Hi Patrik,
Fig. 13 from the aforementioned patent has nothing to do with schematics posted. Maybe you were thinking of Fig. 10. Here is actually shown that the feedback is taken from the switching stage with additional low pass filters in feedback and forward paths.
The very reason I am disputing your opinion, is that I came to the Putzeys circuit by thinking how to improve ICEpower VCOM. I thought there is no reason to use low pass filtered output from switching stage, when there is already output filter. Let' s take feedback from output stage.
Only later I realized that this is the same circuit that Subwo1 had already posted. The only difference was that I thought that feedback compensation resistor RL was not necessary, while Subwo1 even included this one. We discussed this in the middle of a very long thread.
I also share Subwo1's opinion about Putzeys patent, but based on my post in another thread. It shows picture from US patent 6,489,841 which predates Putzeys patent. But then again, actual patent claims my be different.
The sad fact is that this circuit has to be the simplest possible implementation of the class D amplifier, and it is not in the public domain.
Best regards,
Jaka Racman
Fig. 13 from the aforementioned patent has nothing to do with schematics posted. Maybe you were thinking of Fig. 10. Here is actually shown that the feedback is taken from the switching stage with additional low pass filters in feedback and forward paths.
The very reason I am disputing your opinion, is that I came to the Putzeys circuit by thinking how to improve ICEpower VCOM. I thought there is no reason to use low pass filtered output from switching stage, when there is already output filter. Let' s take feedback from output stage.
Only later I realized that this is the same circuit that Subwo1 had already posted. The only difference was that I thought that feedback compensation resistor RL was not necessary, while Subwo1 even included this one. We discussed this in the middle of a very long thread.
I also share Subwo1's opinion about Putzeys patent, but based on my post in another thread. It shows picture from US patent 6,489,841 which predates Putzeys patent. But then again, actual patent claims my be different.
The sad fact is that this circuit has to be the simplest possible implementation of the class D amplifier, and it is not in the public domain.
Best regards,
Jaka Racman
Attachments
Hi Pabo,
in my opinion, any 1 resistor, 1 capacitor Schmitt oscillator and the classic capacitor-delay-on-inverting-pin opamp type are basically the same circuit. The output is taken at the input pins of each. For that matter, an unstable audio amp that oscillates is basically the circuit also, just add the output filter before the feedback to slow down the period and to smooth the output. Maybe I just oversimplify things, dunno.
in my opinion, any 1 resistor, 1 capacitor Schmitt oscillator and the classic capacitor-delay-on-inverting-pin opamp type are basically the same circuit. The output is taken at the input pins of each. For that matter, an unstable audio amp that oscillates is basically the circuit also, just add the output filter before the feedback to slow down the period and to smooth the output. Maybe I just oversimplify things, dunno.
Many doubts, that PHILIPS has plans to pursue somebody, which will be using them patented RC (actually one res & one cap) network, probably they wanna just protect himself from outside patents attack. Usually the PHILIPS business it's a tonnes of chips for TV, homecinema, audio etc. (excluding I2C precedent some years ago, as me remember 
)
)
Jaka Racman said:
I'd like to throw in my cent's worth here. The Takagishi patent explicitly uses hysteresis (Vref1+/-dV in the comparator feedback in the picture). This is unnecessary - the delay of the power chain is wholly sufficient to obtain a well defined oscillation. Adding hysteresis will actually produce unwanted extra frequency modulation and adversely affect large-signal linearity of the modulation scheme. Unfortunately for MPS, the patent makes the hysteresis part of its claims.
Hi Mr. Putzeys,
While Takagishi has been granted three patents on the same circuit it is true that all of them use hysteretic comparator. I stand corrected.
I would hovewer like to use this chance and hear your opinion if posted circuit infringes your patent. IMHO it improves on your circuit in two ways:
first it prevents power supply pumping at low frequencies (well known fact that can be considered public domain)
second, by using CC coupling capacitor in filter it forces both halfs of the amplifier to oscillate at the same frequency and phase. This canceles ripple in differentially coupled load as can be seen in this simulation .
Since I have no commercial interest in this design, I would like to proclaim it as public domain, but only if you feel that it does not infringe your patent.
Best regards,
Jaka Racman
While Takagishi has been granted three patents on the same circuit it is true that all of them use hysteretic comparator. I stand corrected.
I would hovewer like to use this chance and hear your opinion if posted circuit infringes your patent. IMHO it improves on your circuit in two ways:
first it prevents power supply pumping at low frequencies (well known fact that can be considered public domain)
second, by using CC coupling capacitor in filter it forces both halfs of the amplifier to oscillate at the same frequency and phase. This canceles ripple in differentially coupled load as can be seen in this simulation .
Since I have no commercial interest in this design, I would like to proclaim it as public domain, but only if you feel that it does not infringe your patent.
Best regards,
Jaka Racman
Attachments
Jaka Racman said:
Hi,
Not so much up to me to decide that, it's Philips who decides to what extent to enforce the patent.
Legally speaking, when someone invents an improvement on a patented invention they can patent the improvement with reference to the original patent. The improvement is then their ip. License fees are split between the holder of the original patent and the holder of the new patent in an arrangement to be negociated among those two parties.
In this particular case, the improved circuit can no longer be patented because by the simple act of publishing it on this forum you've already made it public (well I've seen similar things float around before too). Another reason would be that we have already been using that arrangement in products (philips mini sets, nothing fancy) for quite some time.
Now "public domain" and "public domain" are two things. Inventors don't normally enforce their patent rights on individual experimenters. That would go against the spirit of the system, which is to encourage information sharing while still allowing inventors to get a fair return for their work. Now, if a company were to use the circuit in mass-produced items, Philips would most certainly knock on their door for royalties based on the original patent. In short, patenting means "to publish while asserting your rights as an inventor."
All this means nobody will stop you from actually publishing circuits based on a patented invention. Making money from them is a different affair.
Cheers,
Bruno
Hi Mr. Putzeys,
while I admire your work, I can not share some of your legality thoughts.
First I find very unlikely that anyone patenting improved circuit would be willing to share royalities with the holder of the patent that was improved. Anyone writing a patent application is smart enough to try to present his invention as something completly new and best thing since sliced bread. For one thing, I do not think that you consider your circuit as an improvement of Takagishi patent, and that Philips is willing to pay royalities to MPS.
Second, for real patent protection, WO (PTO) patent number means almost nothing. It only means that you have 18 months priority time to file national patent applications for each country where you expect your sales profit to be larger than the cost of the national patent application and maintaining fee. While this would be no problem for Philips, I still think they will not do that in every country of the world.
Best regards,
Jaka Racman
while I admire your work, I can not share some of your legality thoughts.
First I find very unlikely that anyone patenting improved circuit would be willing to share royalities with the holder of the patent that was improved. Anyone writing a patent application is smart enough to try to present his invention as something completly new and best thing since sliced bread. For one thing, I do not think that you consider your circuit as an improvement of Takagishi patent, and that Philips is willing to pay royalities to MPS.
Second, for real patent protection, WO (PTO) patent number means almost nothing. It only means that you have 18 months priority time to file national patent applications for each country where you expect your sales profit to be larger than the cost of the national patent application and maintaining fee. While this would be no problem for Philips, I still think they will not do that in every country of the world.
Best regards,
Jaka Racman
Jaka Racman
You may be right in practice but in theory the patent holder of the improved version of a patent may not have the right to use his own patent since it is based on a previous patent. Because the holder of the previous patent may not use the improvements made by the new patent holder an agreement could very well be made between them. I have had such discussions between my company and patent holders.
You may be right in practice but in theory the patent holder of the improved version of a patent may not have the right to use his own patent since it is based on a previous patent. Because the holder of the previous patent may not use the improvements made by the new patent holder an agreement could very well be made between them. I have had such discussions between my company and patent holders.
Bruno Putzeys
Nice to see that such famous designers are to be found in this forum.
One thing that I don´t understand about your technique is how you can avoid having capacitive loads affecting the switching frequency. Lets say that you drive a electrostatic speaker with a uF capacitance, wouldn't that lower your switching frequency to below 50% of the original switching frequency?
Nice to see that such famous designers are to be found in this forum.
One thing that I don´t understand about your technique is how you can avoid having capacitive loads affecting the switching frequency. Lets say that you drive a electrostatic speaker with a uF capacitance, wouldn't that lower your switching frequency to below 50% of the original switching frequency?
This is quite correct. However, the holder of a previous patent may feel otherwise about the new invention and claim compensation or file an infringement lawsuit.Jaka Racman said:
This is by no means an automatic mechanism. If you have a patent but you don't defend it, nothing happens! Worse still, failure to vigorously defend a patent significantly weakens your position for subsequent infringement suits concerning that patent.
Correct. Each national application is handled by the local patent office and is subject to examination every time. It is not uncommon for patents to get accepted in some countries and thrown out in others.
Now, please note I'm only telling you how the system works. It's not a reflection of my personal sentiment which lies more along the same lines as most posters' here (it would give me more satisfaction to see UcD become a kind of public domain standard circuit). However, both the company I work for and our licensees are better served by the IP protection put in place. It allows them a return on their investments and it allows me to continue my research (=hobby) as a daytime job.
This to say I don't find the whole discussion about patents awfully thrilling. Legal matters are out of my hands anyway.
Now, if everyone would stop calling me "Sir" or "Mr. Putzeys", we'd be talking.
Cheers,
Bruno
I have seen tests performed on the UCD by a German mag (I can't remember if it was "Stero" but Bruno will most probably know, because he visited them back then) showing very little load-dependancy (amongst all the other excellent performance figures).
While this is definitely a must for an amplifier that will be used by John Average (connecting anything but an "decent" load to them) one could even take things a step further:
I am a fan of active concepts (and I assume that I am not the only one here) where the load on each power-amp is previously known. In this case each output-filter can be purpose-built. It could even be deliberately dimensioned to get the desired performance into a capacitive load only (i.e. an ESL). The extreme solution in this case would indeed consist of a series inductor only (O.K. some more circuitry would be needed in order to pass any EMC tests, but I assume it is clear what I want to say).
Regards
Charles
P.S. Bruno: Welcome to this forum.
While this is definitely a must for an amplifier that will be used by John Average (connecting anything but an "decent" load to them) one could even take things a step further:
I am a fan of active concepts (and I assume that I am not the only one here) where the load on each power-amp is previously known. In this case each output-filter can be purpose-built. It could even be deliberately dimensioned to get the desired performance into a capacitive load only (i.e. an ESL). The extreme solution in this case would indeed consist of a series inductor only (O.K. some more circuitry would be needed in order to pass any EMC tests, but I assume it is clear what I want to say).
Regards
Charles
P.S. Bruno: Welcome to this forum.
That loudspeakers "expect" to see a voltage source is largely a matter of convention. I've heard it argued (Hawksford?) that electrodynamic drivers exhibit lower distortion when connected to a current source. Of course all electroacoustic parameters change, so the design method of a system changes too, but that would be a prime candidate for an application specific amp - a current source.
Electrostatics are another point in case. They thrive best when driven by an equalised (differentiated) current source. When in rest, the ideal case would not allow any charge to flow from one stator to the other, electrostatically suspending the diaphragm. The diaphragm would be stable, theoretically, even when no mechanical tension is applied. It would become less acoustically transparent, thus making it hopefully less sensitive to placement.
Who takes up the gauntlet?
Electrostatics are another point in case. They thrive best when driven by an equalised (differentiated) current source. When in rest, the ideal case would not allow any charge to flow from one stator to the other, electrostatically suspending the diaphragm. The diaphragm would be stable, theoretically, even when no mechanical tension is applied. It would become less acoustically transparent, thus making it hopefully less sensitive to placement.
Who takes up the gauntlet?
This has already been discussed endlessly. I generally agree with you but I still have to come across a clever and simple idea to achieve control around fs (O.K. I had some ideas but didn't experiment with them).
At the "upper end" the LPF pole built by Lvc would be cancelled as an additional feature.
But for an amp/driver combination being capable of taking advantage of the latter, one must use a very high switching frequency if one wants to do it with class-d.
Regards
Charles
Hi Charles,
You may want to look at the following Karsten Nielsen's AES article .
You will find an interesting concept of mounting filterless class d amplifier directly on the loudspeaker. Since amplifier is filterless it can be made voltage or current source very easily.
I have once changed my modified Grandson of Ampzilla (stable into less than 1 ohm loads, 50 A of output current) into current source. While my intention was to run a high speed electromotor, I have connected my Maggies to it. Although Maggies are almost perfectly resistive speakers, I did not like the sound. But then again, sound was not so radically changed, less if I would change the amplifier for another brand.
Bruno, I have a gut feeling that full bridge version of UCD is not so sensitive to capacitive loads, since oscillating frequency is dependent on filter common mode capacitors which remain the same. Since Philips is using full bridge in production allready, it has priority and I think there is still chance that you include full bridge version into your national patent applications.
BTW, there was mistake in the schematics I posted, feedback shoud have been taken from output capacitors, not before the inductor.
Best regards,
Jaka Racman
You may want to look at the following Karsten Nielsen's AES article .
You will find an interesting concept of mounting filterless class d amplifier directly on the loudspeaker. Since amplifier is filterless it can be made voltage or current source very easily.
I have once changed my modified Grandson of Ampzilla (stable into less than 1 ohm loads, 50 A of output current) into current source. While my intention was to run a high speed electromotor, I have connected my Maggies to it. Although Maggies are almost perfectly resistive speakers, I did not like the sound. But then again, sound was not so radically changed, less if I would change the amplifier for another brand.
Bruno, I have a gut feeling that full bridge version of UCD is not so sensitive to capacitive loads, since oscillating frequency is dependent on filter common mode capacitors which remain the same. Since Philips is using full bridge in production allready, it has priority and I think there is still chance that you include full bridge version into your national patent applications.
BTW, there was mistake in the schematics I posted, feedback shoud have been taken from output capacitors, not before the inductor.
Best regards,
Jaka Racman
Hi Jaka
I already know that you can drive dynamic drivers by a PWM signal directly, if it is tri-state this works even better (as proposed in this paper).
Without seeing actual figures from other sources as well I am sceptical however. It is a B&O paper after all !!!!
Furthermore you don't overcome the limitations given by the Nyqvist frequency, so you still need a high switching frequency if you want to compensate for a tweeter's (or FR's !!!!!!) Lvc.
Regards
Charles
I already know that you can drive dynamic drivers by a PWM signal directly, if it is tri-state this works even better (as proposed in this paper).
Without seeing actual figures from other sources as well I am sceptical however. It is a B&O paper after all !!!!
Furthermore you don't overcome the limitations given by the Nyqvist frequency, so you still need a high switching frequency if you want to compensate for a tweeter's (or FR's !!!!!!) Lvc.
Regards
Charles
Here I agree with you 100%. As an illustration of their technical competence I attach picture from their patent application. It seems they like half the supply voltage present on their speaker at all times
. However, it would be a relatively simple task to create single supply filterless current source amplifier by use of triangle type class BD modulator and average current mode control. Loop compensation is simple because of the single LR pole rolloff.
Best regards,
Jaka Racman
