I tried that . Problem is 1 A pull down to qualify . He obviously is doing the trick by a bias stand off . I doubt the patent would be granted for the pull up / down CCS as I remember that form the 1970's when 741 . Looking at the link below I was fooled , I just asked the right question in Google this time . Hence Pat Pending ( I suspect it will not allowed ) . LT show a pull down CCS in application notes and I am sure many others do , they use a J FET . Ironically a pull down in not needed in two stage mic or phono amps as the first stage current should fall inside the first op amp class A range ( 1 mA I would guess ) . My phono stage only would need 12 uA output at stage one ( 700 uV x 62 )/3750 R . I still use the class A trick and feel it is better ! LT show it for an RIAA two stage preamp . I wonder if that is a crowd pleaser or like me they though it did sound better ? I can not think the op amp has less than 100 uA standing current in it's output stage ? In my second stage I choose not to do it as I feel it has more punch . How my pre amp works is to almost get the output to MM level in stage one . The final stage has a gain of 17 at 1 kHz .
Pull down I think is a 741 legacy . The pull down slightly out performed pull up . If a PNP NPN output stage it is supposed to knockout the PNP is the better choice . A 4k7 resistor usually is a good approximation to the needs . Listen and see what you think . This resistor is a great way to teach DC feedback and how an amplifier reacts .
Current feedback whilst I remember . My friend John says current feedback assisted early op amps that had poorly matched LTP . Refereeing everything to 0 V helps . He says the obsession with current feedback continues long after the problems have been solved . I could never find a big deal reason to use it . Always measured the same more or less . It causes additional problems , phase and drive being two . In an RIAA stage it just about makes sense . 50 R input resistance and double inverting stages . The second stage DC block capacitor useful doing extra work if so , 22 uF in may case and 1K8 . Useless for MM .
This was shifting the crossover distortion point in Cambridge A840 . There is more info since I last looked when the A840 came out . Here is an extract from the link .
" Now, from the Cambridge Audio whitepaper, we learn that the new Cambridge Audio's patent-pending-proprietary-XD™ technology adds a slight twist: the constant-current source’s current is not constant; instead, the current varies with the output swing, drawing the most current when the output swings down close to the negative rail voltage, and drawing the least current when the output swings close to the power supply rail voltage. The formula that the whitepaper specifies is "
Problem is I was put off by the journalist who described a rather more elegant solution that in fact doesn't exist in the A 840 . I like it that a tube site has to tell use the how and the why .
Amplifier Class-XD™
Last edited:
Back to post #1, I have been doing some crossover distortion testing. I found little distortion caused to sine wave type instruments, like REAL acoustic strings, wind instruments, etc. But put modern synthesizer music to the test with square and triangle waves and their near harmoncs. Yuk, it gets ferocious.
Our systems are likely not sounding much like the source. Take Phillip Glass's 1000 Airplanes cd. It sounds very different from system to system. It is all synthesized and FULL of non sinusoidal waves plus harmonics. It even worries me this one might be rough on the tweeters, but so far it hasn't fried my Linaeum ET-19's, a sort of easy to fry tweeter.
Our systems are likely not sounding much like the source. Take Phillip Glass's 1000 Airplanes cd. It sounds very different from system to system. It is all synthesized and FULL of non sinusoidal waves plus harmonics. It even worries me this one might be rough on the tweeters, but so far it hasn't fried my Linaeum ET-19's, a sort of easy to fry tweeter.
Back to post #1, I have been doing some crossover distortion testing. I found little distortion caused to sine wave type instruments, like REAL acoustic strings, wind instruments, etc. But put modern synthesizer music to the test with square and triangle waves and their near harmonics. Yuk, it gets ferocious.
Our systems are likely not sounding much like the source. Take Phillip Glass's 1000 Airplanes cd. It sounds very different from system to system. It is all synthesized and FULL of non sinusoidal waves plus harmonics. It even worries me this one might be rough on the tweeters, but so far it hasn't fried my Linaeum ET-19's, a sort of easy to fry tweeter.
Our systems are likely not sounding much like the source. Take Phillip Glass's 1000 Airplanes cd. It sounds very different from system to system. It is all synthesized and FULL of non sinusoidal waves plus harmonics. It even worries me this one might be rough on the tweeters, but so far it hasn't fried my Linaeum ET-19's, a sort of easy to fry tweeter.
Was this by measurement ? Did you find a fix ?
The CCS class A trick is sometimes to try ( Cambridge A 840 ) . When I get a spare moment I am going to try to deliberately shift the crossover point to 5V rms kick in . This might mean capacitor coupling , a small price to pay for better sound . A capacitor will be grossly better than the better valve transformers costing a fortune .
The CCS class A trick is sometimes to try ( Cambridge A 840 ) . When I get a spare moment I am going to try to deliberately shift the crossover point to 5V rms kick in . This might mean capacitor coupling , a small price to pay for better sound . A capacitor will be grossly better than the better valve transformers costing a fortune .
All signals are a summation of sine waves. ( just physics) All speakers are a transducer with characteristic transfer functions. ( again, just physics) It is totally invalid to send triangle or square waves to a traducer and expect they all will do the integration the same. Not a chance. If you don't do the integration first ( digital filter or analog filter) and rely on the transducer, obviously the resulting signal will differ. You are likely to get a large sum of HD and IMD that varies with system. To produce even "modern" music without proper output integration is a gross miss-use of the technology.
If you look a little at the distortion characteristics of your crossover components, you will find inductor distortion rises with frequency. Caps, well all kinds of things get in there. Lesson: Don't ask it to something it was not designed to do!
Fourier died as the first steam engines took to the rails . This is remarkable we can use his formula today with confidence .
Sq wave F + 1/3 F3 + 1/5 F5 + 1/7 F7 .......... infinity
Tri wave F + 1/9 F3 + 1/25 F5 + 1/49 F7 .......... infinity
From my tests triangle waves are the best of all . Most amps love them . Flatters my valve design .
One thing I noticed . 10 % THD with exponentially decaying harmonics looks almost like zero distortion on first glance . 5 % THD with 3 dr 5 th and 7 th looks dreadful . How it looks and how it sounds are related .
Sq wave F + 1/3 F3 + 1/5 F5 + 1/7 F7 .......... infinity
Tri wave F + 1/9 F3 + 1/25 F5 + 1/49 F7 .......... infinity
From my tests triangle waves are the best of all . Most amps love them . Flatters my valve design .
One thing I noticed . 10 % THD with exponentially decaying harmonics looks almost like zero distortion on first glance . 5 % THD with 3 dr 5 th and 7 th looks dreadful . How it looks and how it sounds are related .
This is an extract form this below .
More or less correct. The FFT makes channels, but if the signal is not
exactly in the center frequency of the channel, there is some smearing to
the neigbours or something like that. As I know, there is also a DC channel
and subharmonics might show up as a DC component, depending on the phase of
the subharmonic.
Remember that FFTs only work as expected if you have a periodic signal. So
either include the subharmonic so that you get a periodic signal or live
with the misinterpretation.
Subharmonics are frequencies not an integer multiples of the root frequency
like harmonics, but integer divisions of the root frequency. But your
interpretation below is right. Subharmonics are rarely found and are seldom
hearable (except in the Trautonium which was build to generate
subharmonics). You sometimes can find them in coupled resonators, the
german "Kombinationstoene" (combination tones? difference tones?) are
subharmonics if I remember correctly (find your Helmholtz "On the
sensations of..."). But they are rarely audible (in some pipe organs
coupling inside the registers air reservoir). Probably in bowing two
strings on the same instrument at once and in multistring instruments like
the piano.
Fourier knew of these problems !!!!!!!!
[music-dsp] sub harmonics and mains hum
More or less correct. The FFT makes channels, but if the signal is not
exactly in the center frequency of the channel, there is some smearing to
the neigbours or something like that. As I know, there is also a DC channel
and subharmonics might show up as a DC component, depending on the phase of
the subharmonic.
Remember that FFTs only work as expected if you have a periodic signal. So
either include the subharmonic so that you get a periodic signal or live
with the misinterpretation.
Subharmonics are frequencies not an integer multiples of the root frequency
like harmonics, but integer divisions of the root frequency. But your
interpretation below is right. Subharmonics are rarely found and are seldom
hearable (except in the Trautonium which was build to generate
subharmonics). You sometimes can find them in coupled resonators, the
german "Kombinationstoene" (combination tones? difference tones?) are
subharmonics if I remember correctly (find your Helmholtz "On the
sensations of..."). But they are rarely audible (in some pipe organs
coupling inside the registers air reservoir). Probably in bowing two
strings on the same instrument at once and in multistring instruments like
the piano.
Fourier knew of these problems !!!!!!!!
[music-dsp] sub harmonics and mains hum
Dear Nigel!
Your inter-punctuation makes me wonder if You mean one tenth of distortion (0,1%)? The convention is to put the full stop as the next character after the last character in the word ending a sentence Like this. And not like this .
I hope You are not offended by my comment as I truly want to understand what your message is.
/
Your inter-punctuation makes me wonder if You mean one tenth of distortion (0,1%)? The convention is to put the full stop as the next character after the last character in the word ending a sentence Like this. And not like this .
I hope You are not offended by my comment as I truly want to understand what your message is.
/
This flavor of snake oil goes all the way back to Edison with specially trained singers that sounded like his cylinders. Reproduced sound is not live sound. Don't expect it to be. You are not in the same environment.
Passive crossovers are an old technology and on the way out for serious systems. As far as transducers, position feedback drivers are starting to show. That means the electronic crossover can split the frequency bands, then compare transducer cone position to the drive signal and make super sonic corrections to make the cone nearly exactly follow the signal.
You have probably seen it ?
Nearly exact ? How about approximately exact ? Well within a very narrow margin of travel position error. Or to even correct for electrical to mechanical conversion problems. It can measure cone position, and even the sound produced, then make corrections so fast we can't sense it. But we will sense the sound quality. I don't see how it's going to get any better after that. Gonna be very expensive at first, then expensive.
You have probably seen it ?
Nearly exact ? How about approximately exact ? Well within a very narrow margin of travel position error. Or to even correct for electrical to mechanical conversion problems. It can measure cone position, and even the sound produced, then make corrections so fast we can't sense it. But we will sense the sound quality. I don't see how it's going to get any better after that. Gonna be very expensive at first, then expensive.
Last edited:
Yes the VC is the stabilized thing. The diaphragms must be as stiff as possible, but at some point you do need to have a second transducer with smaller area. Even if the emitting surface is sufficiently stiff there are still off-axis radiation issues.
But for near field wide range and large delta x, one can get a long way with the VC positional feedback and appropriate equalization, I think.
I was shipping subwoofers with servo controlled voice coils (servo'ed to DC) in the early 1980's. Its not really difficult and somewhat easier today. But closing the loop above 500 Hz would probably prove nearly impossible. While the magnetic field will react instantaneously the mechanics won't (and the cone will follow the speed of sound at best even if the mechanics did respond instantly).
I think some version of the Beverage speaker had a servo of some sort.
The midrange and the tweeter will still be open loop. A diaphragm that isn't modal probably will be at some higher frequency. Even the ceramic domes will break up.
If memory serves, this is more or less what Philips did with their Motional Feedback active speakers in the late 70ies, and with Grundig hopping on board later on (after it was purchased by Philips)
I do remember that their bass (the only driver they applied it to) was one of the, if not THE best bass I ever heard, and downright amazing for a box of that size. My feeling is that was one of the best Philips products ever in consumer audio.
But Brad is right, it wasn't only the electronics supplied, there was some serious equlization applied as well.
I do remember that their bass (the only driver they applied it to) was one of the, if not THE best bass I ever heard, and downright amazing for a box of that size. My feeling is that was one of the best Philips products ever in consumer audio.
But Brad is right, it wasn't only the electronics supplied, there was some serious equlization applied as well.
. 10 % THD looks nothing like the totally incorrect 0.10 % , one never uses the extra zero especially when a generality . Superfluous zeros would say enough to support your point, people do it I know . If I see it I say nothing because I am bound to do it myself sometimes .
It is shame I didn't keep some 10 % THD tests . The wave looks remarkably like an OK wave . The 5 % of mains distortion doesn't . To my way of thinking this explains the paradox of high distortion valve amps that sound OK and not obviously couloured ( English spelling , not an error ) . Coluration more often than not is frequency response . Some high distortion designs were dreadful in general . Reduce this to 1 % distortion and we get something rather good if the exponential rules are followed . These amps then stand or fall on damping factor . Some speakers are fine with a low damping factor ( > 3 works ) .
I am not a great fan of valve designs . I built one to know the truth . Truth is I could be won over . It took me six months of relentless work to get something worth calling my own . I use a mildly modified Quad 303 . It has very low crossover distortion . My speakers are Magneplanars . The Quad has a mildly MP3 affect on the sound . The detail should be greater , we are talking of the Quad being 98% good enough . The valve design does it better . If someone knows of a way of shifting the bias point to take the crossover distortion to 8 V peak I would be interested . Not a CCS or over biased class AB as I know those . One sided class G is the best I can think of . I will try a 8 V DC offset which has to carry through the amp if so into the preamp or use a level shifter like op amps do . Adding 8 V to a rail seems OK . Common-mode rejection might suffer . I will use TL071 and unbiased dumpers to be sure I know where the distortion sits as my starting point . So easy to fool oneself .
I have the flu today .
- Status
- Not open for further replies.