It's easy to claim witness from the dead, you can make up anything you want. Putting words into the mouth of someone who can not clarify is just a waste of time.
I have three cars ...... a fast road holding 2-door car that I increased the TRQ and HP on it to 700. 400 #ft at 1500 rpm. The other is a 4 door sedan (Caddy) for luxury. I increased the road holding/handling and power on that one ---- it drives like a dream on long trips -- a nice GT. Then there is the little stock 4-banger.... took the rear seats out for more of a van type use (PT Cruiser). That little car is so easy to drive, it almost drives its self. I just cant find a single car that does it all..... same for audio gear. But I keep looking/trying to find that one amp that does it all.
THx-RNMarsh
Do you mean.... as in .... clean amp performance first then gnfb applied sounds best?
-RNM
IME it's the system working correctly that "does it all" - one component looked at in isolation won't cut it, for getting the job done. And if it does come together, completely together, then you can put on anything, absolutely anything, and it works ... that's the beauty of fully optimising, no nasty surprises, ever ...
The only thing that I can control is the open loop linearity, so I try to make my amps and preamps as linear as possible. When I can, I use as little global negative feedback as possible, and sometimes, like the CTC Blowtorch, none. However, I have never tried to make a power amp with no global feedback. Charles Hansen does, however, and his measurements are good, but not special, because he does not use any global feedback.
I use plenty of negative feedback, but I don't go overboard to generate a lot of excess gain at very low frequencies. This is easily done with active loads on the input stage, but I use input stage load resistors, almost exclusively. What I try to do is to extend the gain bandwidth as much as possible, then get the open loop bandwidth as high as possible, and of course, the open loop linearity as smooth as possible. Personally, I think that negative feedback is not necessarily a problem, IF I could always have an open loop bandwidth above 20KHz. The Otala amp does it, but my amps are have about 4KHz or so open loop bandwidth, far more than many power amps, but nothing to brag about.
IF the circuitry was perfectly linear, and the open loop bandwidth extended to way above 20KHz, then adding global negative feedback should be OK, but then you would not need to use it in the first place. I use a lot of negative feedback in order to get good specs, that can be stated on spec sheets, and not because it makes the amps sound better. My competition is with the open loop guys like Charles Hansen of Ayre or Lamm, not with op amp designers.
Reducing FM distortion and its relatives are an important key to better sound quality. Look for more input soon!
I use plenty of negative feedback, but I don't go overboard to generate a lot of excess gain at very low frequencies. This is easily done with active loads on the input stage, but I use input stage load resistors, almost exclusively. What I try to do is to extend the gain bandwidth as much as possible, then get the open loop bandwidth as high as possible, and of course, the open loop linearity as smooth as possible. Personally, I think that negative feedback is not necessarily a problem, IF I could always have an open loop bandwidth above 20KHz. The Otala amp does it, but my amps are have about 4KHz or so open loop bandwidth, far more than many power amps, but nothing to brag about.
IF the circuitry was perfectly linear, and the open loop bandwidth extended to way above 20KHz, then adding global negative feedback should be OK, but then you would not need to use it in the first place. I use a lot of negative feedback in order to get good specs, that can be stated on spec sheets, and not because it makes the amps sound better. My competition is with the open loop guys like Charles Hansen of Ayre or Lamm, not with op amp designers.
Reducing FM distortion and its relatives are an important key to better sound quality. Look for more input soon!
Yeah. The worst parameter is good look. Has nothing to do with sound but matters. I have a little trick. Inside, anything best for the sound (point-to-point wiring, shortest track, etc. Outside, just coat it with fancy material (even fabric/cloth to hide ugly speaker).
And I'm slowly, hesitantly, moving to multi amping system with discrete crossover, mixed with first order passive.
BTW, can anybody help me. I need short theoretical explanation regarding amplifier I'm building today. The amp puts out DC (I used 10Vdc setting in the multimeter so I know it is more than 10V, probably the power supply 30V). First I measured the output AFTER the output capacitor. This is what surprised me. I thought the cap can block DC but it didn't!!? (The caps are 25V, new old stock, different brands). Another thing that amazed me (I feel like dreaming), no sound from the 4" speaker (no plop either), only vibrating. I have checked the meter, it's normal.
I'm not expecting off-topic discussion. Only short answer to show where my assumptions have been wrong. I have ditched the amp and is building another one.
I would agree about the system working best possible together. But some things just need to be better than they are.
However, as in each car I have... each is optimized for a particular goal as a complete system. For an audio system to work at its optimum for my goal, it is more of the 2 seater sport car..... fast, quick, agile, stops and starts quickly, high top speed, plenty of reserve power etc. Others may find they want to optimize for the Caddy sound... smooth, lush, but fast enough etc. Or the optimized PT for fuel economy and easy to drive, light weight etc. Optimized has different results depending on goal.
THx-RNMarsh
However, as in each car I have... each is optimized for a particular goal as a complete system. For an audio system to work at its optimum for my goal, it is more of the 2 seater sport car..... fast, quick, agile, stops and starts quickly, high top speed, plenty of reserve power etc. Others may find they want to optimize for the Caddy sound... smooth, lush, but fast enough etc. Or the optimized PT for fuel economy and easy to drive, light weight etc. Optimized has different results depending on goal.
THx-RNMarsh
Last edited:
What do you think of the amp designs here using error correction rather than gnfb. DADOD here has done some really interesting designs which are not dependant on gnfb yet have very low distortion.
THx-RNMarsh
Last edited:
By that analogy I would go for the 2 seater sport car as the best initial solution; the Caddy sound is a "fake", it will be tripped up by the "wrong" recordings; and the light weight variety implies a lack of capacity to deal with strong transients. So, I would take the 2 seater, and make sure it delivered "lush", when that is what's on the recording, and then improve the controls interface so the "easy to drive" aspect fell into place ...
Last edited:
Ah, yes. Like avoiding class-A monster amp. The Extrema? Yikes!
About the amp I'm building, bad caps? Probably, tho hardly because I have plenty of them and they are fine in other amplifiers. But thanks anyway, at least you didn't say that caps do not block DC. May be I'm just tired building since last night (yesterday).
I think GNFB is simply a tool, much like a say hammer. With a hammer, you can idly crack nuts, nake a house or kill someone; like any tool, it depends on the hand holding it.
GNFB is really no different. You can use it indiscriminately to iron out the bumps, or use it to simply improve an already good circuit. How much of it do you need will thus depend on your design objectives and how much is required to make the amp completely stable under any condition within reason. GNFB by itself is neither good nor bad, it depends on how you use it.
For example, looking at H/K line of integrated amps from 20 years ago, you will find that GNFB was used at a value from 26 to 12 dB, and listening to them will tell you that they do not sound quite the same - similar, but not the same (beside the obvious difference in the output voltage).
I don't much like generalizing, but if I were to generalize, I'd guess that lower GNFB factor amps have a better chance of sounding good, assuming their open loop bandwidth is reasonably wide, meaning from at least say 40 kHz upwards. Better chance, no certainty, no guarantees.
In my view, the more output devices one has to share the workload, the less GNFB one can get away with. Obviously, this hinges on what level of distortion one sees as the acceptable minimum.
Let's not forget that the less amplification we require from our circuits in order to make them very wide bandwidth is offset by the fact that the less amplification they have to do will by itself reduce all sorts of distortions, although certainly not all.
Otala & Lohstroh pushed their amp out to 100 kHz open loop. And remember, in 1973, their choice of components was ridiculously small compared to them today.
GNFB is really no different. You can use it indiscriminately to iron out the bumps, or use it to simply improve an already good circuit. How much of it do you need will thus depend on your design objectives and how much is required to make the amp completely stable under any condition within reason. GNFB by itself is neither good nor bad, it depends on how you use it.
For example, looking at H/K line of integrated amps from 20 years ago, you will find that GNFB was used at a value from 26 to 12 dB, and listening to them will tell you that they do not sound quite the same - similar, but not the same (beside the obvious difference in the output voltage).
I don't much like generalizing, but if I were to generalize, I'd guess that lower GNFB factor amps have a better chance of sounding good, assuming their open loop bandwidth is reasonably wide, meaning from at least say 40 kHz upwards. Better chance, no certainty, no guarantees.
In my view, the more output devices one has to share the workload, the less GNFB one can get away with. Obviously, this hinges on what level of distortion one sees as the acceptable minimum.
Let's not forget that the less amplification we require from our circuits in order to make them very wide bandwidth is offset by the fact that the less amplification they have to do will by itself reduce all sorts of distortions, although certainly not all.
Otala & Lohstroh pushed their amp out to 100 kHz open loop. And remember, in 1973, their choice of components was ridiculously small compared to them today.
On the subject of resistors. Some years ago, I had the opportunity to change my standard Beyschlag resistors for Dale units. I did not measure anything, I simply sat down and tried to find any sonic differences. Changing input stage degeneration resistors did nothing for the sound I could find.
But changing the GNFB voltage divider resistors did produce a difference. Small, true, but it was just audible, and for the better. I underline the values were kept the same, and all resistors were checked by measurement for their values, just in case, and both were well withn their 1% tolerance.
But changing the GNFB voltage divider resistors did produce a difference. Small, true, but it was just audible, and for the better. I underline the values were kept the same, and all resistors were checked by measurement for their values, just in case, and both were well withn their 1% tolerance.
Jay, you might be interested in capacitor forming.
Basically, it's a tedious process of taking a new, never before used capacitor, and charging it for say 5 minutes with a low voltage, say 5V. Then discharge it with a power resistor of adeqaute handling. Repeat the sequence again, but using say 15V this time.
The idea is to polarize it prior to soldering it in. I cannot guarantee it will sound any different, but experience has taught me that caps so formed will usally last longer than virgin caps simply dropped into a circuit.
- Status
- Not open for further replies.