Haven't decided on the new version cos I still can't define the 'perfect' speaker. But why not DSP when your source is gonna be mostly evil digital?
Certainly helps with the best microphone for this millenium
6 is not enough, you either need more, or each 6 motor-controlled.
However, new media is needed, to contain special digital track to control motors.
Thanks for this John. I'm sure you do. But I was asking about Listening Tests. Didn't even use the rude "Bl*nd" word.
Anyone got John's 'proof' of zillion V/us with MC mistracking? Oops! I see he's posted it. Thanks John.
I've done a bit of work on the subject so am really interested. Gave up after Otala's student cos it was far more comprehensive than my stuff.
Thanks for this John. Looks entirely in keeping with the Shure data (which I think you quote) and summarized by the THX chap in Audio. Also consistent with Otala's student. I note you only guess what happens with HF mistracking while he measures it.
While HF mistracking does give probably [*] the highest slew rates with vinyl playback, it is not surprising that you don't get zillion V/us as HF mistracking is a very severe form of Slew Rate limiting.
[*] There IS one other case which is known to give higher slews but I won't complicate the issue.
Hi John,
Its useful that you put this up. It is one of the more reasonable points you made in that unfortunate fiasco that took place over thirty years ago.
It is always worth noting for perspective that the normalized slew rate for a 20kHz sinewave is 0.125 V/us/Vpeak. My original paper in TIM, called "Another View of TIM" can be found on my website at CordellAudio.com - Home. I believe that the points made in that paper have stood the test of time. Likewise my paper on PIM can be found there as well.
Cheers,
Bob
No need to remind-us about that, sir Cordell. I believe everybody on this forum and every engineer involved in audio have-it in his favorites and your contributions in his library. Your clever and clear contributions as well as your generous and modest way to share your knowledge and experience are universally recognized.
I say clever because all what you had published, or correlate with things we have experienced previously or can be verified by future experiences.
Not the case of all 'so called' gurus.
This words to express our gratitude and pay tribute.
John did you even read the results? There are many modern op-amps that have no PIM by Ron's own measurments. So why distrust global feedback when Ron's tests show no evidence for this at all? Get up to speed no PIM no TIM no FIM, why not address the results?
Last edited:
Mr. Marsh, there used to be a website with schematics of practically all hardware decoders but this died with Geocities.
I can only find http://decoy.iki.fi/dsound/ambisonic/motherlode/source/Surround%20Sound%20Decoders%20Pts%20567%20WW%201977.pdf This is a very old design but in fact ALL hardware decoders followed this topology as they were done by Dr. Geoffrey Barton under guidance from Michael Gerzon. The best in terms of sound quality would be the Minim AD10s.
But today, it is possible to use DSP to get far more accurate decode without passing through a zillion evil OPAs.
VVMic is one such beast. Do you have a good multi-channel soundcard on a convenient computer to play music?
Though VVMic is primarily used to decode Ambisonic B-format into stereo, it is also a surround decoder and can decode to 4, 6, 5.1, zillion.1
Simple 4.0 decode works well on most 5.1 systems.
An Ambisonic B-format recording doesn't record what should come out of the speakers. It records 'what you should hear'. The decoder turns this into speaker signals which, when they reach your ears, recreate the best approximation to what was happening at the microphone. To do this, you need to tell the decoder where your speakers are.
The problem is that for other than rectangles, regular polygons and other simple shapes, its quite difficult to dream up the correct decode. It's only this century, that the computing power has been available to do the optimisation at home. The BLaH team, Benjamin, Lee and Heller, have presented a series of papers on both Classic Ambi decoders as well as new ones which have arbitrary speaker positions.
But most people will get good results with simple rectangles and/or square decodes on their 5.1 systems.
And where you are in that anechoic chamber, sitting in the single point between that speakers.
sound field
Thank you.... I'll look into it further... I dont mind, for this appl using the evil opa... just newer opa models would be fine.... still better than 2 mic stereo. Or multi-mono mic techniques. Would love to find someone interested in dsp approach (not me-- too steep learning curve) -RNM
Thank you.... I'll look into it further... I dont mind, for this appl using the evil opa... just newer opa models would be fine.... still better than 2 mic stereo. Or multi-mono mic techniques. Would love to find someone interested in dsp approach (not me-- too steep learning curve) -RNM
This old sea snake !
No one was able to listen a difference between 5 OP260 in serial and a strait wire in a blind tests made in my studio, 20years ago.
Not to forget that most of the record we love to listen to are recorded and mixed with desks full of Op amps.
Op Amps are bad. It is well known.
Mostly because they are cheap and easy to use: so no mystery and difficult to sell zillion dollars hbv* preamps with them.
You need to find arguments instead of maths, for that, and global feedback evil can sound credible near some "idiophiles". And you can even convince them, using 741 or other slow devices to compare with your product.
Then, you open the boxes, and they see, one side a little cheap black Opamp lost in the middle of a printed board, and the other a huge assembly of good looking BIG unoptanium parts, with strange looking wires for more mystery, and gold everywhere...
*handcrafted by virgins
Last edited:
Amplifiers have no knowledge of symmetrical signals or not. They track levels.
More they are accurate, less distortion.
Of course, if they cannot follow a signal because they are not fast enough, they will make errors. So obvious.
If a signal is not statistically symmetrical accross a symmetrical amplifier, we will have some temperature disparities, then some non symmetrical increased distortions.
Global loop can correct that if slew rate have enough margin. And, on this point of view, Op amps have all the chances to behave better, because chips are little so temperature transmission is faster between silicon parts. And, because sizes are smaller, they can be faster than their discrete equivalent.
It is very interesting to look at the frequency and phase response curve in the input of any VAS stage of a closed loop amp. It is flat up to... 1 or 2 Khz for the fastest amps with >Mhz of bandwitch and 1000V/µs.
Not fast enough for perfection, yet...
More they are accurate, less distortion.
Of course, if they cannot follow a signal because they are not fast enough, they will make errors. So obvious.
If a signal is not statistically symmetrical accross a symmetrical amplifier, we will have some temperature disparities, then some non symmetrical increased distortions.
Global loop can correct that if slew rate have enough margin. And, on this point of view, Op amps have all the chances to behave better, because chips are little so temperature transmission is faster between silicon parts. And, because sizes are smaller, they can be faster than their discrete equivalent.
It is very interesting to look at the frequency and phase response curve in the input of any VAS stage of a closed loop amp. It is flat up to... 1 or 2 Khz for the fastest amps with >Mhz of bandwitch and 1000V/µs.
Not fast enough for perfection, yet...
Last edited:
I suppose all of them were configured for unity gain, right?
What if to get say 60 dB from each, with resistive voltage dividers between them?
IMO the optimal approach is the use of monolithic matched parts where possible, and thermal isolation of parts that undergo significant dissipation shifts with signal. Alternatively, make the dissipation shifts small or zero. And where possible use parts that have intrinsically small parameter shifts with temperature, e.g. FETs biased close to the zero tempco region.
We can do this with a hybrid of discretes and monolithic duals and other multiples. It took a while to appreciate the problems, but for many years now IC designers have done a great job of taking temperature excursions into consideration because they have to, not because they want to. It is a significant drawback to having power and small signal live in close proximity, not a benefit. On the other hand, for a dual part, signal-induced thermals are usually never fast enough due to proximity to cancel all that effectively. Ambient shifts or shifts from a local heat source are dealt with well if the layout is properly symmetrical, but transient differences between halves of a pair are messy. Fortunately, in a good design, at least the shifts due to signal at the front end are small and short-lived given adequate feedback.
Now when things have to go at blinding speeds, but require global feedback for accuracy, there is practically no alternative but to place the parts in close proximity. And for ICs, costs must be controlled, hence the more working parts off a wafer the better. In that regard diy folk will always be limited, although those limitations have diminished with the advent of surface mount packaging. Fortunately, audio bandwidths are fairly small, although doing a first-rate job at the interface between the symbol/numeric (aka "digital") domain and the analog domain is very demanding of both speed and accuracy.
Wrong.
i tried different girls before to be married too.
(With no comment on +/- inputs)
Nothing to add. Golden tips.
Or may-be one, thinking about terrific slew rates, and as you point... the output of a digital to analog converter !
About "audio bandwidths are fairly small", i agree, but if you want no phase shift up to 20Khz in a closed loop configuration, you need > Mhz of available bandwidth.
Last edited:
200 kHz 1st order LP: 2 ns group delay distortion @ 20 kHz !!
can't have "no phase shift" - physics doesn't allow
can't record with MHz system bandwidth - due to mic roll-off first, recording media limits too
practical geometry of measuring sound frequencies with wavelengths much smaller than diaphragm diameter suggest what is captured is so sensitive to relative orientation as to be characterized as “uncontrolled, chaotic” by 20 kHz
Earthworks does sell one 50 kHz "recording" mic - but they also advertise a "Drum Mic kit" with only 25 kHz mics
can't reproduce with "no" phase shift - LR4 crossovers with full cycle phase rotation thru mid to hi are still popular, as are various technology cone, dome dynamic tweeters - how does anyone with audiophile pretensions justify accepting anything but Electrostats?
there is plenty of practical evidence that "large" phase variation in upper octaves is inaudible, or presently completely accepted throughout the recording, playback chain - as long as it matches in both channels
why suddenly in just the audio electronics (especially SS, monolithic - not tube amps) is this "phase error" flag waving so popular?
Otala's “flat open loop gain” prescription for PIM reduction has been shown to be irrelevant - by measuremnets on real hardware – other choices can do better – as Scotts' summary of Ron Quan's latest paper shows
can't have "no phase shift" - physics doesn't allow
can't record with MHz system bandwidth - due to mic roll-off first, recording media limits too
practical geometry of measuring sound frequencies with wavelengths much smaller than diaphragm diameter suggest what is captured is so sensitive to relative orientation as to be characterized as “uncontrolled, chaotic” by 20 kHz
Earthworks does sell one 50 kHz "recording" mic - but they also advertise a "Drum Mic kit" with only 25 kHz mics
can't reproduce with "no" phase shift - LR4 crossovers with full cycle phase rotation thru mid to hi are still popular, as are various technology cone, dome dynamic tweeters - how does anyone with audiophile pretensions justify accepting anything but Electrostats?
there is plenty of practical evidence that "large" phase variation in upper octaves is inaudible, or presently completely accepted throughout the recording, playback chain - as long as it matches in both channels
why suddenly in just the audio electronics (especially SS, monolithic - not tube amps) is this "phase error" flag waving so popular?
Otala's “flat open loop gain” prescription for PIM reduction has been shown to be irrelevant - by measuremnets on real hardware – other choices can do better – as Scotts' summary of Ron Quan's latest paper shows
Last edited:
...until found obsessive-compulsive one that threw at midnight away stuff from the fridge that is expiring. I know. Try simple "whispering test": connect some fair microphone, like your favorite Neuman, and some very nice speakers, with class A tube amp, whisper close to the mic and slowly back up. Hear the difference: with 1 opamp whisper will go softer slowly and gradually, with 10 it will go more abruptly and less natural. Again, they must be connected with 60 dB gain and -60 dB pads between them. You may use headphones if prefer.
- Status
- Not open for further replies.