What's it all about?

[planet10]

As we dive further off topic, what would you use? I would personally like MDF/asphalt/MDF sandwiches, but that is overkill for my speakers.

I use BB plywood... thin, well-braced plywood. Abandoned things like the MDF sandwhich you mention some time ago.

dave

 

[dfdye]

Re: this thread is 50 years old

I am building tube amps because:
- I don’t know what to do with sands and only know the basics of tube (thanks M. Jones & DIYAUDIO)
- I do like the bright light
- I do like the sound
- I do like building them

I must admit I like the bright light too! 😀

These are all great reasons to build amps, and many of the same reasons I have an old bench plane in my shop. I have a ton of power tools that work better doing the same job, but I often find the plane is more pleasing to use for a particular job.

 

[planet10]

I think it is safe to assume that they are not sonically perfect systems

It is probably safe to say that even the very best audio systems are so far from sonically perfect that 2 systems could score the same on some perfect metric & still sound totally different (ie from a set-theory point-of-view in an n-dimensional space of all possible attributes, the set of attributes for one system could have little interesection with the attributes of the other system. and yet the scalar of the sets could have identical magnitudes)

dave

 

[planet10]

'm no engineer, but.... tone bursts. Square waves.

Here (http://www.t-linespeakers.org/fivecycles/cosine.html} is one that Bill Perkin's built an entire analog test system around... all the stuff after the initial burst tells huge amounts about the DUT if you know how to interpret it (don't ask me -- i'm just providing an example)

dave

 

[Gluca]

Good points SY ... really trying to learn here and I got great suggestions in the past from yourself and others

but in practice what tests would you suggest to investigate for

"good load tolerance and outstanding overload recovery, in addition to competent linearity"

If a set of tests proved to be indicative of good sound and quality and finally estabilished and, let me say, standardised well it would a great benefit.

Ciao
Gianluca

 

[Tubelab_com]

I was recently reading an article in the Audio Xpress magazine about some proposed testing method to determine an amplifiers response to "counter EMF" generated by a loudspeaker in response to a transient. The test involved running a sine wave through the amp under test, while simultaneously applying a signal through a resistor to the speaker terminals of the test amp. The resulting distrotion spectrum is then analyzed. The author presented the test results for 3 different amp. The results were very dirrerent. He then postulated a correlation between the results and the sound qualities of the amplifier. The results were also correlated to the type and amount of feedback used in the amplifier design.

I am currently at work, and the magazine is at home, so I can't tell you exactly which issue the article was in. I remember thinking that I would like to repeat some of these measurements to see if there is anything to it.

 

[SY]

gluca, I don't pretend to be a measurement whiz, but some improvised stuff does seem to basically do the job. These are far from comprehensive tests, but they do let me know that the amp on the bench will at least sound reasonably good in a real-world situation.

First, the basic measurements, THD, IM, power bandwidth. If an amp can't perform OK on easy stuff like this, it will be a disaster on music.

Stability. I use square wave tests into resistive loads, RC combinations, and actual speakers. This is done at a few different levels and a few different frequencies.

Stability after overload. For this, I use gated sine wave bursts, set for a level a bit above clipping. Again, I try to use a few different loads. The silence following a burst will hash up if there's any dynamic stability issue. It does require a tone-burst generator or a very clever gating circuit following a sine wave generator.

Blocking. For this, I use a similar waveform, but with a sine wave below clipping filling in the "spaces." Amps with poor blocking behavior (and this is nearly universal in RC-coupled feedback amps) will show very evident choking between clipping bursts. The lower-level sine wave may show very obvious crossover distortion or an exponential envelope. Bad. I generate this waveform with a function generator and a tone burst generator. Very jury-rigged, but it works.

I'd like to go a lot further, but am instrument limited. I invite anyone who wishes to gift me with an arbitrary waveform generator or an Audio Precision or Audio Critic's "Test Cube" to feel free to do so. I will not be offended.

 

[dhaen]

Sy,

Tone bursts will check for some blocking problems, but what you really need is some asymetry in the waveform... Yes, shock-n-horror, music is asymetrical.😱

 

[jeff mai]

Geddes's papers don't quite say that. What they say (and what I agree with fully) is that there's no correlation below a certain threshold , and that a more sophisticated metric is necessary. Gordon Holt said the same thing 40 years ago...

Careful there SY - the threshold below which there is no correlation is tremendously high, such that even most SET amplifiers would slide under it quite comfortably.

See Geddes' own results: http://www.gedlee.com/results.htm

Have a look at the THD chart.

 

[valveitude]

Well, I would say that I'm fasionably late...but since the thread is at page 13, I'll just except that I'm late 😀

A point that seems to have been missed (at least by me) is the inherently simple circuit a tube amp presents relative to their SS brethren. If you except that all passive components are in fact combinations of all passive components,(i.e. a resistor is in reality a resistor, a inductor, a capacitor, and a microphone…ok the microphone bit aint passive), it would seem clear that the fewer of these thingies we have in the circuit, the fewer interactions we have to deal with…ergo, the less chance we have of mucking up the signal integrity.

It seems to me, that if replacing a single capacitor in the chain can affect a surprising change in the sound, the fewer the better…I’ll go with 2-3 tube stages over 6-9 SS stages, along with all their support components, everytime.

Casey

 

[EC8010]

Although it's true to say that a simple circuit tends to produce simple faults, the standard transistor amplifier is actually only three stages; input differential pair, voltage amplification stage, output stage. All the other bits are there to make it work better. And valve amplifiers can use the same techniques of constant current loads (to reduce distortion due to varying ra), constant current tails in differential pairs (to improve CMRR and distortion), HT regulators, and output stage bias servos. The interesting bit is deciding what to use, and when. And how.

 

[rdf]

I was recently reading an article in the Audio Xpress magazine about some proposed testing method to determine an amplifiers response to "counter EMF" generated by a loudspeaker in response to a transient.

I read the same article. Very interesting results. I recall the SE performance weighted for masking threshold bested the SS amp. Then there are suites such as Audio Precison's Fasttest measurement systems which hammers the DUT witha block of different tones simultaneously. Could the technology be on the threshold of new standard metrics?

 

[EC8010]

Bear in mind that as far as professional audio is concerned, analogue died years ago. Modern multi-tone measurement systems are designed to test (digital) compression systems, but they might be useful for our testing.

 

[valveitude]

Although it's true to say that a simple circuit tends to produce simple faults, the standard transistor amplifier is actually only three stages; input differential pair, voltage amplification stage, output stage.

I think it would be more accurate to say a “basic” as opposed to a “standard” transistor amplifier. I have seen precious few simple 3 stage SS amps…maybe I have just led a too sheltered life.

All the other bits are there to make it work better.

Exactly, because they NEED to work better in order get anywhere near acceptable…and along with all the easily measurable improvements, a multitude of not so easily measured artifacts are introduced.

 

[EC8010]

No, "standard" was right. Most amplifiers are based on a circuit originated by H C Lin, and it's a three-stage design. It looks more complex because of the bells and whistles, but it really is a three stage design.

Your second statement is a little more contentious. I know how I can answer it. I recently watched a TV programme called "Rumpole of the Bailey" about a barrister who dealt in criminal law. (By the way, the books, by John Mortimer are fun.) Anyway, to get back to the point, he commented that, "Well, that rather puts you in the position of a vegetarian advising on how to cook boef bourbignon."

I'm convinced that any audible effect is measurable - we just need to make the right measurements.

 

[valveitude]

I'm convinced that any audible effect is measurable - we just need to make the right measurements.

I agree in theory. The problem I’m trying to illustrate is one of thresholds. If for example, you replaced one of say 12 good resistors in a circuit with a great resistor, it is unlikely you would notice a difference. If on the other hand, you replaced all of them with the improved component you likely would…you reached the threshold, or *pie-libreum, of noticeable difference. When the sonic differences between two seemingly similar components is so slight that you have to magnify the effect to hear it, I would say measuring it would be most difficult indeed. We don’t even know what it is we want to measure.

I’m not saying that simpler is ALWAYS better, just that by limiting the variables you have a greater chance of success. I would concede that it is entirely possible to build a very complicated, and good sounding amp…I’m just not good enough to do it.

I also would say that I’m just looking at one of many factors…the fewer imperfect passive components the better.

* Pie-libreum- - The last bite of a slice of pie taken before it rolls over onto its crust.

 

[rdf]

Bear in mind that as far as professional audio is concerned, analogue died years ago.

The irony of that too-true statement rings loudly in these analogue ears.

 

[leadbelly]

Anyway, to get back to the point, he commented that, "Well, that rather puts you in the position of a vegetarian advising on how to cook boef bourbignon."

* Pie-libreum- - The last bite of a slice of pie taken before it rolls over onto its crust.

Hey, quit that, it's dinner time over here, and I can't to dinner for another hour or two 🙂

 

[gingertube]

Standard SS Amps a 3 stage design

Quite correct.
The Linn, the Bailey (speeliing???) and even the Halcro are the basic 3 stage design. This "standard" design is discussed in detail in Doug Self's "Distotion in Power Amplifiers" article.

Interesting to note the very expensive but hughly regarded Halcros are VERY tubelike in their philosophy. Each of the "standard" 3 stages is optimised to the "hilt" with lots of local feedback and the whole amp runs ZERO global feedback. Very common in tube amps but it stands pretty much alone in modern SS designs.

Cheers,
Ian

 

[Tubelab_com]

The Voxson SS amp that I previously mentioned actually has only two stages in the power amp. A class A single driver transistor directly coupled to a complementary pair consisting of a silicon NPN and a germanium PNP. Barely enough gain so there is very little global feedback. Hey, it was sold in 1969 and 1970, probably designed a few years earlier. Their later amps were more traditional designs, and lost the magic sound.

I'll shut up now before I get called Transistorlab again.