thanks everyone, although a lot of the technical aspects of these posts are outside my grasp i get the idea, i have a 50hz lab scope (lcd) are there any tests i can run on my current amp, was a k-16l but is now modified to run in 8 and 16w configurations.
how and where do i make connections?
what should determine how good the amp is?
how and where do i make connections?
what should determine how good the amp is?
So I says run whatcha like and enjoy.
What you and DF96 said. One can make excellent amps using a variety of technologies- or poor ones, ditto.
I can't agree with that, so many factors are ignored. For instance the example 3 amplifier curves given - no circuits mean that you can't even see how much feedback the device(s) have, how many stages etc.
Additionally what do you mean by SS? Are you talking about IGBTs, MOSFETs? Triacs? They are all completely different - as different as a tube is from a BJT.
Also how do you measure accuracy?
It's not that simple. In the first place, how does one define "more accurate in matter of signal amplification"? That means two very different things in the time domain and the frequency domain. You can improve performance in one domain, but at the expense of the other. You can't optimize both at the same time.
As for sound reproduction, you add an EEEEEEEENORMOUS variable to the equation: the personal preferences of every listener, and these vary widely. The head-banger, the Classical fan, the Jazz fan, the Hip-hop fan, the casual listener who puts on a tray full of CDs for background noise while he does something else, the serious listener who'll sit there for an hour or two while really listening to and paying attention to the details of the music -- none of these are ever going to agree that the same system is the best system. Not. Gonna. Happen.
If it were that simple, we could publish a schemo of that Perfect-Amp-Everyone-Likes (regardless of implementation -- solid state, hollow state, hybrid) and be done with it, and shut down the forum forever. Again: Not. Gonna. Happen.
"Unless it s a loosy design, a SS amp..."
Unforch, "loosy" designs predominate in the SS world. It's possible to get really good sound from a SS amp. I know, because I dunnit myself. Of course, it wasn't easy, and a helluvalot of completed circuit boards landed in the trash bin after a few minutes of listening. It requires discarding a lot of "folk wisdom" concerning solid state design, devices, and topologies.
SS is not as suited to signal amplification (as hollow state is not suited to switching applications). Doesn't mean that the deficiency can't be overcome, but it's gonna take some extra work to get that sandy amp sounding really good. Otherwise, there really is no excuse as to why production SS amps sound as "loosy" as they do. All those "loosy" SS amps makes understandable the attitude you see among so many bottle-heads that all SS (and NFB and pents as voltage and power amps) are the Spawn of Satan. I don't buy into that nonsense (or the folk wisdom among the sand-heads that 0.000000001% THD figures automatically mean perfect sound reproduction).
Solid state, hollow state, NFB: pick the right tool for the right job, and "Git 'err dun!"
Last edited:
I can't agree with that, so many factors are ignored. For instance the example 3 amplifier curves given - no circuits mean that you can't even see how much feedback the device(s) have, how many stages etc.
Additionally what do you mean by SS? Are you talking about IGBTs, MOSFETs? Triacs? They are all completely different - as different as a tube is from a BJT.
Also how do you measure accuracy?
The topology doesnt matter much provided it s well designed,
but i admit that choice is rather limited in this matter.
As for devices, BJTs or Mosfets are both adequate , although
i settled for lateral fets for power stages but still using BJTs
for low power designs.
As for accuracy , logically , the amplified signal must be as close
as possible as the original one , safe the amplitude , of course ,
and in this matter , solid state cant be bested by tube amps.
I'm sorry but if you are comparing device linearity the topology is critical.
For instance you could be testing an open-loop tube and a darlington BJT with feedback.. you must show the test rig or the comparison is about as useful as comparing a bicycle and a car..
Oh and _which_ solid state device do you consider more linear than a triode, and why?
Last edited:
Since designing a SS amp that works is quite easy , there s scores
of badly designed ones that are used as a prove that SS is no good...
For a tube amp, a loosy design is less probable because :
a) design guideline for good results is way narrower.
b) badly designed means often either flashing of the beast
or blatantly bad sounding amp with scores of noises...
This is an erroneous statement.
Only SS designs have definition that match high quality digital sources,
by whatever parameter you look at it , not only THD.
Using tube gear with such sources is equivalent to using bicycle wheels
coupled with a Mercedes V12 engine..
What matters is the linearity of the whole design , not the individual
linearity of a single component.
As a matter of fact , tubes are branded as more linear individualy
than SS devices , yet, a tube amp will be far less linear than its
transistorized counterpart.
Well the topic does not say if it's tubes vs solid state or tube amps vs solid state amps, but I think a tube amp can easily be made as linear as a solid state one - with less feedback too!
If one were to weigh any measurements of any given circuit or execution under the same criteria as the ear hears it..a definite correlation to human perception of 'distortion' and electronic 'distortion' would emerge. The two would finally meet.
I mean to take a complex music signal under load..and then compare the input vs output.
In that act, compare only the leading edge of any peak transient and following into an opposite swing.
Compare the timing between all transients and their levels with respect to one another. This, as a single complex measurement. You would be looking at about 10% of the overall full complex sine signal. But 100% of the signal (for distortion analysis) would be considered to be in that 10% recorded and compared to the original's same 10% (10% of the full 'music signal' input).
This would inherently require a very fast and accurate digital capture system with high sensitivity on it's analog front end. We're talking about 1M sample rate and exceedingly low jitter. With about a 24(minimum) to 32 bit word length (comparatively speaking).
Remember, 10% of the full original signal would be just the positive peak transients and the follow up as they go into negative swing. That's it.
To the ear, the rest is unimportant, it never hears the other 90% nor recognizes the other 90% of the signal.
The proof in that, is that horn speakers produce 10-30% or even more distortion, yet people feel they are 'low distortion'. This is because the one thing they DO get right, is the positive leading edge of any transient.
if one does this with a complex music signal and a high resolution digital capture and analysis system, then measurements will lead directly to being fully correlated to what we humans hear.
Remember, as an engineer, or technician, or even scientist, one deals with numbers when measuring.
Up until now, the numbers looked at have had very very little to do with correlating to how the ear hears things, or how the ear looks at things. We have used 'whole signal' and used 'standard weighting' (full signal) to such calculations and measurements.
Remember, what I'm saying is that the ear hears only peak transient levels and the timing between them. The rest it does not hear or give two fecals about.
To the ear, this all important 10%, is 100% of the music signal. Thus it is very sensitive to distortions in transients, levels of transients and timing/level differences between them.
Therefore ---MEASURE THE SAME WAY.
It really is that simple.
I mean to take a complex music signal under load..and then compare the input vs output.
In that act, compare only the leading edge of any peak transient and following into an opposite swing.
Compare the timing between all transients and their levels with respect to one another. This, as a single complex measurement. You would be looking at about 10% of the overall full complex sine signal. But 100% of the signal (for distortion analysis) would be considered to be in that 10% recorded and compared to the original's same 10% (10% of the full 'music signal' input).
This would inherently require a very fast and accurate digital capture system with high sensitivity on it's analog front end. We're talking about 1M sample rate and exceedingly low jitter. With about a 24(minimum) to 32 bit word length (comparatively speaking).
Remember, 10% of the full original signal would be just the positive peak transients and the follow up as they go into negative swing. That's it.
To the ear, the rest is unimportant, it never hears the other 90% nor recognizes the other 90% of the signal.
The proof in that, is that horn speakers produce 10-30% or even more distortion, yet people feel they are 'low distortion'. This is because the one thing they DO get right, is the positive leading edge of any transient.
if one does this with a complex music signal and a high resolution digital capture and analysis system, then measurements will lead directly to being fully correlated to what we humans hear.
Remember, as an engineer, or technician, or even scientist, one deals with numbers when measuring.
Up until now, the numbers looked at have had very very little to do with correlating to how the ear hears things, or how the ear looks at things. We have used 'whole signal' and used 'standard weighting' (full signal) to such calculations and measurements.
Remember, what I'm saying is that the ear hears only peak transient levels and the timing between them. The rest it does not hear or give two fecals about.
To the ear, this all important 10%, is 100% of the music signal. Thus it is very sensitive to distortions in transients, levels of transients and timing/level differences between them.
Therefore ---MEASURE THE SAME WAY.
It really is that simple.
People on both sides of this argument seem to be allowing those with the opposite view to back them up into a corner so that they then feel they have to make more and more ridiculous claims/counter-claims in self-defence.
Components and architecture are both relevant, together with how well they are combined. Too many people seem to believe in 'magic bullets': use device X or circuit Y and you are guaranteed success no matter how incompetent your design skills. This may work in fashion, but not engineering.
Components and architecture are both relevant, together with how well they are combined. Too many people seem to believe in 'magic bullets': use device X or circuit Y and you are guaranteed success no matter how incompetent your design skills. This may work in fashion, but not engineering.
No. it isn't. Compare any transistor with even a bad triode like the 12AV7 or 6SF5. Transistors have much worse linearity, higher THD, and nastier clipping behavior on overdrive. There's no getting around that.
A source is a source, regardless. As for this claim: "Using tube gear with such sources is equivalent to using bicycle wheels coupled with a Mercedes V12 engine", it's absurd. I use "such sources" all the time, and there is much more detail revealed when listening with hollow state designs that simply is not audible with the usual, Big Box SS rig, even if you know such passages are there, and you're listening for them.
That depends on the topology and gain, no? For example, in common emitter versus common cathode *at the same gain*, the transistor may well be more linear. In common anode versus common collector, the transistor will generally be more linear.
As soon as you say "linear," you have to invoke topology, it's inescapable- transistors and tubes do nothing without other stuff hooked to them and voltages applied. That's why they each have their uses.
At last someone puts it into perspective - i've wanted to say all that for years but could not find the words.
Also no one ever mentions their listening rooms with concrete floors or wood floors with a sub floor void that booms it's nuts off, various furnishings, room shape etc ..never get mentioned.
It's so hard to interpret peoples descriptions of the sound of tweaks and mods and all the crap HI Fi Reviewer style raptures that get posted.
Sick of hearing comments relating to ' deeper bass ' and ' a veil has been removed '
I've heard good valve systems sound amazing and good SS systems sound....well....amazing but in fact no one, not one of you.....can tell me what my brain does with what goes ' into ' my ears
Anyone else out there like me or am I just spoiling the party ?
If it's done right in the first place it should be ' right '
Irrespective of the bits used.
Sorry for being so simple - interesting thread though
This, is not possible, even with thousand dollars parts , let alone "easily".
Linearity is solid state amp s stronghold.
I notice you have stated this several times, could you now demonstrate why you believe this?
As a designer of semi-conductor ATE hardware I can well see your point, and believe your assertion is essentially correct wrt to linearity, but the real question is whether or not the ear cares about the same linearity criteria as we do when doing technical design to that criteria, I'm not sure it does, in fact I am pretty sure it doesn't.
I design in solid state where I need accuracy for measuring things, there clearly is no comparison, however when I design something I listen to I often use tubes, and the comparison is much murkier, with tubes being fully viable contenders for my ears - and that may be true for me and possibly others as well. As much as I like good numbers in the end what my ears tell me rulez..
Also given the title of the thread it is not necessarily the case that the OP was thinking at all about such issues as the technical superiority of high feedback solid state circuits wrt to linearity as opposed to something he was hearing which would (could?) be much more nebulously connected to the concept of linearity as discussed, such as the difference in sound quality that he perceives.
Erm - you calibrate your solid state measurement tools so linearity is not so important, also I suspect you have never compared them to tubes - you use _only_ transistors for cost and packaging reasons.
I expect repeatability (lifetime) is better for solid state, but I have seen zero evidence for a more linear solid state device than a simple tube triode - sorry.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.