No coincidence, you talked to each other! The descriptions were highly non-specific and vague.
Here's a possibly skill testing question: can prejudices, especially highly vague ones be spread by people talking to each other?
Fourier actually won't usually help tell you what you can hear and not hear. It is a step along the way, though because it helps clarify what you are trying to hear.
For that you need to study psychoacoustics, starting with Fletcher and Munson and progressing through Zwicker and Fastl.
You can short cut all this by simply doing some time-synched, level-matched, double blind listening tests.
Unfortunately, I still haven't scaled the discussion down to things that most audiophiles and even any number of self-proclaimed audio experts can grasp.
Note: There is an existing sharp cut-off LPF in the CD player....... but the high filter ripple above cut-off is letting thru HF junk.
View attachment output filter.pdf
THx-RNMarsh
View attachment output filter.pdf
THx-RNMarsh
Last edited:
Something like this ? depends on DAC design what filtering is needed/used. The filter noise levels and frequencies shown go higher than I thought on some CD players (and DVD).
Usually you only see the traditional affects of a filter on a signal. And digital filter techniques etc. Nothing about 'noise energy' within the ripple and how a PA responds to it re distortion increase in audio band-pass. Clearly low ripple is needed but not always provided........ I recently read of dig and analog filters together are sometimes use. I am just looking at the level of energy above the filter and its affect on various amplifier topologies.
THx-RNMarsh
Last edited:
perfect. May-be we will learn something usefull about the two sides of the problem (souce and effect).
I have never read a serious study about the exact effects of parasitic HF signals (noise or high order harmonics) on amplifiers of different closed loop bandwidths.
I'm specialy curious of the CFA/VFA differences in that matter. Removing any input filter in thelm Is this part of your project ?
yes.
Neither (CFA/VFA) may be especially better than the other if neither have increased nonlinearity/distortion in those HF regions.
It may be just coincidence that designs which have lower distortion above audio also sound better. Not so much because the amp is doing anything directly to the input signal/music but what happens when you have also HF signals to contend with that are moderate levels (not at the noise floor). Would apply to mistracking phono cartrdiges also.
It might help explain why so many people hated the sound of CD early on and some still do. And might explain why certain amp topologies and design decisions Sound better. Certainly is a contributing factor, IMO.
THx-RNMarsh
Neither (CFA/VFA) may be especially better than the other if neither have increased nonlinearity/distortion in those HF regions.
It may be just coincidence that designs which have lower distortion above audio also sound better. Not so much because the amp is doing anything directly to the input signal/music but what happens when you have also HF signals to contend with that are moderate levels (not at the noise floor). Would apply to mistracking phono cartrdiges also.
It might help explain why so many people hated the sound of CD early on and some still do. And might explain why certain amp topologies and design decisions Sound better. Certainly is a contributing factor, IMO.
THx-RNMarsh
Last edited:
I for one agree with you, Scott- It can't be a coincidence that for the last 30+ years I have shown such preference for wide bandwith ampfiers in comparison to other amps which needed a lot of GNFB to even work. Of course, there's more to it than just speed, but speed seems to be the key ingredient.
An easy statement to question. You can't add gain or speed to an amplifier with global or even local NFB. The basic capability has to be there before the NFB is added.
NFB always involves throwing away gain. That's why the N stands for Negative.
OTOH you can flatten frequency response with NFB, and by flattening FR you can move HF - 3 dB points higher. But under the covers, you gave away gain to achieve this. Furthermore, obtaining adequate stability can involve even giving away HF gain and reducing bandwidth.
I've never seen NFB make a non-operational amp start working. The amp may not be very practical due do its high open loop gain, but by attenuating the input voltage, it has been possible to have some kind of a working amp.
Most cases where an amp running open loop has really poor frequency response, the poor frequency response is the result of a lag network and other circuit features that were added to make the amp appropriate for applying NFB. Prior to the lag network & etc., the amp was probably a fair working amp.
Any statement is open to questioning. In the end, with small or large GNFB factor, ultimately we add about the same reacion, the question is will we pump the amp up and then throw lots of GNFB at it, or will we add local NFB loops and make less GNFB necessary. Add it up and you're at about the same od similar.
No doubt the GNFB will influence the amps's stability, however I'd like to point out that even soma mass manufacturers have long ago learnt how to make their amps stable into even evil loads with 12-15 dB of GNFB (Harman/Kardon, for example, I own three of them, one has 15 dB of GNF, two have 12 dB of GNFB). I have made them work into pretty reactive loads (e.g. 3 Ohms in parallel with 2 uF) and they were completely stable, although of course their output THD was greater than into nice and esy loads. I prefer a degraded THD figure to clipping and/or instability, not to even mention oscillation. In reality, I never go that far high up the power scale as these amps wirk into an unusualy wasy load (ma speakers are nominally 8 Ohms, minimum 6.5 Ohms, worst case phase shift -25 digrees).
As Andrew pointed out here some time ago, there's really no great problem achieving a full power open loop response of 40 kHz, and even as much as 70 kHz if you try hard. Otala and Lohstroh achieved 100 kHz in 1973 with those components. Problems start when you have to market the amp, when like it or not it will be copared to similar amps starting a THD race, where users will take it for granted that a lower THD factor will make for a better sounding amp.
No doubt the GNFB will influence the amps's stability, however I'd like to point out that even soma mass manufacturers have long ago learnt how to make their amps stable into even evil loads with 12-15 dB of GNFB (Harman/Kardon, for example, I own three of them, one has 15 dB of GNF, two have 12 dB of GNFB). I have made them work into pretty reactive loads (e.g. 3 Ohms in parallel with 2 uF) and they were completely stable, although of course their output THD was greater than into nice and esy loads. I prefer a degraded THD figure to clipping and/or instability, not to even mention oscillation. In reality, I never go that far high up the power scale as these amps wirk into an unusualy wasy load (ma speakers are nominally 8 Ohms, minimum 6.5 Ohms, worst case phase shift -25 digrees).
As Andrew pointed out here some time ago, there's really no great problem achieving a full power open loop response of 40 kHz, and even as much as 70 kHz if you try hard. Otala and Lohstroh achieved 100 kHz in 1973 with those components. Problems start when you have to market the amp, when like it or not it will be copared to similar amps starting a THD race, where users will take it for granted that a lower THD factor will make for a better sounding amp.
Agreed. In the end low amplifier distortion is the consequence of the basic gain and linearity of the devices used to make it, knowing full well you can trade gain for linearity and vice-versa if something is there to trade.
I think I discovered this when I first experimented with unbypassed cathode resistors. ;-)
I was a bit faster. That Otala/Lohstroh amp from 1973 was a clear turing point in my view evolution. It seemed logical to me to first leanrize each and every single stage and then apply enough GNFB to simply clean op what's left over. Since then, I never once needed more than 26 dB of GNFB, and that on occasion only, I am usually happy with about 20 dB of GNFB. However, I do not have the same reasoning of the mass proudcing manufacturers and their numerous bean counters - I test for what I feel are the edges of performance I expect from it and use as many power device pairs as required, and damn the cost. I have the luxury of working for my own self, and can afford power devices rated at 200, 230 and 250 Watts, rather than industry standard decies such as Toshiba'a 2SC5200/2 SA1943 simply because they are cheaper.
- Status
- Not open for further replies.