I'm thinking that we should have a stickied poll for members to give score for, say, Top 10 DIY amplifiers from 0-10 so that we know what to expect when we build them. This can be separate for each forum: Tubes/Valves, Solid State, etc.
Of course we expect only members who have actually built and listened to them to give score. Even better, members who have compared at least two of the amplifiers on the Top 10 list.
I'm getting sick of reading someone making comments like: "X amp sounds wonderful but when i made this change, it sounds even much more wonderful". How do you quantify "much more"?! How is it any different with "slightly" when it comes to something so subjective?
Discussing audio quality can be frustrating, even could very well be flame-inducing on internet forums. Why not we rely on statistics to solve the problem?
Of course we expect only members who have actually built and listened to them to give score. Even better, members who have compared at least two of the amplifiers on the Top 10 list.
I'm getting sick of reading someone making comments like: "X amp sounds wonderful but when i made this change, it sounds even much more wonderful". How do you quantify "much more"?! How is it any different with "slightly" when it comes to something so subjective?
Discussing audio quality can be frustrating, even could very well be flame-inducing on internet forums. Why not we rely on statistics to solve the problem?
I agree audio quality is very subjective. That's why i propose we should make statistics to average it out. One member saying "X Amplifier" sounds fast and wide and another saying "Y Amplifier" sounds warm and tubey is less useful than 25 members averaging the score of "X Amplifier" to 8.5 below "Y Amplifier" which gets 8.75 average from 15 members.
Those of you familiar with Mobile Network RF Measurement might recognize something called SQI (Speech Quality Index). SQI is a Mean Opinion Score system developed by Ericsson to assess the quality of calls made on mobile network. Basically Ericsson varies the Frame Error Rate, Bit Error Rate and Speech Codec on a radio frequency used to transmit speech and a scoring poll is taken by a number of listeners. RF environment with the the least BER, FER and best Speech Codec would statistically show high SQI score. Why can't we do the same with DIY Amplifiers?
As popular as the F5 has been -- I will bet that there were more SWTP "Tiger" amplifiers built than anything on DIYAUDIO or ELEKTOR. It was a really popular design and discussed in the now-defunct "Popular Electronics" -- if you mention any of its shortcomings on DIYAUDIO its proponents emerge from the woodwork.
In the list, one would have to include Hafler, although these were more "bolt-em" together with minimal soldering.
There were probably quite a few Williamsons made as well.
In the list, one would have to include Hafler, although these were more "bolt-em" together with minimal soldering.
There were probably quite a few Williamsons made as well.
If you can make a list of 10 power amplifiers that amongst all the members here you can find 10 of each built, I'll be very surprised.
Sure, there will be some really easy ones, F5, Aleph 30, BrianGT chipamp... But 10 different amps with 10 of each constructed? That's a taller order than you may think.
And there are many, many differences in construction and parts selection / availability, so at what point does that come into play ?
Sure, there will be some really easy ones, F5, Aleph 30, BrianGT chipamp... But 10 different amps with 10 of each constructed? That's a taller order than you may think.
And there are many, many differences in construction and parts selection / availability, so at what point does that come into play ?
Last edited:
If a thread for a certain Power Amp has reached, say, 1000 replies, i doubt that less than 10 people has actually build it.. So perhaps we should make a certain condition that has to be fulfilled before an amp makes it into the Top 10? Also it doesn't even have to be Top 10!
I guess one should take the poll result with a grain of salt. I mean, as long as the main topology is followed, isn't that worth something? I would then expect AKSA and P3A to have near score as they're basically similar topology.
I guess one should take the poll result with a grain of salt. I mean, as long as the main topology is followed, isn't that worth something? I would then expect AKSA and P3A to have near score as they're basically similar topology.
even objectively this is not likely the case
there are implementation issues at every point from the mains power entry to the speaker output
amps can be built with bad or better construction, parts placement, wiring pairing/twisting, gnd hierarchy...
and that would be assuming identical parts rather than some built with EI and others toroidial mains xfmrs, different value, specs and brands for diodes, caps...
then you think subjective commentary on amplifiers is somehow going to be "valid" with differing speakers, rooms, no common reference to compare against even if you could convince more than a few to use any psychoacoustic controls?
If a thread has 1000 responses you assume there are at least 10?
I am sure that's not so. I've been selling amplifier parts kits for which there are PCB available, as well as PSU pcb, and multiple chassis options, including some that require no tools other than screwdriver and soldering iron.
Of those kits I am sure the completion rate is less than 10%, and this is of people who have purchased the parts!!
I admit this is a long shot. It's just that those overrated comments about the sound of some amplifiers here are sickening. I doubt that in a controlled blind test environment, those amplifiers would stand out from the rest of the amplifiers here.
Or will they? Hence my idea of having this score polls. I guess i'm just itching to say "Shut up, your amplifier is not "gazillions" better than the other amps here".
If someone should ask "If there's only one amplifier that you can build here, which would it be?" so far i'm a firm believer that the answer should be "Choose randomly and build it properly". All of these "1000V/uS slew rate", "current feedback", "diamond buffer", "current mirror", etc and yet a simple Lin topology amplifier like P3A remains famous? Surely there's some conclusions to be made from this fact?
Or will they? Hence my idea of having this score polls. I guess i'm just itching to say "Shut up, your amplifier is not "gazillions" better than the other amps here".
If someone should ask "If there's only one amplifier that you can build here, which would it be?" so far i'm a firm believer that the answer should be "Choose randomly and build it properly". All of these "1000V/uS slew rate", "current feedback", "diamond buffer", "current mirror", etc and yet a simple Lin topology amplifier like P3A remains famous? Surely there's some conclusions to be made from this fact?
I agree with "build it properly", but don't necessary agree with "Choose randomly"
System building must be to requirement, and most people's strongest requirement is their speakers and the SPL they listen to.
Somebody with LS 3/5A speakers who is desperate wanting to build a 2A3 single-ended amplifier needs to understand that the combination will not work. Just as somwbody with insanely efficient horns is not going to find a chipamp even listenable...
Anyway, build the amp to your speakers. It always works out well if thought of that way.
how about a poll to decide whether we should make a poll? 
By the way, jackinnj mentioned about some article in Linear Audio about Mean Opinion Score in amplifiers. Would be great if i can get the link to it. I tried searching to no avail. Sorry if this is becoming out of topic.
Yes of course i meant random to an extent. Nobody sane would build Apex's 1000 Watt PA amplifier for elevator music.
By the way, jackinnj mentioned about some article in Linear Audio about Mean Opinion Score in amplifiers. Would be great if i can get the link to it. I tried searching to no avail. Sorry if this is becoming out of topic.
I was thinking something simple, like opinions from those actually building different DIY versions, not necessarily or any need for ABX testing. Build experience and perception of Performance would suffice, well unless those involved would like to send a sample of each to Jan, there his ABX scientist could perform such an endeavor for print ...
what purpose would these builder's opinions of their own work, in isolation from any reference standard be useful for?
Isn't that what many threads are already full of?
how about high res ADC of the amp terminal output driving real speakers with a agreed on cut of music?
then we could share around the files, listen on good headphones - after standardizing level, weeding out obvious fails with DiffMaker
Isn't that what many threads are already full of?
how about high res ADC of the amp terminal output driving real speakers with a agreed on cut of music?
then we could share around the files, listen on good headphones - after standardizing level, weeding out obvious fails with DiffMaker
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.