I never said I am against it, I have said I have used it many times when I wanted to create a amp quickly and they sounded great. All i said was that 1. It's 20 years old and uses a quasi complementary output stage which is obsolete. Its THD is unimpressive at 0.03%. Its saving grace was the very low IMD, low noise and exceptional linearity. Things have progressed over 20 years and If a IC designer was designing a class B chipamp today they can easily use a fully complementary output stage and use other techniques which can give THD+N of less than 0.0002% even. Sadly that will never happen as there is no commercial market for such a chip as every manufacturer has moved onto class D amps. So the LM3886 and the TDA7293 is all we will have until they are obsolete very soon. GaN FET based class D amps are the the latest and soon there will not be a convincing reason for mainstream manufacturers to use anything other than Class D for any power level.
Considering the 0.03% thd figure is at 20khz, it's not that bad. The biggest problem of the lm3886 is highlighted in those measurements of the Akitika by Stereophile. At highish power into 4r, thd will rise badly at lower frequencies, probably due to the protection system.
But used within its limits, it makes for a decent little amp.
But used within its limits, it makes for a decent little amp.
although I woul use dc-servo-opamps with this. But then maybe they are under the PCB as SMDs. Who knows......
https://www.akitika.com/documents/AssemblyManualGT101rev1p50.pdf Schematics on pages 39-40. Again, a completely different design to anything else.
The only real originality of the Akitika is the regulated PS. The amp itself is a datasheet implementation of a single supply lm3886.
The use of a regulated PS is questionable imho. From the published measurements, it doesn't offer a lot compared to Tom's LM3886DR or what I've measured on unregulated supplies.
The use of a regulated PS is questionable imho. From the published measurements, it doesn't offer a lot compared to Tom's LM3886DR or what I've measured on unregulated supplies.
"All parameters are measured using a ±30 V regulated power supply unless otherwise noted."
Tom used a regulated PSU for most of the published measurements of his LM3886DR.
What I mean is that every designer has developed different solutions. I am not saying the solutions are good or something I would use.
Tom used a regulated PSU for most of the published measurements of his LM3886DR.
What I mean is that every designer has developed different solutions. I am not saying the solutions are good or something I would use.
It's the expected power from a lm3886 on +/-36v rails. The only advantage I see with regulated supplies and the lm3886 is that you have no sag and thus can extract a bit more power without needing even higher rails (and thus higher dissipation and problems). That's what I had in my mind with "doesn't offer a lot" because honestly 80w vs 100w doesn't mean much in real life use.
And at 72v supply, the akitika might already be running into heat/current problems with the spike protection into 4r way before 100w at low frequencies, as exemplified by stereophile measurements.
I'm not saying it's a bad implementation, it's one of the best. But there's no going around the limitations of the lm 3886.
@Mark: I'm pretty certain Tom said somewhere that the lm3886dr doesn't perform very differently on unregulated supplies and that's been my experience too.
I agree with all you say. I Just think it is interesting to see the different approaches. It doesn't get much better than your design, if you are looking for a simple single chip solution. Have you had much feedback from DIY builders? Would be nice to see a 4 or even 6 channel build.
Last edited:
There is a 10000uf capacitor in series with the output. This is going to cause increased distortion and impaired low frequency response. It's a bad design choice necessitated by the single supply rail decision.
A number of Nelson’s design are capacitor output and are stellar. But given the choice of single or dual rails for the 3886 (ie cap or no cap), i’d choose no cap.
If the LM3886 is similar to LM3875, there is a rail voltage optimum for no current liiting into 4Ω +/-24=-25v for the 3875). With modern speakers, that is a good idea. I see little point in goin more. I do have a 3875 retrofitted into a Technics amp, and it is really only suitabe for 16Ω speakers.
dave
If the LM3886 is similar to LM3875, there is a rail voltage optimum for no current liiting into 4Ω +/-24=-25v for the 3875). With modern speakers, that is a good idea. I see little point in goin more. I do have a 3875 retrofitted into a Technics amp, and it is really only suitabe for 16Ω speakers.
dave
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.