So, DVV is the guy taking burgeois vacations here?
While Nige is the guy who takes pedestrian vacations?
Yes, if you prefer instead of freezing to feed mosquitoes.
You can live there for free, no need for $200. Berries, mushrooms, fish, meat, honey, cedar nuts, you name it. But be advised that The Master of Taiga (siberian forest) is Bear.
Yes, if you prefer instead of freezing to feed mosquitoes.
You can live there for free, no need for $200. Berries, mushrooms, fish, meat, honey, cedar nuts, you name it. But be advised that The Master of Taiga (siberian forest) is Bear.
Hence the saying "Russian Bear"?
A few days ago, I caught a program on Nat Geo (which I like watching on occasion) about a guy living in the wild in USA for research purposes, whose best friend and inseparable partner is a grizzly bear. Really looks surreal, you see a man walking and a damn big bear trotting next to him like a puppy. They live together, they fish together, they even hunt together, and share equally each and every time.
The man says bears are great friends, so long as you always remember that they are bears and you think like a bear, rather then attributing them human characteristics.
That's a tip for you, Nige, if you should get to Siberia.
If I had lived under communism I am sure I would have been sent before now .
Bears , who needs them ? They fish for salmon so can't be all bad . It was recently discovered that the Panda is a bear . I think we might have guessed . DNA does tell us the most unlikely things !
It is said running at them is the only defense . It is so rare anything does it might just catch them off guard ( reportedly it did work ) . They are said to run faster than Usain Bolt . They can smell you 50 km away .
Bears , who needs them ? They fish for salmon so can't be all bad . It was recently discovered that the Panda is a bear . I think we might have guessed . DNA does tell us the most unlikely things !
It is said running at them is the only defense . It is so rare anything does it might just catch them off guard ( reportedly it did work ) . They are said to run faster than Usain Bolt . They can smell you 50 km away .
Unfortunately that guy met an early death:
"Treadwell was discovered dead and partially eaten by one of his beloved grizzlies at his campsite in Katmai National Park in the fall of 2003. His girlfriend, Amie Huguenard, was also killed."
Zoom lens is your friend. Bear (or moose, elk, etc) isn't.
"Treadwell was discovered dead and partially eaten by one of his beloved grizzlies at his campsite in Katmai National Park in the fall of 2003. His girlfriend, Amie Huguenard, was also killed."
Zoom lens is your friend. Bear (or moose, elk, etc) isn't.
Any of you guys coming to http://www.diyaudio.com/forums/club...-mountain-audio-fest-2012-a.html#post3196195?
We have a combined stand Elektor USA, AudioXpress and Linear Audio.
Bob Cordell will show his VinylTrak phono/preamp.
jan
We have a combined stand Elektor USA, AudioXpress and Linear Audio.
Bob Cordell will show his VinylTrak phono/preamp.
jan
What is the general opinion on bandwidth shaping?
I refer to the technique in which a bandpass filter is placed in front of the power amp, typically using 12 dB/oct. filters, to cut out subsonic signals (e.g below 15 Hz) and ultzrasonic signals (e.g. 50 kHz).
Assuming you set the low pass filter BELOW your open loop full power bandwidth limit, then theoretically you should have a TIM free device, as the amp cannot be outpaced by any reasonable input signal which could appear in real life operation; I mean, 20V at 200 kHz just ain't natural.
The trade-off will be the natural phase shifts introduced by any filter, plus of course the added distortion and noise from added electronics.
It seems to me both arguments have merit, perhaps the added distortion can be made to be truly insignificant, but the phase shifts will not go away. On the other hand, cutting out the effectively useless parts of the bandwidth (not too many loudspeakers out there which will faithfully do even 30 Hz, let alone 15 Hz, what with the current craze regarding small enclosures) does offload the amp, at least in theory.
So, what do you think?
I refer to the technique in which a bandpass filter is placed in front of the power amp, typically using 12 dB/oct. filters, to cut out subsonic signals (e.g below 15 Hz) and ultzrasonic signals (e.g. 50 kHz).
Assuming you set the low pass filter BELOW your open loop full power bandwidth limit, then theoretically you should have a TIM free device, as the amp cannot be outpaced by any reasonable input signal which could appear in real life operation; I mean, 20V at 200 kHz just ain't natural.
The trade-off will be the natural phase shifts introduced by any filter, plus of course the added distortion and noise from added electronics.
It seems to me both arguments have merit, perhaps the added distortion can be made to be truly insignificant, but the phase shifts will not go away. On the other hand, cutting out the effectively useless parts of the bandwidth (not too many loudspeakers out there which will faithfully do even 30 Hz, let alone 15 Hz, what with the current craze regarding small enclosures) does offload the amp, at least in theory.
So, what do you think?
You make two assumptions:
1. The filter does not ring. You can have a 12dB/octave filter with a huge peak near the corner frequency. This could increase TIM problems. You would not do this, but worth mentioning. A filter needs to be specified by more than its final slope and corner frequency; for 2nd-order there is also Q. 1/sqrt(2) for Butterworth (maximally flat), 1/sqrt(3) for Bessel (most linear phase). People who worry about waveform appearance would go for Bessel, but I believe I am correct in saying that listening tests suggest Butterworth sounds better. This may be because our ears are more sensitive to amplitude than phase.
2. You can go well above the open loop full power bandwidth. It is the closed loop full power bandwidth which matters. You may prefer the former, but the latter is what the maths tells us.
All these out-of-band frequencies will be removed somewhere in the chain. The question is where. It might sometimes be better to remove them in a nice linear filter than expect a transducer to cope with them.
1. The filter does not ring. You can have a 12dB/octave filter with a huge peak near the corner frequency. This could increase TIM problems. You would not do this, but worth mentioning. A filter needs to be specified by more than its final slope and corner frequency; for 2nd-order there is also Q. 1/sqrt(2) for Butterworth (maximally flat), 1/sqrt(3) for Bessel (most linear phase). People who worry about waveform appearance would go for Bessel, but I believe I am correct in saying that listening tests suggest Butterworth sounds better. This may be because our ears are more sensitive to amplitude than phase.
2. You can go well above the open loop full power bandwidth. It is the closed loop full power bandwidth which matters. You may prefer the former, but the latter is what the maths tells us.
All these out-of-band frequencies will be removed somewhere in the chain. The question is where. It might sometimes be better to remove them in a nice linear filter than expect a transducer to cope with them.
DVV . I find as I am sure you have that playing with the input capacitors gives you plenty of options ( and LTP cap ) . My instinct tells me that 3 Hz - 3db and 200 kHz - 3dB should be aimed for . It is then a simple task to add filtering if required . I think you yourself said it should be inside the pre-amp if anywhere ? I think you will find what I have said contradicts what you might perceive my preferences to be . I have always preferred wide bandwidth amps ( makes piano sound real , don't know why ) . Sometimes the typically British way suits very well . That is perhaps 20 Hz to 20 kHz - 1 dB ref 1kHz . I have a very early Quad 33 / 303 which is my everyday amp . Mostly I listen to MP3 style stuff via TV . It never is less than pleasing . I have learned to like it very much . At first I used it because it was the only bit of hi fi I didn't sell when feeling the pinch . Now it is my honest friend . It lacks detail is my only criticism ( much more detailed than any would guess ) . I have adjusted it very carefully so as to use it's capabilities well . The 33 is not the best , even so mostly that is about matching and not disaster of performance by bad design ( my original assumption ) .
You might have detectors that kick in filters if excessive ULF or HF detected . These could be reset and have a warning light . You might also have PA mode .
You might have detectors that kick in filters if excessive ULF or HF detected . These could be reset and have a warning light . You might also have PA mode .
I call it passband adjusting.
I firmly believe that the passband of a power amplifier should be determined by a pair of passive filters at the input to the power amplifier.
One decides either by listening, or by measurement, or maybe a combination of both, what the required passband should be to let the audio signal pass through without corruption. I set my filters quite a bit wider than F(-1dB) from 20Hz to 20kHz, but that is my choice. Each builder/designer/manufacturer can choose what they want as the passband.
I do not use active filtering inside the amplifier's global loop to determine the passband of the power amplifier.
Andrew I would say the same . It is a free lunch . I tend to use non-polar large electrolytic for the LTP feedback cap . Sometimes even 1000 uF 16 V ( Nikkai ) . I can see some throwing their hands into the air . All I would say is try it . I dare say if using J-FET's no capacitor required on LTP -ve ? Dvv as you use a servo I guess you have options .
The point I am making is if the LTP -ve input is 1000 uF and 1K we can say almost DC ( 0.16Hz - 3dB ) . If we say 10 uF polyester 100V and 22K input we have about 1 Hz - 3dB combined . Correct my maths if required ( often is ) . 1 uF then is 10 Hz to a good approximation . We can then play with 1000 uF to see if it helps to reduce it ( ears ) . My guess is it will be retained . Non polar's work very well in this application and surpass many so called Audiophile types . They work especially well if kept below 0.4 V ( pk to pk ) of signal . I would say if 0.4V is reached after 10W rms all is well . This means fine for all applications I would say . This is an unexpected bonus of high gain amps , the feedback cap voltage is kept low . Electrolytic's do not work well when biased contrary to opinion . They need it is other applications , not this one . As DC offset is often only about 25 mV I haven't included it . As far as I know the electrolytics will not notice DC of less than 0.4V . The bad effect at 0.4V is a diode like characteristic .
I try not to use feedback compensation filtering .
The point I am making is if the LTP -ve input is 1000 uF and 1K we can say almost DC ( 0.16Hz - 3dB ) . If we say 10 uF polyester 100V and 22K input we have about 1 Hz - 3dB combined . Correct my maths if required ( often is ) . 1 uF then is 10 Hz to a good approximation . We can then play with 1000 uF to see if it helps to reduce it ( ears ) . My guess is it will be retained . Non polar's work very well in this application and surpass many so called Audiophile types . They work especially well if kept below 0.4 V ( pk to pk ) of signal . I would say if 0.4V is reached after 10W rms all is well . This means fine for all applications I would say . This is an unexpected bonus of high gain amps , the feedback cap voltage is kept low . Electrolytic's do not work well when biased contrary to opinion . They need it is other applications , not this one . As DC offset is often only about 25 mV I haven't included it . As far as I know the electrolytics will not notice DC of less than 0.4V . The bad effect at 0.4V is a diode like characteristic .
I try not to use feedback compensation filtering .
DF96, I make absolutely no assumptions whatsoever, I simply asked what was the general feeling here regarding active bandpass filters, no more. I mentiond 12 dB/oct as an example - I see less as not so desirable, and more as dicey exactly due to phase and response funnies.
Also, I see a second order filter as giving me more lattitude than a simple first order filter, it lest more of the bandwidth pass through before it takes effect and gets it all down to the same level as a first order filter would.
As an example, I tend to agree with reVox and Sony on this matter - they first get the amp to have a wild bandwidth, typically 350 kHz and over, and then they throw in a first order filter at the input, typically cutting by -3 dB at 200 kHz (reVox) or 150 kHz (Sony).
What prompted my initial question was a schematic from a UK company, I forgot which one. Not very famous. What they did was to use op amps to get the 12 dB/oct slopes at (if memory serves) 15 Hz and 70 kHz.
The good Lord knows one has his choice of filter types, from Bessel and Butterworth all the way to Smith-Papoulos filters, with slopes of 375 dB/oct, as used in some MacIntosh tuners. Of course, in tuners it's a different matter, we don't need that in power amps, so a nice 2nd order should be just fine.
In practice, neither is a problem, really. The low pass could be a simple series resistor followed by a capacitor to the ground, pick your cutoff frequency as you like it. UK manufacturers generally opt for 100 kHz or less, though there are exceptions. The high pass is similar, if it's an AC design, throw in a series capacitor before the resistor setting the input impedance, and/or use a smaller value cap in the NFB path.
Ultimately, I expect (but have not actually tried) one could even adjust a DC Servo circuit to act as a filter, though that does raise the question of using it in the first place, it's point being NOT to use capacitors.
Let me rephrase the question - do you think subsonics and ultra high frequency signals (say from 100 kHz up) should be attenuated? If so, do you feel the first order filters we can install right at the amp's input do the job just fine, or could they use being 2nd order filters?
Just to remind you, W. Marshall Leach did use a 2nd order low pass filter in his early 70ies amp project published in USA's Audio magazine, and in the subsequent Double Barrel 200W project later on.
- Status
- Not open for further replies.