this, here: Cheap Output Transformers )
And this is another twist. For some of us with thin wallets, the "best" tube is one that is on the $1 list at one of the large US vacuum tube vendors. 😀
Only unpopular tubes make it to the $1 lists. So you won't find any NOS RCA 6V6s there. But if you know how to interpret a valve datasheet, you might find something with potential, but unpopular because it was never used by Leo Fender or Jim Marshall.
Which is exactly why I was fiddling with load lines for a push-pull pair of 6BK5 valves earlier in this thread. They seem to be nice little beam power tubes, like a 6V6, but rated for about half the power. Which is still more than enough for any guitar amp I can actually use.
-Gnobuddy
NO, it doesn't. I posted that out of an interest in the electronic subject. It has nothing to do with mine is spiffier than yours but rather some small knowledge of the variations possible in a bunch of transformers all dubbed 5KΩ:8Ω.
My thing has always been tone. I put far more effort into developing tone than working on fast fingers and it showed in the kinds of comments I'd get at gigs.
I don't care about spiff, my comment that I went after that best tube was meant to show that I didn't know anything, didn't understand what the page host was talking about, and wandered off with the crowd in search of that mystical best tube. (Good 300Bs were $30 then!) Sorry you thought I was politicizing a current view.
Well, it depends.
If you are not pushing things, that plan works.
If your tubes don't have much current to spare, that works.
If your SE design runs high THD before it actually clips, the distortion cancellation in push-pull may allow you to run a lower load impedance and get higher power without higher THD.
Look at 2A3. Yes, the self-bias PP suggestion is 2X the SE suggestion, and gives more than double the power. But with a lower load (and fix-bias) the power is even more, and the THD is much lower than either other condition. Over 4X the power! This is like: buy two beers, in a cheaper mug (OT), and get 4X as drunk as one beer. An engineer's happy-hour.
"Pick transformer first and design the amp to it" has a flaw. The transformer may work best with an absurdly strong amplifier. Example from a different application: the "telephone modem" transformers are rated 600Hz, but "can" work to 80Hz, IF the amplifier is about 10X stronger (more current) than the modem application needs. In this field: I have seen power transformers give really excellent audio response, but need like 70V 150mA tubes to stay out of saturation and to feed the hungry inductance.
Attachments
Last edited:
My apologies, I had no intention of offending you. As I understood your post, it wasn't your viewpoint that the transformer is the most complex, blah blah, component in the entire amp, and should be the centerpiece of the design, but rather, the viewpoint of the author of that website.
I don't share that particular viewpoint, but it's a big world, and there's plenty of room for differing opinions.
-Gnobuddy
No offense taken, but thanks. My motive for that reply was to try and keep the view open to more possibilities.
I do hear you on the audio bling pride thing (I worked in a local audio store for a few years and saw it often), but that doesn't by default make everything that's expensive a waste of money.
I can't tell you whether I agree with that particular website's design view as I didn't understand much of it at the time and now I can't check it.
It does seem fairly simple though . . . . . leakage inductance, stray capacitance, layer winding, interleaving, bobbin winding, split bobbin, c-core, r-core, EI, lamination materials , insulation materials. . . . . . There can be a lot of variety hiding behind a single specification.
My experience tells me that everything makes a difference. The thing for me is whether or not I can hear it and if I can, what do I want to do about it, if anything.
I do hear you on the audio bling pride thing (I worked in a local audio store for a few years and saw it often), but that doesn't by default make everything that's expensive a waste of money.
I can't tell you whether I agree with that particular website's design view as I didn't understand much of it at the time and now I can't check it.
It does seem fairly simple though . . . . . leakage inductance, stray capacitance, layer winding, interleaving, bobbin winding, split bobbin, c-core, r-core, EI, lamination materials , insulation materials. . . . . . There can be a lot of variety hiding behind a single specification.
My experience tells me that everything makes a difference. The thing for me is whether or not I can hear it and if I can, what do I want to do about it, if anything.
Hold on a second there bubb ! If you want me to accept your argument I need to know what kind of beer we're talking here.
Well, now you're making my case for me, which is that the way you drive the transformer makes a difference, so knowing your transformer and thinking about/finding out what it takes to drive it is a good idea.
- Status
- Not open for further replies.