I dare anyone here to prove that they can hear the low, low distortion in a high quality signal transformer. I would say that some of the great recordings of the past have decidedly less distortion.
Steve, you ought to keep in mind that a lot of people have trained themselves (or rather forced themselves) to like equipment based on the measured performance. I have been to demos of such systems, their owners filled with pride, and can only feel embarrassed about the results.
Having said that, John Curl's and Charles Hansen's amplifiers are superb, in my opinion. Right up there with the best transformer-coupled tube gear.
John
fizzard said:
Maybe. Maybe not. I can't really say one way or the other with absolute confidence. Do you have any conclusive evidence that the distortion produced by a good quality transformer such as the Jensens when run at or below their rated input levels is even audible?
And as an aside, at the end of the day, what quality of the sound is more meaningful than the perceived quality of the sound? Are you suggesting that people should choose something which doesn't sound as good to them because it has a better looking spec sheet?
Ok. Just be ready to hit the off switch in the even that your eardrums start trying to claw their way to the middle of your skull in an attempt to get away from that awful, distorted transformer sound. 😀
se
Hi, PMA,
I just realize you are asking the right question. From what I learn from making classD output chokes, at high frequencies the inductor becomes capacitive because of the proximity of the magnet wires. Samewise, capacitor will have non-neglectible inductance at HF.
What measurement equipment can show that?
I just realize you are asking the right question. From what I learn from making classD output chokes, at high frequencies the inductor becomes capacitive because of the proximity of the magnet wires. Samewise, capacitor will have non-neglectible inductance at HF.
What measurement equipment can show that?
I am using Rohde&Schwarz FSP7 spectrum analyzer, with BW = 10kHz - 7GHz, with dynamic range up to 120dB. Marconi signal generator, from audio band up to 2GHz.
Steve, fizz, jlsem, do you guys remember the spectra I showed in my Heretical article? The thru-preamp (JT-11P1 input transformer) spectrum was noticeably cleaner than the loopback. Ground isolation, CMR, and maybe a bit of useful bandwidth-limiting (100kHz) can all be positives. The 1V level at these impedances is a sweet spot for transformers.
SY said:
This is no surprise when compared to SE loopback, and limited to 100kHz. Have you also tried exactly balanced input for loopback spectrum?
PMA said:
Yes. But consider why that is the case.
First, they only use just a handful of turns as opposed to the thousands of turns you'd find on an audio transformer. They do this not only because they can due to coupling efficiency being so much higher at such high frequencies, but also because if they used thousands of turns, the RF signal would be eaten up by the winding capacitance.
Second, they use powdered ferrite cores. That's because laminated cores would be much too lossy due to eddy currents induced into the laminations.
Just the handful of turns means that the transformer's primary inductance will be exceedingly small and at audio frequencies would be little more than a short circuit.
Bottom line, just as RF transformers make for useless audio transformers, audio transformers make for useless RF transformers.
That's not to say that an audio transformer can't couple RF. They can, but it's not by way of transformer coupling. That path is just too lossy at RF. Instead they can couple RF through capacitive coupling due to interwinding capacitance. But then that's why the better input transformers have electrostatic shields between their primaries and secondaries.
Sure. Just send me your 2GHz signal generator and your 2GHz 'scope.
Either that or I can send you a transformer and you can do it.
se
For the record, transformers INCREASE in low frequency distortion at lower levels. EVEN JENSEN. They saturate at high levels, and HAVE to be driven in a special way to remain low distortion in general. Did you know this, STEVE?
Actually, while transformers are necessary and even convenient in some pro recording situations, I doubt the Crystal Clear or Wilson Audio ever used a coupling transformer in making a record or CD.
Actually, while transformers are necessary and even convenient in some pro recording situations, I doubt the Crystal Clear or Wilson Audio ever used a coupling transformer in making a record or CD.
well, perhaps this bears relevance to the thread. I've done my best to try and figure out the level of coloration that is brought to the audio table by just about every ancillary 'signal path' and part known to mankind.
Since I can't design a circuit to save my life (went another direction-and successfully-contrary to SY's thoughts on it)...I've concentrated on the parts quality, electrical noise (as in all ways) and mechanical aspects, vibration, etc) solders, wiring, solid state, tube, BJT, FET, Related power supply implementation as well, obviously...uhm..all connected to just about any given circuit..and mostly done as near or at 'single cause analysis on each unit or device type. Common sense to test one's mods out - before proceeding with the next. When it breaks after too many changes, where's the problem? So single cause analysis ends up being a sensible thing to do.
Now, I've tried tube active and passive and split RIAA, and same for differing SS circuits. This, on about 100 phono units, over the years.(Single cause analysis re-builds on each-via the extensive modification route)
I've never heard one outperform a transformer, with regards to low output moving coils. I've been as honest as I can with regards to what I'm hearing, as one would like to do away with transformers, as it's cheaper and has it's own form of neater-as simply a circuit. But I've never come across the pure SS unit that sonically outperforms a transformer, when it comes to the first stage of LOMC amplification.
At such a point we get into the argument of how people listen and what they listen for. In my experience, folks who prefer an all SS LOMC amplification are used to and listen for a very specific type of 'signal package', with regards to high end audio. The transformer type folk, actually listen differently and for different things.
Which is why it's so difficult to find consensus. As someone who uses Transformers, I believe (with some basic knowledge of how the perception model works with regards to audio reproduction in electronics) I know that answer to some degree, and that the satisfaction of the idea of 'musical (electrically based reproduction) truths', it leads me to the 'transformer' before the idea of 'active' first stage in LOMC amplification.
That this question gets constantly bandied about leads one to believe that fundamental issues have not been addressed, as the question would then answer itself, if such where the case. Ie, self evident- to all. It is my contention, after observation over a period of greater than 20 years, that it is the basic function of the given specific human hearing model-in the individual...this is where this whole human variable, as an answer, comes into being.
It's not the measurements, per se, it's the way the given person has learned to listen, and how that is applied to choices in design and equipment purchasing (with regards to the end user).
The human ear, hears by listening primarily to the transient edges of the note, and -over time. thus, linear weighted, full signal analysis will fail to capture the point the ear listens for, with regards to measurement correlation with the ears. It's basically incorrect signal analysis weighting -- as the core issue.
Now...speakers and other transducers have a problem. Low S/N and high levels of distortion. Now, one point that a speaker gets across fairly well, is the specific micro leading edge of the note. The rest can end up being highly distorted hash,as the box and other components of the system explode with generated noise after the initial pulse.
A VERY key point in the speaker and the electronics is that slurring, in time, over time, of the transient components can be, and is very low in level..but this is the key point the ear is listening for. The human ear can attribute a micro smear of a transient as being more 'open' or 'revealing'. Which, when analyzed on it's own, with respects to the original transient...turns out to have notable distortion..but will show up as being quite negligible..when weighted with respects to the entire signal-which is the 'current measurement model'.
The human ear is a device which can and does learn - how to hear. How to decode. We,as audiophiles, teach ourselves a way through the muddled mess that transducers bring to the table. As well, distortions, for the most part, with regard to electronics, can and do primarily take place in the transient domain. Exactly where the ear uses 100% of the signal, when listening. The transient domain.
now, as audiophiles, for the most part, when they get further down the road in their ability to 'hear', or listen, they invariably find that keeping control of the signal after the main transients..actually sounds better to them, than a clean or exaggerated (time smear, etc) transient. More natural, more integrated, with more correct and natural micro detail (micro transient information that is correctly weighted in relation to the rest of the signal whole that is occurring at the same time) after the initial transient(s).
Under that stated umbrella of re-weighting the measurements so it correlates to what the ear is looking for..I suspect you will find that even with the lower 'whole signal analysis' distortion characteristics that a SS circuit has (with regards to the very low level LOMC signal)..that..the transformer..in this specific case...when properly loaded..brings less distortion to the table. Once again, when you actually measure what the ear is looking for, instead of a basic distortion measurement-as we know them today.
Begin looking for the right or correct thing, begin framing the question correctly, by looking at the human hearing model and inserting that point in the measurement system..and this enigma fixes itself.
This is my contention for many a year..after playing extensively with the MC cart itself..and every single material and aspect (in near thousands of cases now) all the way to the customization of drivers, as a minimum.
I mention this, as it bears a considerable amount of relevance to SS circuit design, with regards to dealing with distortions and feedback schemes or layout and loading schemes, transistor choices, etc.
If one has and utilizes a fully digital measurement package, it is possible to design a 'modified algorithm' with regards to distortion measurement..and that is to only analyze data with a high delta..but in the context of whole signals passing through the given device, in their normal audio signal manner, not test signals. Real signal loading scenarios. 'In Situ' so to speak.
Under that whole package of analysis, I suspect that transformers may show themselves to be the equal or exceed that of SS qualities, when it comes to first stage LOMC amplification.
How such a thing would play itself out in general SS design, is for you guys to find out.
After all is said, here..I think that this point is one of the (current) fundamental changes or considerations that needs be addressed when it comes to audio circuit design.
I have more reasons that I can cast about from my 'quiver of logic arrows' on this exact subject (that explain further), but those reasons lead to other thoughts, experiments,and answers in my past..and the patent lawyer has those on his desk literally right now, and I can expound and expand on them at a later date. Ie, after filing.
Since I can't design a circuit to save my life (went another direction-and successfully-contrary to SY's thoughts on it)...I've concentrated on the parts quality, electrical noise (as in all ways) and mechanical aspects, vibration, etc) solders, wiring, solid state, tube, BJT, FET, Related power supply implementation as well, obviously...uhm..all connected to just about any given circuit..and mostly done as near or at 'single cause analysis on each unit or device type. Common sense to test one's mods out - before proceeding with the next. When it breaks after too many changes, where's the problem? So single cause analysis ends up being a sensible thing to do.
Now, I've tried tube active and passive and split RIAA, and same for differing SS circuits. This, on about 100 phono units, over the years.(Single cause analysis re-builds on each-via the extensive modification route)
I've never heard one outperform a transformer, with regards to low output moving coils. I've been as honest as I can with regards to what I'm hearing, as one would like to do away with transformers, as it's cheaper and has it's own form of neater-as simply a circuit. But I've never come across the pure SS unit that sonically outperforms a transformer, when it comes to the first stage of LOMC amplification.
At such a point we get into the argument of how people listen and what they listen for. In my experience, folks who prefer an all SS LOMC amplification are used to and listen for a very specific type of 'signal package', with regards to high end audio. The transformer type folk, actually listen differently and for different things.
Which is why it's so difficult to find consensus. As someone who uses Transformers, I believe (with some basic knowledge of how the perception model works with regards to audio reproduction in electronics) I know that answer to some degree, and that the satisfaction of the idea of 'musical (electrically based reproduction) truths', it leads me to the 'transformer' before the idea of 'active' first stage in LOMC amplification.
That this question gets constantly bandied about leads one to believe that fundamental issues have not been addressed, as the question would then answer itself, if such where the case. Ie, self evident- to all. It is my contention, after observation over a period of greater than 20 years, that it is the basic function of the given specific human hearing model-in the individual...this is where this whole human variable, as an answer, comes into being.
It's not the measurements, per se, it's the way the given person has learned to listen, and how that is applied to choices in design and equipment purchasing (with regards to the end user).
The human ear, hears by listening primarily to the transient edges of the note, and -over time. thus, linear weighted, full signal analysis will fail to capture the point the ear listens for, with regards to measurement correlation with the ears. It's basically incorrect signal analysis weighting -- as the core issue.
Now...speakers and other transducers have a problem. Low S/N and high levels of distortion. Now, one point that a speaker gets across fairly well, is the specific micro leading edge of the note. The rest can end up being highly distorted hash,as the box and other components of the system explode with generated noise after the initial pulse.
A VERY key point in the speaker and the electronics is that slurring, in time, over time, of the transient components can be, and is very low in level..but this is the key point the ear is listening for. The human ear can attribute a micro smear of a transient as being more 'open' or 'revealing'. Which, when analyzed on it's own, with respects to the original transient...turns out to have notable distortion..but will show up as being quite negligible..when weighted with respects to the entire signal-which is the 'current measurement model'.
The human ear is a device which can and does learn - how to hear. How to decode. We,as audiophiles, teach ourselves a way through the muddled mess that transducers bring to the table. As well, distortions, for the most part, with regard to electronics, can and do primarily take place in the transient domain. Exactly where the ear uses 100% of the signal, when listening. The transient domain.
now, as audiophiles, for the most part, when they get further down the road in their ability to 'hear', or listen, they invariably find that keeping control of the signal after the main transients..actually sounds better to them, than a clean or exaggerated (time smear, etc) transient. More natural, more integrated, with more correct and natural micro detail (micro transient information that is correctly weighted in relation to the rest of the signal whole that is occurring at the same time) after the initial transient(s).
Under that stated umbrella of re-weighting the measurements so it correlates to what the ear is looking for..I suspect you will find that even with the lower 'whole signal analysis' distortion characteristics that a SS circuit has (with regards to the very low level LOMC signal)..that..the transformer..in this specific case...when properly loaded..brings less distortion to the table. Once again, when you actually measure what the ear is looking for, instead of a basic distortion measurement-as we know them today.
Begin looking for the right or correct thing, begin framing the question correctly, by looking at the human hearing model and inserting that point in the measurement system..and this enigma fixes itself.
This is my contention for many a year..after playing extensively with the MC cart itself..and every single material and aspect (in near thousands of cases now) all the way to the customization of drivers, as a minimum.
I mention this, as it bears a considerable amount of relevance to SS circuit design, with regards to dealing with distortions and feedback schemes or layout and loading schemes, transistor choices, etc.
If one has and utilizes a fully digital measurement package, it is possible to design a 'modified algorithm' with regards to distortion measurement..and that is to only analyze data with a high delta..but in the context of whole signals passing through the given device, in their normal audio signal manner, not test signals. Real signal loading scenarios. 'In Situ' so to speak.
Under that whole package of analysis, I suspect that transformers may show themselves to be the equal or exceed that of SS qualities, when it comes to first stage LOMC amplification.
How such a thing would play itself out in general SS design, is for you guys to find out.
After all is said, here..I think that this point is one of the (current) fundamental changes or considerations that needs be addressed when it comes to audio circuit design.
I have more reasons that I can cast about from my 'quiver of logic arrows' on this exact subject (that explain further), but those reasons lead to other thoughts, experiments,and answers in my past..and the patent lawyer has those on his desk literally right now, and I can expound and expand on them at a later date. Ie, after filing.
Lumanauw, please read Jensen application notes. It is not possible to explain every trivial issue. Or chapter 6 of Walts lectures.
john curl said:
Yeah?
Here's a plot of THD+N versus frequency at fixed levels for the Jensen JT-11P-1:
An externally hosted image should be here but it was not working when we last tested it.
Admittedly my eyes aren't what they used to be, but I see low frequency distortion decreasing at lower levels.
What do you see, John?
"Driven in a special way"? Just what "special way" is that, John?
se
john curl said:
Yeah? Do you know everything they use in the recording chain? What microphones do they use? That's the first place you're likely to find transformers used.
se
Jakob2 said:
Even if then it still reads that he's saying that low frequency distortion increases with lowering of level.
se
- Status
- Not open for further replies.