TC Electronics Switch amp

[bwaslo]

EVA, is that MC2 amp yours by chance?

[Eva]

No, but I'm working in something with some similarities

[Charles Darwin]

I presume the MC2 amp has been originally designed by XTA.
That is I remember XTA saying that they have their own class D design in the pipeline.
Since then they merged with MC2 Audio and since power amps are not part of XTAs product line while MC2 makes nothing but it would only seem logical.
The psu may well be the same as the switch mode psu MC2 has been using for some time in the rest of their E series amps.

[Kurt von Kubik]

Originally Posted by AP2

from : Kurt Von kubik

Taking your FB signal outside the postfilter, you can correct in the unlinearities
postfilter, but you can not avoid degradation of performance due to phase shift in the
filter itself.
-------------------------------------------------------------
yes,
You say right, post-loop filter is essential, but then where's the problem?
(leaving out important devices for now), it remains very simple diagram, suppose comparator, dt / logic stage and driver stage. all configurations in this company has many difficulties to enter good FB. (all amplifier would be perfect or the same performance).
Problem is in the stage pre-drive/drive architecture. all performances depends on it, crystal clear sound quality slew / rate etc..
B & O could not enter directly FB first stage of conversion because of the delay (total)
can not think of the sound driver as a step motor control.
example, suppose you change .. after comparator on Hypex, with traditional drive circuit you think it works the same? I remove entry-point of FB .. but back to the problem of other amp.
vorrei provare un TC module, è possibile?
...Philips has had great intuition...

I would try a TC module, is it possible?
Regards.
Sorry for my english

Where is the problem - hmmmm!
To me it looks like the output filter problem is a matter of the "filtered feedback signal.
If you take your feedback after the outout LPF, you can only provide audioband feedback to the input of the amplifier. This is a very important matter, because FB in addition to correction, also provides stability, but at higher frequencies.
The phase shift in the output filter will be malign when phase has shiftet sufficiently.
On the other hand, FB taken before the output LPF will not correct the unlinearities in the filter itself.
Doing both will give you both stability and correction of the LPF imperfections, but still the post LPF FB is audioband only.
The third loop was meant to compensate exactly that.
But anyway, according to experts, level shifting is also cruisial, which I believe is true, and also jitter is a very important matter, leading to layout issues.

I do not know if you have any experience with timing issues, but I do.
I´ve seen and heard several cd players and DACs modified with high precission clocks from i.e. LC, Tricord and others.
They were mounted insede the enclosure, as close as possible to where the original crystal was dismantled. Every single time the DAC/CD sounded a lot worse than with the original crystal, and that in spite of, that the new clock indeed was a high precission clock. But that will only go right if monted right at the very feet of the component needing timing. Long leads will make you drop the whole improvement on the flor.
To this day, I´ve only experienced improvement in sound quality, when these clockmodifications were removed, and substituted with the original cheap crystal oscillator.
I´m not in any doubt about that this must be due to layout, because a very precise timing device of course will be an improvement. But it is probably hardly possible at all, to implement that kind of stuff as aftermarket devices.

[AP2]

Hi,
what I said before .. you put flea in his ear?
ok, I think I have knowledge of time and synchronization circuits in real time ... some of my circuit ... still flying in the skies ... but this does not say anything about class D.
then, you're right, I totally agree with you. (fb, etc.).
in physics to solve a problem, often this is seen in reverse.
The truth is that we want to entrust a very complex task to Class D. OK, this is a great challenge, I'm really interested, but first I tried to understand what are the critical points in reference to an esoteric Class AB.
I am sure that this is the real point that big companies do not understand.
when we listen to a song, if you had proper instruments (very advanced for recording for post-procesing), you would understand why true esoteric Amplifier uses bjet with 250Mhz FT. if you change this with others, feeling wonderful .. disappears.
In other words, simultaneously amplify various instruments, requires an ability to modulate very high fast (first stage on class D, translate this to PWM).
next stage must be able to follow this speed.
FB can not be examined alone. (Function, pre-filter or post-filter).
is clear that it is very difficult to do very good for class D audio.
many digital amplifier, when I listen to them I think that is playing MP3 music ..

[Kurt von Kubik]

Originally Posted by AP2

Hi,
I am sure that this is the real point that big companies do not understand.
when we listen to a song, if you had proper instruments (very advanced for recording for post-procesing), you would understand why true esoteric Amplifier uses bjet with 250Mhz FT. if you change this with others, feeling wonderful .. disappears.
In other words, simultaneously amplify various instruments, requires an ability to modulate very high fast (first stage on class D, translate this to PWM).
next stage must be able to follow this speed.
FB can not be examined alone. (Function, pre-filter or post-filter).
is clear that it is very difficult to do very good for class D audio.
many digital amplifier, when I listen to them I think that is playing MP3 music ..

I do not believe that big companies do not understanding the importancy of components and topology. In fact I think they often do a lot better than small ones. But if you produce an amplifier in larger numbers, it has to stable and only require a minimum of service. In the opposite case, huge spendings for service covered by warranty might be necessary.
So they do have some other problems to cope with, than small manufacturers, where customers mysteriously to some degree, better can understand the need for expensive service.

I.e. Mark Levinson 331 etc. this amp very often has leaking electrolytic caps, due to their orientation and lack of quality. But still customers have them replaced at enormous costs, this in spite of electrolytics being pretty cheap.

About the FT of output devics, I completely disagree. To the best ones IMHO are Motorolas/ON semi they do have a lot of beauties to choose from, and the really can work when necessary.

The almost universely used Sanken ringemitters with FT a lot higher than the ON semi is IMHO not a well sounding device.
Looking at data one also will find, that they might be fast, but will not do any work at higher frequencies.
So when looking at power bandwith in an amp with both the Sankens and the ON´s the one utilising the ON´s are clearly superior.
I´v seen that many times, and I´ve also heard it.
I´d never go for the fast ones, but the rugged linear ones instead, and then bias them sufficiantly.

About the class D amps, which I must emphasize, are not digital at all, there are many problems to overcome.
These are tried solved in various ways in known products.
Looking at solutions carried out, some seems mor elegant than others.
Most DIYérs has heard about dead time, decreasing deadtime is an easy way to bring down THD, but doing so also brings the amp in danger of break down, due to tolerances and temperature deviations in various components.

The multiple FB loop technology elegantly handles this and a lot of other important problems. As far as I know, already the dual loop technology converts the output filter from a higher order sytem with higher order filtration, to a first order system with first order behaviour, but still with higer order filtration.
The tripple FB loop system carries this even further, because the FB signal taken after filtration is no longer frequency limited by the filter.
Thus unlinearities, ripple, and frequency dependant rising of the output impedance is cancelled out.

Now it seems as I´m fond of this technology itself, but that is not the case, it would not be of any interest to me, if it wasn´t for sound quality.
The TC modules are surprisingly natural sounding, but the downsides are, that you have to build your own PSU´s for both 50-70V and for 12V, and then you need a very high quality buffer.

Lately a friend brought around the Hypex amps, which were supplied from a huge UI core transformer, 1.5KVA I think, and smoothed by BHC slitfoil caps, but the TC module with ONE single 100VA toroid and less than 10.000µF smoothing capacitance, seemed both more powerfull and a lot more natural and present.

[AP2]

HI,
I'm curious to hear a TC module, (resolution and power at low frequency).
I may be surprised ..!
I know only that .. it took me 2 years to develop a new type of modulation on class D amplifier. (for now I can not mention the name)
I constantly during development compared a MosPower class AB, including the way of clipping, (up to natural pure square rail voltage), listening and measures with regard to class AB continuously.
I finished only when some audiophiles with anechoic chamber, have not recognized the digital amplifier between esoteric class AB.
As I said before, not enough fb and low-pas filter.se enough, all amplifiers will sound great!
I heard some sounds clean (up to 60% of PWM) but ... without big, deep bass (typical of good class AB) or with low resolution in the upper part of the sound.
I see many chip Class D up to 600W. also 600 small radio 1W .. do 600w!
during development I realized that despite THD (0.02% flat up to 40KHz), the sound was still digital amplifier, the mospower had a very different sound.
Therefore, I remain of the opinion that it is not easy to emulate the good old class AB
PS.
I think that will be marketed soon and I can send you one for test (without money). I think you have good speakers for your big class A

[Kurt von Kubik]

Hi AP2

I do know it is not easy to make a good class D amp, but it is not at al easy to design neither a good class AB amp, nor a good class A amp.
Everyone of them are difficult in their own ways.
Linear amplifiers need very good power supplies, and IMHO very thorough regulation af the supplies for the voltage stage. They need to be masive oversized in every stage to perform effortlessly in any load.
That is both expensive and heavy, and worst of all, in class A it creates substantial amounts of heat, making listening to music a winter hobby.
My class A amp elevates the ambient temp by as much as 6-7 dgr. C.

The class D amp is a quite different challenge, which you probably know very well already.

What I do know something about is listening, and IMHO this amp will never come out of any listening test as a digital device. The first reason is, that it isn´t really digital, next is, that it performs very different from other PWM designs I´ve heard.
It does in fact sound pretty much like a very well designed class A amp, it has the smoothness and it also has the authority and effortlesness.
Excactly that has been my reason, to stay put with my huge class A amp consuming 1,2KW and delivering 2*100W of class A power @ 6 Ohm.

The only downside is in fact, that upstream components really have to perform, otherwise it´s a pain in a sore place.

In other words, this might not be an amp, which you with ease can put into any audio chain, this because it really seems as a wide range low Q component. Unfortunately I cannot give you any clou of what it sounds like, because I simply do not know any amp to compare with, but deep bass with awsome control seems to be one of he easy tasks for it, byt try it out yourself and be aware of the need for a good buffer.

[KLe]

Originally Posted by Kurt von Kubik

... The TC modules are surprisingly natural sounding, but the downsides are, that you have to build your own PSU´s for both 50-70V and for 12V, and then you need a very high quality buffer.

... but the TC module with ONE single 100VA toroid and less than 10.000µF smoothing capacitance, seemed both more powerfull and a lot more natural and present.

Hi Kurt

Are you saying that you are running two TC modules with one 100VA PSU which has less than 10,000uF of smoothing? And that it is achieving comparable results to your Class A amp? What is the Class A that you are using?

What very high quality buffer are you using?

Is your 12V supply regulated?

Also, what is the PSRR of the TC input stage?

How do you think the TC modules would perform with an SMPS with 60V or 70V DC output? Actually, do you think that the TC modules sound different with 60V or 70V DC PSU?

thanks

[Kurt von Kubik]

Originally Posted by KLe

Hi Kurt

Are you saying that you are running two TC modules with one 100VA PSU which has less than 10,000uF of smoothing? And that it is achieving comparable results to your Class A amp? What is the Class A that you are using?

What very high quality buffer are you using?

Is your 12V supply regulated?

Also, what is the PSRR of the TC input stage?

How do you think the TC modules would perform with an SMPS with 60V or 70V DC output? Actually, do you think that the TC modules sound different with 60V or 70V DC PSU?

thanks

Hi Kle

Not excactly.
The TC module is a stereo amplifier, and it runs by now on 1 pcs. of 100VA toroid with 18 pcs. of 560µF Panasonic FC capacitors.
The 12V supply is by now done by 1 pcs. of 10VA UI core transformer and 2*1.500µF Sanyo awx capacitors.
One must understand, that oversizing the PSU for this module to achieve less ripple, will not bring you any performance gain. We tried that long ago.
The matter of concern in its supply is impedance rather than muscle.

The Class A amp is a Dali Gravity, which I consider the strongest 2* 100 Watt class A amp ever build, and it is an honest class A amplifier delivering its power in real class A.

The switching amp do have properties that I´ve never found on any amp, and so does my class A amplifier. But when it comes to power and strength, they are comparable, the module can deliver 200 watts in both channels, but will not do so whith sinusoidal signals for an extended period of time. The Class A amp will do that with ease from now on untill christmas, even @ 2 Ohms.
But with music signals the switching amp is not limited at all.

Yes my 12V are regulated, and we are working on an improvement in that, because our buffer also need 12V and we want to regulate seperately for that, as well as we also want a two step regulation, where the last one will be discrete and very fast.

Our buffer is a discrete unity gain design with no feedback and an one point DC servo circuit with very low cut of frequency, but the schematics are undfortunately not yet closed, since we want to implement more thorough regulation.

I do not know the excact data for PSRR on the module, but it is my impression, that it is huge. The designer himself told me, that ripple is no matter of concern, but as always, one has to listen for oneself.
But as I already wrote, I´ve found the impedance to pretty important, which points at some sensitivity for HF or something.

We did not yet try out SMPS to compare with the linear ones, but we will do so. It is my belief, that SMPS couldd a good job in this amp. Especially if 12V supplies is nicely regulated and free of any noise from the switching procedure.

The sensitivity for voltage should be very low, we decided to go for about 150-180Watts @ 8 Ohm. At this time, we are running it on 50V RMS.
We do not want to get to close to the electrolytic caps max rating (63V), because we then cannot use Panasonic FC anylonger. And that is IMHO pretty important to this module.