I don't necessarily dispute that, jd, I was just too lazy to look any further for an explanation.
w
The advertized THD is 5.50 ppm, at 1kHz, 1 watt into 8 ohm. This is from the spec sheet on the web link provided in post #1. Close to the 1ppm league, thus. Truly amazing ! Is a NE5532 really capable of this ? Not on a light load, but delivering a few milliamps into the output like in our case ? Hey Mister Douglas Self, what's the magic behind this ? Did you intend getting linearization using massive parallelization ? Did you intend a Class A domain extension using the 64 individual DC offset mismatches and the 1 ohm coupling resistors ? Looks very promising. Elektor has not delivered a single clue about this yet. Will you tell us ?
Last edited:
Hi nmiljac, this design made by Jan Meier dating back from 10 years ago looks more elaborate than the Elektor - Douglas Self design. There is a differential control amp with moderate gain at the input working with a reduced common mode voltage, used as global feedback. Each paralled voltage follower buffer is involved into a distributed symmetric class A extension scheme.
Hi Mooly, can you tell if Douglas Self is mentionning this prior art in his newest book ?
Last edited:
Only the use of 5532's in parallel were given a passing mention... I never expected to see a fully worked design appear anywhere... I'm sure there was no reference to the amp in the link... which I had never seen before actually
me too, it has some weirdo qualities and I would expect the design coming rather from an amateur than from a professional or engineer.
regards
I like the idea of using multisource opamps so one can chose in function of taste and budget. I like the idea of putting 64 to 256 opamps in parallel. I like the idea of using SMT devices for getting a compact footprint. I like the idea of one to four x86 processor radiators for the cooling, with or without fan, with or without heatpipe, with or without liquid cooling. I like the idea of using a modified 120 watt laptop switchmode power supply delivering 36V instead of 19 V. I like the simplicity and the power extension of the single supply floating bridge. I like the Jan Meier overall feedback and distributed class A extension. I would like to combine all those features into a small brick amp, specified at 1ppm THD. Looks feasible. But, at the end of the day, will I like the sound that gets delivered ?
Can you explain ? What are the fundamental distinctions or criterions between professional, engineer and amateur ?
That's the killer question isn't it.
256 opamps, say each draws 10 milliamp... I'll let you work it out
I wondered whether this design could be developed to say include "helper" transistors to provide peak current demands into adverse loads when needed, yet keeping the opamps as the main devices for all other times/conditions.
At the moment I'm in a big doubt about this. Maybe we need to distinguish between the subjective results and between the 1ppm THD goal. It could be that the inverting version, not subjected to common-mode voltage, hence possibly delivering less than 1ppm distorsion, sounds ugly in comparison with a non-inverting version delivering around 4ppm THD. And we now have the Jan Meier design, featuring a middleman approach with a global feedback enduring moderate common-mode (the moderate gain differential amp) encompassing non-inverting buffers seeing a strong common-mode. On top of this, the Jan Meier distributed class A symmetric extension needs to be carefully assessed. Regarding the power supply, Douglas Self seems to be using a 300 VA transformer for two channels, hence 150 VA per channel. Isn't this overkill ? Say there are 64 opamps per channel, each eating a 5 mA quiescient current. This is only 320 mA in total hence 12 watt on a 36 V supply. Where do the remaining 120 watt go ? Step to 256 opamps per channel, and you get a 1.28 amp quiescent current. On a 36 V supply, this means 46 watt waste. That's not horrific. Consider each opamp being able to deliver 25mA current. With 256 opamps in parallel, you get 6.4 amps. This is decent. A class A design would sink alot more quiescent current.
Last edited:
In the Elektor article they said that the original design contained two electrolytic capacitors that were removed due to "issues". I couldn't work out whether the Elektor team had changed the design, or whether Self did it himself.
In the later article where they tested electrolytics for distortion they came to the same conclusions as Self; they don't cause distortion as long as you make them 'big' enough. But, that being the case, they still didn't say why the original circuit was changed! Did Elektor change it for no good reason, and prove itself guilty of audiophoolery?
In the later article where they tested electrolytics for distortion they came to the same conclusions as Self; they don't cause distortion as long as you make them 'big' enough. But, that being the case, they still didn't say why the original circuit was changed! Did Elektor change it for no good reason, and prove itself guilty of audiophoolery?
one difference is striving for efficiency...no waste of materials, labor, cost, energy etc.
The Self design features:
-high parts count
-high parts cost
-costly and space wasting PCB
-high quiescent power consumption
-low output power
it seems like being a nice case study or gimmick for a magazine like Elektor, no complains from me...yet I wouldn't expect it coming from Mr. Self
regards
The Self design features:
-high parts count
-high parts cost
-costly and space wasting PCB
-high quiescent power consumption
-low output power
it seems like being a nice case study or gimmick for a magazine like Elektor, no complains from me...yet I wouldn't expect it coming from Mr. Self
regards
An Amateur is someone who is following a topic from which he/she does not earn a significant income and who may or may not have a great knowledge of the Topic.
A professional is one who earns a substantial proportion of his income from working with the Topic. He may know much or nothing or anywhere in between about the Topic.
An Engineer is qualified in the Topic or a branch of that same topic and has proved to his/her institution that they are competent in the topic and/or branch. The Engineer must also prove that he/she is following a course of continuous self development in the topic they are currently involved with.
These qualifications and self development demand that the Engineer knows what he/she is commenting on and that they will never deliberately mislead/defraud any one studying the topic and work for the good of the community they serve.
It is for these reasons that EUR ING and similar are held in some esteem in many countries, except the UK, where everyone that dabbles with a screwdriver can call themselves an "engineer" due to the lack of protected status for the "Engineer" title. c.f. "Doctor", "Architect", "Accountant" etc which are protected in the UK.
Last edited:
I have noted the strange local supply decoupling scheme, each NE5532 being decoupled with one capacitor connected between V+ and V- without touching the ground. However, in the Elektor article I can't find any info about two electrolytic capacitors needing to be removed. Can you provide a web link ?
Last edited:
That's standard practice.
It was in the paper edition (I don't have a copy, I read it at the library). Somebody here will have a copy though, I expect.
Last edited:
Dead right ! Would be interesting to know the inside story, and why and how Elektor got involved in this NE5532 project knowing there is already a company specializing in the Blameless PCBs, connecting to Douglas Self. The Signal Transfer Company: Compact Blameless Power Amplifier
Over there, they feature the Compact Blameless amplifier, conservatively specified at 6ppm THD 1kHz 25 watt into 8 ohm, and measured using an Audio-Precision SYS-2702 at 4ppm THD 1kHz 25 watt into 8 ohm. Measurement bandwidth is 80 kHz. A 22 kHz BW would deliver a lower THD figure. This, using +24V and -24V supplies. They don't say if they need to be regulated for getting such low THD figure.
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.