Waly its not specific to my implementation. When I design I well call it spy a little on what other manufacturers are doing. I noticed they have exactly the same situation. I used to see compensation used when Ken Ishiwata from Marantz was using triples with LTP based designs coupled to folded cascode vas. From about 2004 he exclusively uses CFB designs and the compensation is no longer used. I notice this with NAD and Accuphase as well. I havent been able to pinpoint the reason yet.
This may be look to other amplifiers that UFOs are to planes.
Aim :
The starting point was the MJR6, which was just an attempt to determine the simplest possible class-B (or class-AB) amplifier circuit with adequately low distortion. [...]
MJR7-Mk5 Mosfet Power Amplifier Design Notes
It happened that, regarding harmonic distorsion, the design, using 7 active devices only, is on par with the best... Or even better, using 9 :
MJR9
It shows that what is important is not the topology but what you are able to do with it.
I've built 3 stereo MRJ7s and like them very much.
No truth at all in the asserion that so-called "CFAs" are more linear than VFAs. No truth whatsoever. Period.
Michael stop playing semantic games which are more applicable at a kindergaren playground. The assertion is not whether a complete CFB design is more linear or not, its the diamond buffer input circuit is more linear than a LTP, period. This increased linearity allows CFB to have similar distortion performance to VFB with less loop gain.
How many of the amps you quoted have fast output devices (like the 30MHz RETs)? A triple has barely any stability problems when using slower devices.
Could you please post the schematics you quoted?
No but I can get close and with reduced component count, using a newer CFB topology which is unfortuneatly patented I can but this would then be more complex than your amp. This is all arbitary as if you look at a completed amp like astx effort you get a better idea of what is going to be possible in real life. This goes for my CFB too, housed in a chassis and complete I cant get it measured to better than 12 ppm and this with just a conventional CFB design and the most evil of compensation techniques which is shunt.
Last edited:
All of them use recent Sanken devices as do I. The ones I use are rated to 60 Mhz.
Many of these schematics are available online, Im not sure if its legal I post them here. Go over to hifiengine and download any marantz amp design from 2004 onwards.
Last edited:
Is it not the Self's lesson that the low power stages non-linearities should be almost entirely swamped under the output stage distorsions ?
I'm going to build MJR7 as well. Nice design with 'no sand' in the feedback summing node.
Which is true but why change the subject of the post, it doesnt change the fact that a diamond is more linear than a LTP.
it appears some are operating from a "CFA fanboy" perspective - any discussion of real trade offs, any practical limits, placing technical performance in broader context is an "attack" in the fanboy world
really counterproductive in a thread supposedly intended to sway Doug Self's opinion as a technology writer on the whether he should include CFA analysis in his Audio Power Amp Design book in the future
really counterproductive in a thread supposedly intended to sway Doug Self's opinion as a technology writer on the whether he should include CFA analysis in his Audio Power Amp Design book in the future
It uses either TMC or TPC which is diffrent to standard blameless amp and helps its THD performance. It is also rated for high power and low impedance loads, the higher voltage and parralell pairs of output devices contribute effectively to reduce THD.
With fanboy I hope you mean objective views. I wont get envolved there. I ve heard good sounding amps from both toplogies. I just find it easier to design a amp with CFB architecture with class equaling performance compared VFB, stability and less complexity.
Its obvious Self should include CFB, he is old school and unflexible, same happens with tube lovers, they still swear that tubes are the way to go. Once again there is no technical or psycoacoustic evidence why a audio amp should only be built using tubes or for that matter using VFB topology exclusively.
Similar numbers are achieveable and have been for many years, the reason I mentioned the amp by astx. Before the advent of complex compensation schemes like TMC ect and advanced circuirtry as seen on opamps like lme4562 , VFB amps could not get close to high frequency THD performance of VFB for many years. This is the prime reason CFB atracted attention.
A good start is to see patents from Gosser from Analog Devices, also study feedfoward techniques as used in the texas and national CFB opamps, the current mirror youve already discovered.
But then again who wants an audio amplifier with a zillion transistors. For me only it being NGFB could justify.
Hi Damir,
As you probably know, the superTIS with its VFA front-end gives very low distortion: 61ppb @ 20kHz and 3.1ppm @ 200kHz. I'm unaware of any CFA-TIS combo that performs better.
Cheers, E.
Time to give it a rest...Edmond that is because there is none, this is not disputed, ultimate low THD is for the time being achieved with VFA but you pay a heavy price in complexity with a design like yours. Do you believe a commercial concern would be swayed with such a complex monster.
I hope you build it but I think you are going to be dissapointed when you dont achieve those numbers in the real world. Heres a challenge, show if it is possible.
An aquaintence suggested I build it to try the concept, I refused. The acquaintance is unfortuneatly also my boss so Im still hoping he doesnt play that trump card and force me to build it.
I hope you build it but I think you are going to be dissapointed when you dont achieve those numbers in the real world. Heres a challenge, show if it is possible.
An aquaintence suggested I build it to try the concept, I refused. The acquaintance is unfortuneatly also my boss so Im still hoping he doesnt play that trump card and force me to build it.
Last edited:
I checked the 2008 Marantz PM5003. 40W/channel, with only one pair of 20MHz devices (2SC4467/2SA1694), tamed in the base with no less than 10ohm, and a brutal shunt compensation (470pF at the VAS output, practically a short (17ohm) at 20MHz, hence the rather modest distortion spec) it is no surprise it's rock solid stable.
Nothing to do, triple stability wise, with the current feedback topology. Point me to a high power (3 pairs of outputs in parallel), current feedback, with high sped triple, and without any driver BC cap. I can't find any.
Its true that Marantz overcompensate all their amps. The shunt is no surprise for CFB that is the usual and near mandatory compensation. The point is not often even with such drastic measures a VFB amp can get away without using CB in drivers. See also pm9s2 with faster devices and power, Cyrus 3 is another example.