Hi Gaetan,
Good question!
You start here by selecting a quiescent current for the drivers. On the AKSA I use 5.5mA, 8mA for two output pairs. This quiescent current should be at a level producing around 0.2-0.5W dissipation in the drivers.
Then you 'tune' it with a suitable cap. Start with 82nF, and go all the way up to 1uF. Using the drivers and output specified, it's typically 100nF, as Self uses. Once you have the best 'sound', and it should be tuned by ear as the CRO waveforms are not very reliable as sonic indicators, go back to the resistor and tweak it for the best possible sonics.
This is slow, crude work. The resistor will typically be around 220R for a single pair, and gets lower with more output pairs.
The size of the resistor, the drivers and output devices chosen, all work together with the cap to give the best sonic results.
Cheers,
Hugh
Good question!
You start here by selecting a quiescent current for the drivers. On the AKSA I use 5.5mA, 8mA for two output pairs. This quiescent current should be at a level producing around 0.2-0.5W dissipation in the drivers.
Then you 'tune' it with a suitable cap. Start with 82nF, and go all the way up to 1uF. Using the drivers and output specified, it's typically 100nF, as Self uses. Once you have the best 'sound', and it should be tuned by ear as the CRO waveforms are not very reliable as sonic indicators, go back to the resistor and tweak it for the best possible sonics.
This is slow, crude work. The resistor will typically be around 220R for a single pair, and gets lower with more output pairs.
The size of the resistor, the drivers and output devices chosen, all work together with the cap to give the best sonic results.
Cheers,
Hugh
This thread has gone way off topic.
I once had the opportunity to compare a Peter Daniel Gainclone (LM4870, IIRC) to an Aleph 60W Class A amp. My assessment was that the GC was more detailed and precise. Without something to compare it to, I would have had nothing to complain about wrt to the Aleph.
The only problem with the GC is its performance when clipping. It sounds really bad then (referring to the National Semiconductor chips). So, if operating within its limits, I would recommend a GainClone.
I would just like to add that there is a difference between a good sounding amplifier and one that is accurate. The best thing an amplifier can do is to impart no signature to the signal. Technically speaking. Amplifiers that have "tube sound" or are "the best sounding" may not be all that accurate. But this does not mean that many do not prefer them over better, more technically pure designs.
I once had the opportunity to compare a Peter Daniel Gainclone (LM4870, IIRC) to an Aleph 60W Class A amp. My assessment was that the GC was more detailed and precise. Without something to compare it to, I would have had nothing to complain about wrt to the Aleph.
The only problem with the GC is its performance when clipping. It sounds really bad then (referring to the National Semiconductor chips). So, if operating within its limits, I would recommend a GainClone.
I would just like to add that there is a difference between a good sounding amplifier and one that is accurate. The best thing an amplifier can do is to impart no signature to the signal. Technically speaking. Amplifiers that have "tube sound" or are "the best sounding" may not be all that accurate. But this does not mean that many do not prefer them over better, more technically pure designs.
AKSA said:
Hello
You show me very importants facts about the cdom cap.A to big cap will sound slow, and lifeless, a too small would sound bright, tizzy, fatigueing.
So for the switch-off speedup capacitor, there should be some known sonic differences between a too small value capacitor and a too big value capacitor. What would be those sonic differences ?
I presume that a too small switch-off speedup capacitor can not suckup enough from the drivers transistors, and a too big capacitor will suck too much voltage even wen they are switch-on (by bypassing the drivers center resistor at some frequencies) ?
The more I read you and the more I understand the hard works you did to tweak your Aksa, LF and Soraya amps, you did go some step more than just doing maths and simulations works. Tweaking are a crude work that a lot of designers of very costly amp did not do enough.
I did read Distortion In Power Amplifiers by D. Self, it's very theorical and teach a lot, and many diy and commercials power amps designers use some of his ideas. But many guy's did need to tweak his blameless amp for a better result, that confirm me the need of tweaking for making a real good sounding amp.
My memory fail me a bit since I forgot lot of what D. Self wrote in his text Distortion In Power Amplifiers, I need to read it again. Btw, is there part of that Self's text that you are not agree with ?
Thank
Gaetan
gaetan8888 [/i][B]Chip amps are less precise at very low level details said:
Actually, the data sheets quote very good distortion figures. Your comment about low level detail and distortion is applicable to all amplifiers. Think about it.
Also, refer to the original post in this thread. He wants a simple amplifier that sounds good.
Anyway, I wonder if by "sounds good", he does not perhaps mean "does not sound bad". For "sounds good" one should look at tube amps and other distortion-generating types.
D. Self has gone to great lengths to address low level distortion, I think and WRT chip amps, well, if ever there were a comparison to be had...
I do like chip amps, though I wouldn't use them with sensitive speakers, or where their power capabilities weren't being used, (or for my main system but this is just my preference).
I do like chip amps, though I wouldn't use them with sensitive speakers, or where their power capabilities weren't being used, (or for my main system but this is just my preference).
Switch off /speed up capacitors
"....you 'tune' it with a suitable cap....."
Maybe one can look at the FFT of the signal while trying out the different caps ? The amp must be connected to a real speaker load of course.
A faster solution to pick the right value ? If so one will need to determine what frequency ( frequencies ?) and amplitude is to be used.
"....you 'tune' it with a suitable cap....."
Maybe one can look at the FFT of the signal while trying out the different caps ? The amp must be connected to a real speaker load of course.
A faster solution to pick the right value ? If so one will need to determine what frequency ( frequencies ?) and amplitude is to be used.
Shaun,
Are you being abrasive, or is this just your normal style?
Gaetan is always extremely polite.......
Ashok,
Yes, certainly, now consider how much money, time and effort is required to correlate FFT with the 'sound' to the human ear.
I'm sure it can be done, but it's a project above and beyond the simple tuning procedure I recommend, which although subjective, matches precisely what the consumer goes through when he chooses an amplifier......
Up on my soapbox: Fifty years of complex mathematical analysis and careful oscilloscope and FFT work has failed to nail the connection between these measurements (particularly THD) and the perceived, subjective 'sound' of an amplifier.
This tends to indicate that more work in this connection needs to be done, and possibly that we are measuring the wrong things. You invariably find after a potential buyer has read off the specs, he asks 'How does it sound?', and wants to have a listen. This speaks volumes, I feel, for the validity of such measurements.
I'm sure correlations exist, and Geddes has done a lot of work in this area. But until there is a defined, accurate and 100% correlatable standard, the listening test will reign supreme, as consumers listen to music and want it to make them weep.....
I use CROs, meters, distortion analysers, but find them largely inconclusive except at the engineering limits of operation. Clip peformance, and stability margin, for example.
Please don't take offense at this; I've done it this way for years, can't see a way out, would love some sort of test as it would save a LOT of time.
Cheers,
Hugh
Are you being abrasive, or is this just your normal style?
Gaetan is always extremely polite.......
Ashok,
Yes, certainly, now consider how much money, time and effort is required to correlate FFT with the 'sound' to the human ear.
I'm sure it can be done, but it's a project above and beyond the simple tuning procedure I recommend, which although subjective, matches precisely what the consumer goes through when he chooses an amplifier......
Up on my soapbox: Fifty years of complex mathematical analysis and careful oscilloscope and FFT work has failed to nail the connection between these measurements (particularly THD) and the perceived, subjective 'sound' of an amplifier.
This tends to indicate that more work in this connection needs to be done, and possibly that we are measuring the wrong things. You invariably find after a potential buyer has read off the specs, he asks 'How does it sound?', and wants to have a listen. This speaks volumes, I feel, for the validity of such measurements.
I'm sure correlations exist, and Geddes has done a lot of work in this area. But until there is a defined, accurate and 100% correlatable standard, the listening test will reign supreme, as consumers listen to music and want it to make them weep.....
I use CROs, meters, distortion analysers, but find them largely inconclusive except at the engineering limits of operation. Clip peformance, and stability margin, for example.
Please don't take offense at this; I've done it this way for years, can't see a way out, would love some sort of test as it would save a LOT of time.
Cheers,
Hugh
Gaetan
That is no worse than I expected. I believe that at low output levels crossover distortion will start to dominate, that is why the distortion figure gets worse in that region (the amplitude of the distortion products remain constant, whilst the signal strength is small). Any Class B amplifier would have the same tendency.
But what I would like to point out is that those are not bad figures at all. I believe that it will suit the purpose of the original poster. The request was not for the ultimate amplifier because, understandably, there are constraints. I was not being funny when I suggested re-reading the original post. In that context a lot of the discussion above becomes moot.
Please don't mistake me for a GainClone fanatic: I have been running a Blameless Load Invariant amplifier for the past 5 or 6 years.
That is no worse than I expected. I believe that at low output levels crossover distortion will start to dominate, that is why the distortion figure gets worse in that region (the amplitude of the distortion products remain constant, whilst the signal strength is small). Any Class B amplifier would have the same tendency.
But what I would like to point out is that those are not bad figures at all. I believe that it will suit the purpose of the original poster. The request was not for the ultimate amplifier because, understandably, there are constraints. I was not being funny when I suggested re-reading the original post. In that context a lot of the discussion above becomes moot.
Please don't mistake me for a GainClone fanatic: I have been running a Blameless Load Invariant amplifier for the past 5 or 6 years.
Shaun said:
Hello
I agree that at low output levels crossover distortion will start to dominate for most amps, but there is good amps who succeed to have much lower crossover distortion than a lot of others amps, as an exemple, 20 years ago I did have a 60 W Crimson amp kit who did have between .008 % and .01 % distortion from .1 watt to 57 watt.
But my message was more on a general basis about good sounding amps and not only about the original poster need amp.
I did myself a TDA7294 chip amp for a friend, it was sounding good but my Radford HD250 sound much better.
Yes, doing amps comparaison can sometime be very subjective but not all the time. There is amps who do a better job than others.
Bye
Gaetan
Well, this is where I begin to have a problem: these days any IC-based amplifier gets called a GainClone. National Semiconductor is probably the only chip manufacturer actively trying to meet and retain high-end standards with linear power chips...
But I think your experience also brings us to another point: had you not had the Radford amp for comparison, do you think you would have been satisfied with the sound quality of the TDA7294 amplifier? I think we sometimes fool ourselves into reaching ever higher, ever better when perhaps "good enough" would have done just as well.
I have found the LM3875/86 ilk to be somewhat fatiguing and have sought to find an improved sound. Personally, I could accept these as a bass only amp, a computer amp, and I would consider them for home theatre duty, partly due to their convenience. I would not use them in an office, bedroom, garage, or anywhere where I'd like to be able to tune them out and/or relax.
Shaun said:
Well, this is where I begin to have a problem: these days any IC-based amplifier gets called a GainClone. National Semiconductor is probably the only chip manufacturer actively trying to meet and retain high-end standards with linear power chips...
But I think your experience also brings us to another point: had you not had the Radford amp for comparison, do you think you would have been satisfied with the sound quality of the TDA7294 amplifier? I think we sometimes fool ourselves into reaching ever higher, ever better when perhaps "good enough" would have done just as well.
Hello
Well it's now a question of personal opinions, so we have different opinions.
Bye
Gaetan
My DIY amp, 50w Mirand sound good even at full blast, so something tells that you should measure amp at full load ... its also stable in 1ohm and play happily into 2ohm ... I really dont see the point if some amp distorts badly if used above half its capability, so whats the other half good fore
tinitus said:
Hello
Soundstage and other low level informations are also important and they still at very low level even wen you use your amp at full power. So if they are lost in the crossover distortions you will lose some part of the ambiance and soundstage who made a good cd sound alive.
Bye
Gaetan
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.