D
Deleted member 375592
I'd say EQ is mandatory with both voltage and current drive to get a ruler-flat (on-axis) loudspeaker. Inverting IR is the easiest. But you do need a decent DSP+codecs (not the flaky and shaky junk we get from the internet) or PC+EqualizerAPO+soundcard... none of which existed in the 80s.
Moreover, with the current drive you may be able to cancel H2 for lower frequency by Volterra adaptive filtering, which is not possible with the voltage drive.
So far, I have not seen a tweeter benefiting from the current drive. Ribbon/AMT too. Sub/woofers for f<100Hz too.
Oh but they had MFB in various flavors and nearly perfect analog circuits. Still it didn’t really make it. Weakest chain thing.
D
Deleted member 375592
I agree - the weakest chain link, for sure.
Infinity did make it using the "army surplus" piezo they got from their friends in NASA. Velodine was making MFB subwoofers for a while (which cost $$$). Yamaha has been making "direct" MFB subs. Zami Schwartzman from Israel has been making them in his garage ... while complaining bitterly that he could not find any vendor or investor who would take his field-proven design to the market. As a normal genius inventor, he bumped and still bumps into outright rejection, time and again...
I also experimented with piezo sensors... and found that my old shaky fingers are of no good:-( I would gladly cooperate with anybody who has golden hands.
Infinity did make it using the "army surplus" piezo they got from their friends in NASA. Velodine was making MFB subwoofers for a while (which cost $$$). Yamaha has been making "direct" MFB subs. Zami Schwartzman from Israel has been making them in his garage ... while complaining bitterly that he could not find any vendor or investor who would take his field-proven design to the market. As a normal genius inventor, he bumped and still bumps into outright rejection, time and again...
I also experimented with piezo sensors... and found that my old shaky fingers are of no good:-( I would gladly cooperate with anybody who has golden hands.
Velodyne marketed motional feedback with their subwoofers.
Systems were designed using a coil in front of the woofer in line with the VC, and also with a winding on the voice coil. You can't afford the space on the VC for anything but the driving winding. I heard of a laser feedback system as well, not sure if it ever hit the market.
Systems were designed using a coil in front of the woofer in line with the VC, and also with a winding on the voice coil. You can't afford the space on the VC for anything but the driving winding. I heard of a laser feedback system as well, not sure if it ever hit the market.
We over here actually have a forum “mfbfreaks.nl”. Heritage of Philips systems, but they broadened their views. Modern DIY piezo boards plus accompanying electronics are for sale if you know where to look. German high end brands like Backes & Müller and T&A had their implementations too, back in the 80s.
Hi markbakk,
Should be good once they get it going.
Somewhere I have the Philips speaker databooks. I have some mid-tweeter combination plates and back then designed systems for a store using Philips (and other) components.
Should be good once they get it going.
Somewhere I have the Philips speaker databooks. I have some mid-tweeter combination plates and back then designed systems for a store using Philips (and other) components.
D
Deleted member 375592
No wonder you Dutch folks are THE world leaders in Lorenz actuators and run ASML if you have an MFB forum just for fun🙂
No, I don't know where to look for piezo boards and accompanying charge amplifiers, etc. Could you advise?
I think you and some others are mixing somewhat different things, and what you're opposing is not quite what others are suggesting.
If you've ever seen/heard a piano tuner at work, you may have noticed how they use a tuning fork to get calibrated and start the tuning. The tuning fork obeys the laws of physics and is absolutely, highly reliable. Then they tune the entire rest of the piano by ear, only using the fork as a sanity check from time to time.
No tuner that I've seen relies on a full scale of tuning forks or to check each octave. Piano harmonics are stretched, and each string is its own unique instrument, so an octave and the other intervals must be tuned so they sound right, not so they look right on measurement equipment.
~~~
Going back to designing by ear, I would take it to mean certain things like:
Balancing a long-tail pair amplifier stage and letting the ear be the final judge, at least in a DIY setting. I guess in a professional setting if your biggest concern is consistency at high manufacturing volume, then you probably would not pick a design where you have to balance LTPs in the first place. Even so, the owner of a DIY system gets unique insights with specific listening material and the position of the volume control, whereas a manufacturer can only guesstimate how the equipment will be used (so what are they fine-tuning?)
If a DIYer goes to the trouble of getting a distortion analyser and setting distortion levels visually, in spite of their ears, I think it leaves them vulnerable to errors in the theory. (Potentially something even worse: damaging the psychology of their hearing, as it may amount to negative training if they get something wrong and don't know about it.) While focusing on reducing harmonic distortion, they may miss an increase in intermodulation, which, eg looks 20dB lower but sounds worse. To my knowledge, there is only some soft empiricism (Voodoo) regarding what is likely to sound better or worse based on visual graphs, but no hard formula that guarantees it.
I'm guilty of this, too. Just the easy availability of THD estimates in simulators skews ('scuse?) my decisions, and frankly reduces the scientific robustness of the design process. I've heard people say it's OK as long as we are aware of the limitations of the simulated system, take it with a grain of salt, etc. But we are not fully aware of the limitations. Known unknowns vs unknown unknowns.
Part "rolling" often seems to border on snake oil, but there are areas where I try to keep an open mind. For instance with capacitor types. Upgrading my active filters from polyester caps to polypropylene opened my eyes, and now I don't blink even if someone wants to explore oil and paper dielectrics. TINA won't tell me how C0G, mica, or polypropylene will sound different in a feedback loop. It's 'lab' homework that has to be done by ear.
Hi abstract,
A piano tuner is gauging frequency and is highly trained. Some people can't do this work. I have watched many at work. But this isn't distortion or level where we fail terribly, it is frequency and some of us are pretty good at that. I've tuned many organs with a frequency counter - then listened to it. No pianos, no thanks.
If the LTP is not balanced, you will have higher distortion (given everything else is okay). Your DC offsets will be higher than they should be, and they will drift with temperature. Manufacturers cannot afford the time to match parts, so they came out with DC offset pots first, then servos to eliminate the tech. It is all about money there.
A THD pointer (as in 1970s) is only one indication and you also need IMD. Heathkit made tube models so they were around for ages and affordable. My first one was about $2500, and I used it with an oscilloscope and spectrum analyzer. The scope showed crossover distortion. The HP 3580A was available in the 1970s, also the HP 35665A later on. These are audio spectrum analyzers and give you much more information. Today we have the AP AP555x and Keysight U8903B, those cover far more than you can ever detect as a human. The RTX 6001 I use is similar. We look at spectrums for IMD and THD.
After saying all that, something is better than nothing even if it isn't everything. I constantly get equipment repaired by others "working perfectly" with high distortion. They say they can hear distortion. I can tell you that levels well over 1% are returned to customers regularly. The reasons for the distortion would have pointed these "techs" to circuit problems.
I can measure the effects of various components in an amplifier. I can measure parameters in capacitors and inductors using the correct equipment. I little HP 4263A will do it, and I also use two others (HP 4192A and ESI Digibridge). Y ucan actually measure and select component types that will have lower distortion in certain locations. Not coupling caps. Frequency filter capacitors will be audible if they aren't good types, supporting your example. But I can measure them and know before listening to anything. Yep, I listen. An old habit but it is a good check.
Simulators are inaccurate unless they show problems. Sim results are fiction. Board layout, component types and matching all play a large role. Normal simulators don't model that.
Naw, you don't have to do a single thing by ear anymore. Sorry. I listen as a sanity check and to make sure nothing else is going on. Not once in the last 20 years have I heard anything I didn't see on the screen first. This is the state of technology and measurements today. Mind you, if you don't have the equipment you are in the dark a bit at least.
A piano tuner is gauging frequency and is highly trained. Some people can't do this work. I have watched many at work. But this isn't distortion or level where we fail terribly, it is frequency and some of us are pretty good at that. I've tuned many organs with a frequency counter - then listened to it. No pianos, no thanks.
If the LTP is not balanced, you will have higher distortion (given everything else is okay). Your DC offsets will be higher than they should be, and they will drift with temperature. Manufacturers cannot afford the time to match parts, so they came out with DC offset pots first, then servos to eliminate the tech. It is all about money there.
A THD pointer (as in 1970s) is only one indication and you also need IMD. Heathkit made tube models so they were around for ages and affordable. My first one was about $2500, and I used it with an oscilloscope and spectrum analyzer. The scope showed crossover distortion. The HP 3580A was available in the 1970s, also the HP 35665A later on. These are audio spectrum analyzers and give you much more information. Today we have the AP AP555x and Keysight U8903B, those cover far more than you can ever detect as a human. The RTX 6001 I use is similar. We look at spectrums for IMD and THD.
After saying all that, something is better than nothing even if it isn't everything. I constantly get equipment repaired by others "working perfectly" with high distortion. They say they can hear distortion. I can tell you that levels well over 1% are returned to customers regularly. The reasons for the distortion would have pointed these "techs" to circuit problems.
I can measure the effects of various components in an amplifier. I can measure parameters in capacitors and inductors using the correct equipment. I little HP 4263A will do it, and I also use two others (HP 4192A and ESI Digibridge). Y ucan actually measure and select component types that will have lower distortion in certain locations. Not coupling caps. Frequency filter capacitors will be audible if they aren't good types, supporting your example. But I can measure them and know before listening to anything. Yep, I listen. An old habit but it is a good check.
Simulators are inaccurate unless they show problems. Sim results are fiction. Board layout, component types and matching all play a large role. Normal simulators don't model that.
Naw, you don't have to do a single thing by ear anymore. Sorry. I listen as a sanity check and to make sure nothing else is going on. Not once in the last 20 years have I heard anything I didn't see on the screen first. This is the state of technology and measurements today. Mind you, if you don't have the equipment you are in the dark a bit at least.
D
Deleted member 375592
It is not undebatable... I have a friend who is a piano tuner. They are not the same, each one works differently. Many of them use sophisticated spectral analysis instruments which drastically shorten the tuning process. Just a tuning fork is surely a stone age. On different instruments, they may or may not employ this or that. They endlessly argue on their forums about the best way to do the work in the shortest time - which is very critical because sometimes they are called a couple of hours before the actual performance, and sometimes they have to do the work very quickly and efficiently - people are unwilling to pay a lot of money for tuning an old rarely used home piano.
Yet, I agree that objective measurements do not tell the full story.
Yet, I agree that objective measurements do not tell the full story.
The popularity of the LTP as building block belies its complexity. There are competing distortion mechanisms within it, not counting possible interactions with other parts of the circuit.
If the goal is minimum even harmonics (because that's what a basic LTP is good at canceling out), amplitude modulation of those same harmonics could be at a maximum. Conversely, if the LTP is detuned, the ratio of best-case vs worst-case amplitude of those harmonics could be decreased. Put another way, adding a continuous tone to a tremolo with the exact same centre frequency decreases the depth of the tremolo. AKA masking. And a distortion analyser won't tell you the optimum balance between the two.
Taken to its logical conclusion, an emitter feedback circuit could be used instead, eliminating the issue of 2 very similar parts trying-but-sometimes-failing to cancel each others distortion. And then we are back to square one, with "inferior" circuits being chosen by "bad" designers. Or worse: bulky and "inferior" components, which, to some only appear to increase THD, but to others they reduce temperature modulation and related effects.
If the above is discounted or unknown, that could be an example of negative training like I mentioned earlier. The analyzer gives a conflicting report that circuit X ought to sound better because of its lower THD, leading to a questionable interpretation that it was your ears that failed.
Hi abstract,
Interpreting test results is a skill. So far everything I have predicted from test results has been confirmed by listeners. Now I a m nasty. I'll give them products that haven't been touched, have been simply serviced and some modified. They never know and I'm not there. I never assess my own work. Subjective comments agree with predicted test results with a few individuals that disagree with the bulk. In other words, valid testing. I've done this for decades.
Testing is never in conflict with subjective opinion. That is unless you have someone who prefers the sound of a class A single output tube amplifier or similar, then they say things like "this sounds more musical". Orchestral music is unlistenable on those systems. The average person strongly prefers the lower distortion system barring other obvious defects. Many might say they can't hear a difference once performance reaches a certain level for them.
You can argue theoretically if you want, but most practical LPT amps have lower third. This is a function of the entire circuit. The goal is simple as heck. Minimum distortion. Don't care what kind.
Humans will not hear past -100dB below 1 watt in my experience. Maybe a 1KHz tone - possibly maybe. But not distortion in music. -80dB is the limit for common average people. These are extremely rough figures borne from years of observation and measurements.
Right now I'm looking at spectrums from 10 Hz to 96 KHz, noise floor is below 135 dB below 1 watt. There simply isn't anything anyone can hear that doesn't stick out like sore thumb.
Interpreting test results is a skill. So far everything I have predicted from test results has been confirmed by listeners. Now I a m nasty. I'll give them products that haven't been touched, have been simply serviced and some modified. They never know and I'm not there. I never assess my own work. Subjective comments agree with predicted test results with a few individuals that disagree with the bulk. In other words, valid testing. I've done this for decades.
Testing is never in conflict with subjective opinion. That is unless you have someone who prefers the sound of a class A single output tube amplifier or similar, then they say things like "this sounds more musical". Orchestral music is unlistenable on those systems. The average person strongly prefers the lower distortion system barring other obvious defects. Many might say they can't hear a difference once performance reaches a certain level for them.
You can argue theoretically if you want, but most practical LPT amps have lower third. This is a function of the entire circuit. The goal is simple as heck. Minimum distortion. Don't care what kind.
Humans will not hear past -100dB below 1 watt in my experience. Maybe a 1KHz tone - possibly maybe. But not distortion in music. -80dB is the limit for common average people. These are extremely rough figures borne from years of observation and measurements.
Right now I'm looking at spectrums from 10 Hz to 96 KHz, noise floor is below 135 dB below 1 watt. There simply isn't anything anyone can hear that doesn't stick out like sore thumb.
Then a larger source of distortion must be located elsewhere, making LTP performance less critical. Sort of like arguing about amplifiers that achieve 120-140dB signal to HD, even though they only manage to drive a speaker with 60dB signal to HD.
Meanwhile a "musical" class A amplifier could stretch that to 70dB signal to HD (as described in a nearby thread). Yet it is still considered technically worse by those who only measure electrical distortion with 8 ohm dummy resistors, but never sanity-check their findings with real speakers.
It makes sense that people could hear 20dB or so into the noise floor of speaker distortion, thus being able hear subtle differences between various amplifiers. However, there seem to be lots of opportunities to accidentally train your ears to hear 10dB less distortion as worse rather than better.
Your diff pair (LTP) is where the output signal is compared to the input signal. The degree of match determines how accurate this is, so a well matched pair will give you lower distortion. I have proved this experimentally countless times. There is a limit to what it can correct for of course. A low distortion output stage will deliver lower distortion with the same voltage amplifier (again, proved experimentally). So the entire design hinges on the voltage amp, and the diff pair (complimentary included). But the rest also plays a part in that errors elsewhere have to be corrected by the LTP. So you start with the most accurate/linear, high gain input stage possible. You match the required parts as closely as possible and you've done your job there. Now you have to create a current amplification stage that also has low intrinsic distortion. Less correction required. This combination will both perform and sound the best. The test equipment will also read lower distortion and noise floor.
I always sanity check measurements with listening tests. I have some good speakers on the bench, 99 dB/watt (Klipsch THX 6000), some power hungry speakers in the main system, PSB Stratus Gold, 86 dB/watt and 4R, then some others (Klipsch RP-8000) at 98 dB/watt elsewhere. They are all played very loud at times, normally the same time. I have wonderful neighbours! lol! So when I measure something that is mine or will be here for a while, it is subjected to a lot of listening at various power levels. With the high efficiency speakers, no noise escapes detection. I also installed a pair of Klipsch Jubilee speakers at a friends place, electronic crossover. 108 dB/watt and simply awesome. Any noise is front and centre and we tried a few amplifiers. Bryston 4B cubed got the nod. So I am not one of those people who do not listen critically, just reading the instruments. But I hate to tell you this, get the right instruments, run the right tests and know what they are telling you, we pretty much know what something will sound like to a human.
The main problem with a human is the intelligence unit between their ears. It can have the final say on what you think you hear. The high end industry operates on this to make money. Expectation bias is the weapon and it is very, very effective. That's why you should listen to what children and uninterested women say when hearing a system.
These days when a claim is made an inferior technology is somehow "musical", that should shoot up warning flags to any sane person. Anything they sell today was created decades or even a century ago. Those technologies were left behind because they were inferior. That is the cold, hard truth. The other truth is, if you can differentiate yourself with a product, you can sell it. It's all about money and ego.
I always sanity check measurements with listening tests. I have some good speakers on the bench, 99 dB/watt (Klipsch THX 6000), some power hungry speakers in the main system, PSB Stratus Gold, 86 dB/watt and 4R, then some others (Klipsch RP-8000) at 98 dB/watt elsewhere. They are all played very loud at times, normally the same time. I have wonderful neighbours! lol! So when I measure something that is mine or will be here for a while, it is subjected to a lot of listening at various power levels. With the high efficiency speakers, no noise escapes detection. I also installed a pair of Klipsch Jubilee speakers at a friends place, electronic crossover. 108 dB/watt and simply awesome. Any noise is front and centre and we tried a few amplifiers. Bryston 4B cubed got the nod. So I am not one of those people who do not listen critically, just reading the instruments. But I hate to tell you this, get the right instruments, run the right tests and know what they are telling you, we pretty much know what something will sound like to a human.
The main problem with a human is the intelligence unit between their ears. It can have the final say on what you think you hear. The high end industry operates on this to make money. Expectation bias is the weapon and it is very, very effective. That's why you should listen to what children and uninterested women say when hearing a system.
But .... it is worse! A lot worse. These generally sound best on simple music only. Try listening to anything complicated musically. They sound irritating, annoying. The last hi-fi / music show I attended, there was no shortage of single ended art being sold for very high dollars, attached to art they called speakers. Not one sounded good. Not compared to what actually does sound good. They wind you up with a good story. Typically some genius designer using special parts that were "voiced" for the system and a good story. Coupled with the high dollars, the innocent person expects something that was so much trouble and expense to actually be good, maybe the best. The story clinches it. Same old story, and it is romantic.
These days when a claim is made an inferior technology is somehow "musical", that should shoot up warning flags to any sane person. Anything they sell today was created decades or even a century ago. Those technologies were left behind because they were inferior. That is the cold, hard truth. The other truth is, if you can differentiate yourself with a product, you can sell it. It's all about money and ego.
Distortion (many kinds) can hide a multitude of other distortions.. peel back the layers and discover more. Often it can sound worse but that doesn't mean you should stop 😉
If a class A amplifier has been designed as a tone machine, that's one thing.
If it's been designed as a low distortion device then I might disagree (anything in between is anyone's guess).
Hi AllenB,
Agreed. They generally are for "that sound" from what I have seen. That's single ended.
Class A push-pull amplifiers just simply run hot as heck, but can sound good of course. However, the level of bias current is set more by marketing than technical necessity.
Agreed. They generally are for "that sound" from what I have seen. That's single ended.
Class A push-pull amplifiers just simply run hot as heck, but can sound good of course. However, the level of bias current is set more by marketing than technical necessity.
If we are speaking about electronics we are in complete agreement.
When we continue the conversation into the domain of speakers other things creep into the level of auditability and are also very much measurable.
The thing that we think of first is frequency response.
The thing most offensive to me is amplitude modulation (IMD)
And a close second pet peave is Bass Reflex port noise plus mid-range leakage.
Thanks DT