Adjusting the Lighter Note circuit to match the gain structure of associated components is discussed in the instructions. Yes, it makes a difference though it didn't, in my setup, fully overcome an imperfect gain structure.
Speaking for myself, I am well aware of the difference between linear and non-linear distortion. I doubt I'm alone in that.
Thank you for your response, Jan...
NP's description of his test: "These curves were created with a 600 ohm source impedance, and the current through the diode was set to produce attenuation of -0.1 dB, -0.5 dB, -1 dB, -2 dB, -3 dB, -6 dB, and -10 dB which corresponds to 50 Kohm, 10k, 5k, 2.3k, 1.5k, 600, and 277 ohms."
-From this we can deduce that the attenuator is created with the 600 ohm source impedance of the function gen as the series R and the LDR (DUT) as the shunt R to ground.
-He then adjusted the control LED or "diode"[sic] so the LDR is set for 50kohm, 10k, etc...this is done one at a time to produce each of the distortion response curve at each attenuation value -0.1dB, -0.5dB, etc.
-As an example, we'll choose -6dB for simplicity. At 200mV input to the attenuator, meaning the function gen end of the 600R source impedance that outputs to the other end into the 600R shunt LDR to ground, has an output voltage of 100mV across the LDR. Distortion is then noted across the LDR. The function gen is then adjusted to a new input level and distortion noted and so forth until a curve is generated. For 300mV input we have 150mV; distortion noted, etc...
-The x-axis is then the voltage across the LDR plotted with its distortion on the y-axis.
No I have no intention to listen to them, I don't have them. I assume that someone put up those graphs thinking they have some meaningful info. I think they do; I was just bouncing off my interpretation of those graphs. Sorry if I hurt anyone's feelings doing so.
Jan
Thank you for your response, Jan...
NP's description of his test: "These curves were created with a 600 ohm source impedance, and the current through the diode was set to produce attenuation of -0.1 dB, -0.5 dB, -1 dB, -2 dB, -3 dB, -6 dB, and -10 dB which corresponds to 50 Kohm, 10k, 5k, 2.3k, 1.5k, 600, and 277 ohms."
-From this we can deduce that the attenuator is created with the 600 ohm source impedance of the function gen as the series R and the LDR (DUT) as the shunt R to ground.
-He then adjusted the control LED or "diode"[sic] so the LDR is set for 50kohm, 10k, etc...this is done one at a time to produce each of the distortion response curve at each attenuation value -0.1dB, -0.5dB, etc.
-As an example, we'll choose -6dB for simplicity. At 200mV input to the attenuator, meaning the function gen end of the 600R source impedance that outputs to the other end into the 600R shunt LDR to ground, has an output voltage of 100mV across the LDR. Distortion is then noted across the LDR. The function gen is then adjusted to a new input level and distortion noted and so forth until a curve is generated. For 300mV input we have 150mV; distortion noted, etc...
-The x-axis is then the voltage across the LDR plotted with its distortion on the y-axis.
OK, I get that. Then I may assume that at normal listening levels at say -20dB or -30dB the required 'R' of the LDR would need to be even lower than the 277 ohms at the -10dB curve, probably below 100 ohms. Since the graph shows increasing distortion with lower 'R' I expect the distortion in these practical cases to be significantly higher even than the highest shown at -10dB. Am I wrong in this reasoning?
Jan
-20dB or -30dB of attenuation is like negating all the voltage gain of the power amp. Highly likely only if there is a preamp gain stage before the power amp.
Attenuation is -15dB or less with passive preamps...all the more reason that I prefer passive linestages.
An example would be Dire Straits's Brothers in Arms album which peaks at 1.28V. For a 5V peak on an 89dB/W loudspeaker connected to a 26dB (x20) power amp, these would reach a peak of 94dB SPL at 1m. Music being dynamically changing and thus doesn't sustain this peak, it is comfortable listening in a decent sized room. To reach the 5V peak the power amp needs 0.25V at its input.
The book I found that in is, Industrial Noise Control by Lewis and Douglas Bell.
But the knowledge seems to be fairly well known among Audiologists, Oncologists.
If you read my first post I hope its is obvious that I'm on the objective side of things, but I know there is a lot of mystery in this life too. I build audio equipment sometimes for no other reason to see what the hype is all about. I was very skeptical until I build one. The reason is the expansive and compressive distortion characteristics. It's almost like slight loudness effect, it's euphonic, maybe not accurate but definitely unique and interesting sound. It won't replace my dcb1 in my other system, but with a hot rodded chipamp, oh my those dynamics...
Here's a experiment if you have one or access to one (ldr). Use low gain amp preferably less than 15db. Play pink or white noise through it and see when you begin to notice a difference when the noise seem like less distinct particles and becomes a wash very quickly, for me it was around -8db. this wasn't just loudness but a distinct blurring of the noise.
Duly noted and apologies cormeister--reading comprehension last night wasn't as good as it should have been. And I had a look around the net (I don't know what search params I was using before?) and that corroborates with your claims that tympanic displacements are in fractional-Angstrom ranges at high frequencies.
Interesting effect you're observing for sure.
Interesting effect you're observing for sure.
So what? We can measure that which we can hear -- hearing works based on physics and not magic -- not a big surprise.
But to infer from the threshold of hearing (specifically a 1 kHz tone played at 0 dBSPL into young, "undamaged" ears in an anechoic chamber) that there is somehow a mystery, that we cannot measure stuff, or that that would allow people to hear details anywhere near that while playing music at usable SPL would all be a big fat non sequitur.
But there is one thing that people hear that is not measurable in the sound waves. It's caused by biases, expectations, ... and is not a feature of the sound but a product of one's brain.
This happens especially when one buys a new toy. Add being convinced by marketing statements (that are actually nonsense) and some ideological assertions and you've got yourself a following for any piece of equipment.
@jan: Some attenuators may have LDRs going below 100 ohms and may distort horribly, yes. But the graph posted earlier showed a product where min impedance was limited to about 100 ohms.
But to infer from the threshold of hearing (specifically a 1 kHz tone played at 0 dBSPL into young, "undamaged" ears in an anechoic chamber) that there is somehow a mystery, that we cannot measure stuff, or that that would allow people to hear details anywhere near that while playing music at usable SPL would all be a big fat non sequitur.
But there is one thing that people hear that is not measurable in the sound waves. It's caused by biases, expectations, ... and is not a feature of the sound but a product of one's brain.
This happens especially when one buys a new toy. Add being convinced by marketing statements (that are actually nonsense) and some ideological assertions and you've got yourself a following for any piece of equipment.
@jan: Some attenuators may have LDRs going below 100 ohms and may distort horribly, yes. But the graph posted earlier showed a product where min impedance was limited to about 100 ohms.
Last edited:
Especially the point about temperature drift is very interesting and could cause audible deviations.
Another thing to keep in mind is that all these nonlinear effects will probably be significantly different for each individual LDR except if you could match them perfectly. For example, one of Pass' measurements showed that when taking two random LDRs then one can have 5k and the next one over 40% more (7.2k) at the same current.
Regarding 5) In order to significantly change the dynamics you'd need lots of distortion. A dynamics processor is probably a better choice since its more flexible and cleaner.
Last edited:
The scientific study of the normal function in living systems is physiology
not physics.
"A sub-discipline of biology, its focus is in how organisms, organ systems, organs, cells, and biomolecules carry out the chemical or physical functions that exist in a living system"
.https://en.wikipedia.org/wiki/Physiology
Which goes a long way to explain why measurements have
a long way to go, catching up with what we humans can hear.
120 ohms is typical in ldr kits I do, and dead quiet at zero volume, so no need
to stress the LDR. The recipe: No capacitance or resistance directly in
parallel with the LDR anode and cathode. Cathode ground is best independent of
signal grounding, and can use either a potential difference, or at the least star
grounding. Use a single gang pot.
Cheers / Chris
You missed the point.
That smells like another non sequitur.
Lacking explanations for some limitation or feature of human hearing has no impact on what e.g. a microphone or signal analyzer can measure.
For example, we can measure distortion products that human hearing will simply mask, or at levels where human hearing simply does not work properly anymore. Even if we had no idea what masking is we could still measure stuff that is far beyond our hearing capabilities.
How do you think fields like psychoacoustics advance? We can generate stuff outside human hearing capabilities, then we do proper blind tests to determine e.g. hearing thresholds. Of course we can also measure this stuff way outside human hearing capabilities.
---
Regarding LDRs, even if we had no idea what to look for in measurements (which is not true since we know of their impedance and signal nonlinearities), we could simply do a null test.
The remaining difference would be noise or distortion or whatever these LDRs do to alter the signal.
That's good to know and should prevent crazy distortion, but if your kits require manual matching then I'd like to know how these LDRs are matched.
What are the tolerances involved? How is the matching over all volume control positions?
I think this could also account for some of the heard differences.
I'm not surprised. The point is almost as banal as the cod psychology in your post. It amuses me how EEs ridicule audio hobbyists (often justifiably though less often gracefully) but can themselves show an equally striking ignorance of psychology. As soon as I read the term "Blind Test" or, worse, "DBT", I know I'm in all probablility reading someone who wouldn't recognise a properly conducted psychological experiment if it ran them over. Look e.g. at the mileage accorded for years to that absurd Meyer Moran BAS study.
The same is true of pots, SAs, lengths of wet string, etc etc. I don't get your point.
LDRs are always matched when used in passive attenuators. I'm surprised that someone with opinions on the technology as strong as yours does not know that. The techniques the various designers use have been discussed ad nauseam in forums and elsewhere; I see no need to repeat their reports here.
Again in light of this thread and with memory jogging, I just measured the resistance of the series and shunt LDRs in my Lighter Note with the knob at the 8 o'clock and 4 o'clock positions - as low a volume as I can usefully use at one end and higher than comfortable at the other. The values are about 10k at one extreme and 20k at the other for both devices, one, naturally, the inverse of the other. These values may not be entirely typical but I'd suggest they're not too far out either.
That Silonex app note that no-one wants to read reports that distortion at the volume extremes does indeed climb but adds that it is also, for obvious reasons, imperceptible at those points. Absent caveats, assessing LDR distortion performance at values in the region of 200R is likely to mislead.
That much is very clear from your response.
Btw, I did not try to ridicule anyone, unlike you:
This is what we call being dishonest, trying to make me look bad by attacking a straw man.
I even explicitly mentioned the necessity of manual matching ... Doh!
The "if" was in response to Chris talking about his kits, because I am not certain if they are all passive - I have no idea what he sells.
Could you please write down the precise measured resistance of each individual LDR at either those positions?
"about 10k" doesn't contain any useful information.
Ok so you are saying this distortion has no audible effect. We really should start keeping score of the conflicting positions that people (including owners of such attenuators and manufacturers) hold.
So we are back to "people hear what they want to hear" regarding changes in dynamics, pronounced bass etc.
Last edited:
-20dB or -30dB of attenuation is like negating all the voltage gain of the power amp. Highly likely only if there is a preamp gain stage before the power amp.
Attenuation is -15dB or less with passive preamps...all the more reason that I prefer passive linestages.
An example would be Dire Straits's Brothers in Arms album which peaks at 1.28V. For a 5V peak on an 89dB/W loudspeaker connected to a 26dB (x20) power amp, these would reach a peak of 94dB SPL at 1m. Music being dynamically changing and thus doesn't sustain this peak, it is comfortable listening in a decent sized room. To reach the 5V peak the power amp needs 0.25V at its input.
So listening with the volume turned back even more than that would not be an exception? With my equipment I often listen below -20dB on the preamp. From the earlier given graph by Nelson it seems that if I would have such an optical attenuator I would routine sit at around the 1% distortion point.
Not that this is a disaster - things can sound pretty nice at 1% distortion - LPs come to mind.
But I just wanted to get some clarification about all this 'transparency' stuff.
Jan
Another idea, possibly old hat: has anyone tried to use a pair of matched LDRs to make a closed loop system? I have done something similar in an optical isolator, which has two matched photodiodes and a common LED in a DIL package. One opto-diode is in the forward path and the other in the feedback loop of the driving circuit. I got linearity across the isolator of better than 0.1%.
Jan
Jan
I cannot find where I [allegedly] said that so cannot check the context. Who knows? I might have been right. For once.
What on earth for? I was always taught to eschew phoney "precision" in measurements because it tends to mislead. My point was that if either the shunt or series resistance in an LDR-type series-shunt attenuator has a value in the region of 200R, the system is either silent (or very quiet indeed) or it's at/ close to maximum volume. In either case (esp the first), distortion measurements have little meaning in perceptual terms. I really can't see what's difficult about that.
IOW in everyday use in my system, LDR resistances vary in practice between 10K, give or take, and 20K, ditto. They vary according to source material, mood and diurnal mains fluctuation. In a different system, they might vary between, say, 5K and 100K. Even a 5K value is well over an order of magnitude (folk here seeming to like the phrase) larger than 200R. IOW, I agree with Silonex that the device's distortion performance at values in the region of 200R (or Megohms at the other end) tells you nothing useful about its performance in the real world. Or, if you prefer, precisely nothing.
That doesn't negate debate about how LDRs perform, some of which has been useful. I'm suggesting only that it be conducted sensibly.
Note that taking an LDR below 40R will pretty quickly blow it. Low values are in practice pertinent only when the user wants to mute the system. A known issue in setups with sensitive speakers is that they may not mute fully at minimum volume. Obviously, I don't care much about HD when the thing is silent but nor am I unduly bothered how much there is when it's either so quiet I can barely hear it or so loud the neighbours are complaining.
So, when you ask:
I'm saying (with Silonex) that under the conditions described the distortion has no audible effect.
Okay, I'll bite. If you're going to lob grenades at EE's about their ignorance of psychology, then you'd better back it up with some serious critiques. Otherwise, you get points for banalities AND lack of content. Specifics, not generalities, please.
Last time I checked, 3/4 of what I read from psychology academics (this is not my field, although I'll flatter myself by saying I do pretty okay in scientific literacy) I follow is that psychologists don't design good psychological experiments. Or, at least, the interpretations of experiments are generally too broadly applied.
2V (0.1%) is the output voltage for a 6.33V input at -10dB attenuation.
If you mean 2V input voltage then with -10dB attenuation it will be 0.63V output voltage. This approximates to 0.018% of distortion...that's about -75dB! Will that be audible to you?
Good point yes that's much better!