Found out the circuit of THX 789. It's nothing new.

[filo97s]

The patent with schematics: https://patents.google.com/patent/US9071201
Again, nothing new, it's basically a combination of feedforward and feedback on an amplifier as you can find on books of control theory. But there's feedforward, and it's clearly evident.
An interesting news is that there could be a remote possibility to tune the performance and the measurements in the 1kHz region (that's why all the THX amps shows way too much increasing distortion past 1kHz - still better than 99% of amps however) using a tuned RC filter in the feedforward path. I'll keep you informed past the festivities, it's a theory under investigation.

 

[JohnYang1997]

The patent with schematics: https://patents.google.com/patent/US9071201
Again, nothing new, it's basically a combination of feedforward and feedback on an amplifier as you can find on books of control theory. But there's feedforward, and it's clearly evident.
An interesting news is that there could be a remote possibility to tune the performance and the measurements in the 1kHz region (that's why all the THX amps shows way too much increasing distortion past 1kHz - still better than 99% of amps however) using a tuned RC filter in the feedforward path. I'll keep you informed past the festivities, it's a theory under investigation.

There is feedfoward path yes. But it's not a ground up feedforward circuit. It's fundamentally a feedback topology. I have just tested the opa564 with different circuits and have rebuilt a 789 (amplification part) on a different pcb with apx555. I got better performance than 789(noise performance). And the extra circuit components do nothing in reducing distortion harmonics. Basically we have largest harmonic of -130db with 32ohm load and less than -120db with 16ohm load. I got same or even better performance with feedback only topology with less than half components.

 

[JohnYang1997]

BTW the distortion of 789 is mainly due to the differential amplifier section. If you use rca input the distortion is the same as apx555. The stability they were talking about doesn't make sense either. The stability is due to the large value of two 680 ohm in series and a gain of 6db. I can make the circuit stable with 100ohm or no feedback resistor. Also using method from control theory.
In short the 789 is a full on composite feedback amplifier in disguise under the feed forward. You can remove all extra components and achieve exact same performance. It's the feedback that performs all the error correction instead of the feedforward.

 

[KSTR]

A wise guy (Jan Didden) said: It's feedforward when you need trimming or very specifically tailored compoment values to arrive a the deepest distortion null, whereas any feedback scheme doesn't need trimming (feedback never overcompensates but feedforward may do).

Good example, with explanations easy to understand, is the MJR-9 amplifier.
The main idea of this feedforward is that the master amp is considered distortionless (and runs open-loop or at high gain) but the slave amp (unity or low gain) has distortion, which appears inverted and scaled up at its input... from where it can be mixed passively (after the feedback) with the main output to hopefully perfectly cancel. The though part is the matching of the master amp's gain which must be known, fixed and stable in order to find the passive mixer setup for perfect cancelling. Another property of this feedforward is that the output impedance cannot be made "zero".

I find it embarassing the patent got granted, absolutely nothing new or ingenious in it, but we all know the patent system (especially the US variety where almost everything passes -- I've seen patents granted on circuits that were an exact replica of a datasheet front page schematic of some ADI opamp... audicity wins over sincerity these days).

 

[pos]

Basically we have largest harmonic of -130db with 32ohm load and less than -120db with 16ohm load. I got same or even better performance with feedback only topology with less than half components.

Very interesting !
What would it take to increase max current capabilities and get something usable on a 8 ohms load?
With a 15.5V max voltage that could be up to 30W if current allows it!

 

[JohnYang1997]

Very interesting !
What would it take to increase max current capabilities and get something usable on a 8 ohms load?
With a 15.5V max voltage that could be up to 30W if current allows it!

Then you need a lm3886.

 

[Giulio]

@JohnYang1997, how have they achieved the doubling of the tension in the balanced output? By looking at the pictures, it seems there are only two amplifiers, so it isn't using two amps in series to drive the load

 

[JohnYang1997]

@JohnYang1997, how have they achieved the doubling of the tension in the balanced output? By looking at the pictures, it seems there are only two amplifiers, so it isn't using two amps in series to drive the load

There are 4 opa564s for output. So it's actually true balanced output. The inverting sections' inputs are taken from the output of the non-inverting sections.

 

[KSTR]

Very interesting !
What would it take to increase max current capabilities and get something usable on a 8 ohms load?
With a 15.5V max voltage that could be up to 30W if current allows it!

Assuming max RMS voltage of brigded config of 14V (15.5 is overly optimistic) you'd get 25W and that would require 6 to 8 OPA564's to have some margin, especially when driving reactive speakers (with current peaks at up to 3 times the purely resistive current demand). Not very economical except for a lab/measurement amp. And then, to tweak out the best distortion specs (in the -180dB realms) to justify the whole effort you'd need to run each of them as its own composite amp with the corresponding higher order feedback loop/compensation. With only a single master opamp for a group of slaves you can run the slaves only in a precisely fixed gain operation in order to parallel them -- with ballast resistors, of course -- and that limits achievable distortion. Would be a nice design exersize, though.

 

[trl]

And they call this innovative, new, patented "technology". Na.

I have a friend that patented a shawarma plate, based on the ingredients/recipe and plate presentation, so I find the THX patent being a bit more wise than a shawarma plate.

 

[JohnYang1997]

I have a friend that patented a shawarma plate, based on the ingredients/recipe and plate presentation, so I find the THX patent being a bit more wise than a shawarma plate.

After several months. Things getting much clearer. The circuit in the patent is innovative in that background. It's a useless topology for ad8397, opa564 etc, which already biased the output stage adequately. The Ahb2 is where the patented circuit got useful. Under biased output stage reduced quiescent standing power and thus the heat, allowing high power in a relatively small package.
The core of the THX 788 789 888 are all feedback. One unique design is the flying supply in the SP200 and HPA4. It's not new (demonstrated in Douglas Self's small signal book) but it's done quite neatly. In fact it's not part of the design in early brochure because the supply voltage was only +-12V. But the latest thx 888 does have +-18V and 788 has +-15V. Both have higher voltage swings. Basically in THX 888, they've done everything right but not at all thanks to the feedforward circuit.

 

[mt196]

I have a friend that patented a shawarma plate, based on the ingredients/recipe and plate presentation, so I find the THX patent being a bit wiser than a shawarma plate.

As far as I remembered, in EU it is not possible to patent recipes as far as they are not related to a specific plate presentation and features (like it must be very specific presentation, list of ingredients and way of displaying it, and if you just change a bit of them, then the patent will not cover it anymore).
I also remember something like KitKats trying to patent their shape and getting the application rejected because it was too similar to the classic shape of a normal chocolate tab and you cannot patent something as broad as that

 

[sepp2gl]

In my professional live before my retirement I used to work as a invention/patent representative for quite some years.
So I would like to un-deadlock this discussion.

What is an invention?
An invention is a technical solution of a technical problem.
Every invention must fulfill some criteria to enable a request for patent.
It must provide a certain level of novelty and innovation, it must be unique and it must be executable.
So nobody can invent a space-ship to Sirius at the time being.

A patent protects specific properties of the invention for commercial exploitation of the patent.
In any case DIY-rebuild for personal use is unrestricted.
There are slightly different legal definitions all over the world, US is slightly different from EU and China might be way special.
But all should have on in common: There need to be a specific description about at least one specific solution, so that an expert of the respective matter might be able to build and verify it.
The protected features of the patent are to be found in the claims (and not in the schematic diagrams, which have a more descriptive character).

If there is any prior art technology, that preceeds the patent, the patent is void.
Prior art means any kind of publication in any media (incl. other patents) before the new patent has been applied.
If this is recognized during patent application the patent will not be granted.
This means that any publication before the issue of the invention to patent office might be overriding the patent.
Also earlier patents of similar solutions will stay in place and limit the validity of a patent.

Last not least the protection of a patent is limited to the countries, where the patent has been granted.
In all other countries the patented solution can be freely commercially exploited (but of course not patented any more).
So even a granted patent does not mean, that nobody may build something similar or even same.

I hope this helps a little to clear up the confusion about patents

kr, sepp2gl

 

[phoenixdogfan]

Guess I need to put mine on US Audiomart now.

 

[Universal Cereal Bus]

In my professional live before my retirement I used to work as a invention/patent representative for quite some years.
So I would like to un-deadlock this discussion.

What is an invention?
An invention is a technical solution of a technical problem.
Every invention must fulfill some criteria to enable a request for patent.
It must provide a certain level of novelty and innovation, it must be unique and it must be executable.
So nobody can invent a space-ship to Sirius at the time being.

A patent protects specific properties of the invention for commercial exploitation of the patent.
In any case DIY-rebuild for personal use is unrestricted.
There are slightly different legal definitions all over the world, US is slightly different from EU and China might be way special.
But all should have on in common: There need to be a specific description about at least one specific solution, so that an expert of the respective matter might be able to build and verify it.
The protected features of the patent are to be found in the claims (and not in the schematic diagrams, which have a more descriptive character).

If there is any prior art technology, that preceeds the patent, the patent is void.
Prior art means any kind of publication in any media (incl. other patents) before the new patent has been applied.
If this is recognized during patent application the patent will not be granted.
This means that any publication before the issue of the invention to patent office might be overriding the patent.
Also earlier patents of similar solutions will stay in place and limit the validity of a patent.

Last not least the protection of a patent is limited to the countries, where the patent has been granted.
In all other countries the patented solution can be freely commercially exploited (but of course not patented any more).
So even a granted patent does not mean, that nobody may build something similar or even same.

I hope this helps a little to clear up the confusion about patents

kr, sepp2gl

Guten Tag!

Where did you practice? I visited some patent firms in Munich last year. Walking to the Deutsches Museum in my time off, I happened upon the most intimidating building I've ever seen and only by coincidence found out it was the EPO. In contrast, I nearly missed the German Patent Office next door because I thought it was an apartment building!

I agree US patent law is different from the EU--I'd say more than slightly, even. Surprisingly, China is quite similar to EP practice. And not just in the patent realm; as I understand it, in modernizing its legal system and proliferating rule of law, much of China's civil code was modeled on various European civil laws.

 

[sepp2gl]

@Universal Cereal Bus:
Actually I worked on the other side of the wall those days.
I was working for an automotive manufacturer in advanced engineering.
There was a lot of stuff, that people invented.
It was my job (among others) to support business decision making, if or not to apply for a patent.
Doing this I learned a lot about, what really matters with an invention.

Congrats for your alias-name. It is great