I have not seen much mention of the fact that an LDR can do what no simple pot does and that is independently vary the series and shunt resistance. I suppose a stepped attenuator could be designed with a specific system in mind. There are lots of things that could be at play here related to source/load impedances.
This thread reminds me of Beckett's "Waiting for Godot" only less satisfying.
Waiting for LDR
"Have you listened to it? "
"No"
"How does it sound?"
"It can't be good."
"How would you know?"
"How do you know anything?"
"We should listen to it."
"What if we like it?"
"That can't happen."
"So you'll listen to it."
"No. I won't like it."
"Well then, you're right then I suppose."
"Someone said they liked it."
"Then they're wrong of course."
"Yes, I'm sure you're right".
"Good to be right."
"Have you seen one?"
"No. No need".
"Right."
"We could measure it."
"Yes, but the results would be bad."
"Could we hear it then?"
"No, no. Need only look at the measurements."
"Can you hear measurements?"
"Yes...well, no."
"We should listen to one."
"Maybe."
"No. You're right. Let's not."
Tortuga Audio does exactly what you describe. The series and shunt are adjusted independently to arrive at both a fixed input impedance (over most of the 60 dB attenuation range) plus the attenuation level of each of 70 steps.
Let me cut down that lengthy post:
So CaptainWatt, have you tried magic pebbles yet? There are many people with positive impressions...
Regarding impedance in that LDR3 preamp:
(source: tortuga)
Impedance starts rising at about -45 dB, other than that it's just 20k.
So CaptainWatt, have you tried magic pebbles yet? There are many people with positive impressions...
Regarding impedance in that LDR3 preamp:
(source: tortuga)
Impedance starts rising at about -45 dB, other than that it's just 20k.
Last edited:
Thanks, not exactly what I was thinking. I'm not ready yet for the brilliant pebbles analogy and the distortion stuff could still be a red herring. I'd love to see some in situ analysis of signal integrity.
There don't appear to be many measurement results so I will add my two penny worth.
I looked at using a Silonex NSL-37V51 (I think it was) in a notch filter trim a long time ago. In the particular circuit configuration I used with 12.5 V peak to peak across the divider I got 0.22 % distortion mainly third harmonic.
With 1 V peak to peak I got less than 0.05 %.
Obviously with a different attenuator configuration and a different LDR the results will be different.
If any one has an LDR based volume control that I can measure I am happy to do so.
I looked at using a Silonex NSL-37V51 (I think it was) in a notch filter trim a long time ago. In the particular circuit configuration I used with 12.5 V peak to peak across the divider I got 0.22 % distortion mainly third harmonic.
With 1 V peak to peak I got less than 0.05 %.
Obviously with a different attenuator configuration and a different LDR the results will be different.
If any one has an LDR based volume control that I can measure I am happy to do so.
I was quite specific in writing the results that I heard in my audio system with the LDR. Some of my results may have been due to system synergy. Others with different audio systems may have differing specific results than I had, but the overall view is still the same, a preference for the LDR volume control.
I've been using an LDR volume control off and on for 5-6 years now and here are some observations regarding them.
1.They are highly prone to temperature drift, a change of 5 degree of ambient temp caused some devices to drift 4-6%, yes you can use thermistors circuits to correct this,still does not account for non linear behavior.
2.They are inherently non linear. With a Johnson noise tester (Nut and Volt Mag. device) I quit trying to measure it as it was almost impossible to get the meter from moving erratically.
3.Could only get a maximum of -45db attenuation with my set of ldr's =no 2 a.m. listening
4.Seems to be no correlation if it works with amps with high or low input capacitance, sometimes good other times not so well. Chip amps (3886) good results, other amps hit or miss.
5.Dynamic range seems to increase, this is the area in which it is a real winner but not with every amplifier. Low listening level dynamics are fantastic. Look a Pass' article on the expansive and compressive characteristics of ldr's, and probably the highest measured indicator of why they sound they way they do.
6.Sound is fairly clear, sound stage on flat side and imaging good
Better yet built it your self and make your own opinion.
1.They are highly prone to temperature drift, a change of 5 degree of ambient temp caused some devices to drift 4-6%, yes you can use thermistors circuits to correct this,still does not account for non linear behavior.
2.They are inherently non linear. With a Johnson noise tester (Nut and Volt Mag. device) I quit trying to measure it as it was almost impossible to get the meter from moving erratically.
3.Could only get a maximum of -45db attenuation with my set of ldr's =no 2 a.m. listening
4.Seems to be no correlation if it works with amps with high or low input capacitance, sometimes good other times not so well. Chip amps (3886) good results, other amps hit or miss.
5.Dynamic range seems to increase, this is the area in which it is a real winner but not with every amplifier. Low listening level dynamics are fantastic. Look a Pass' article on the expansive and compressive characteristics of ldr's, and probably the highest measured indicator of why they sound they way they do.
6.Sound is fairly clear, sound stage on flat side and imaging good
Better yet built it your self and make your own opinion.
What's the connection between linearity and noise, or are you trying to say the noise is so high the distortion measurement is unstable?
No, What I'm saying is that I couldn't get it to lock to measure thermal noise, but it might correlate to making other distortion measurements at different temps. The Johnson noise is a side issue, Oh how I wish I had an Audio Precision measurement system. some day, some day
Also I'm no discouraging anyone to build or try it, Ive had it hooked up to a chimp with fantastic results, but I put it with a push pull amp with feedback and results where very disappointing.
No problem with that, but the repeated evidence posted that there are gain structure source/load impedance issues points to a quantifiable effect. If it comes down to some secret mojo that does not show up in any "conventional" measurements I'm not too interested, that just goes to the DBT challenge and here that goes nowhere.
Mr. Wurcer, you seem to be the most open minded here. Could you please kindly answer the questions on #183 http://www.diyaudio.com/forums/analog-line-level/284078-ldr-attenuator-impressions-19.html. Then if you could relate it to practical terms of actual use. Keeping in mind Mr. Pass's graphs are % Distortion vs. Attenuator Output Voltage and that modern power amps voltage gain range from x15 thru x30 and peak voltages at speakers per Pano's poll majority listed 5V or less.
I mean I know the answers but I'd rather it be from someone else with your background and reputation. I think LDRs in these particular optocouplers have been overlooked as high audible distortion without analyses of their actual practical use. Naturally every component can be overdriven to higher distortion...
Well that's a problem now is it not. Many on this forum who tout being scientific minded won't look beyond "conventional" measurements. If we were building a radio transmitter tower thermal noise would be a conventional measurement in the tuning circuits. However I'm not one to think everything can be explained or measured, in fact it's just plain arrogant.
I recently read that thru some kind of doppler analysis that the tympanic membrane can detect movement at .1 angstroms less than the width of hydrogen molecule, so no I don't believe everything can be measured.
With that said I can pick out every time my preamp, ldr, dcb1 buffer and attenuator in a blind test, just need an assistant like my wife who just loves to try to fool me.
This thread is ldr listening impressions, sorry if pure subjectivity was in order.
If you're going to hang your ideology on an example like that, may you please at least link the study? That's quite a claim for anything mechanical. Perhaps some kind of Michelson-Morely interferometer setup?
No one is saying you cannot enjoy your system in the subjective way that you will. I mean, yes, end of the day we all listen at home in a sighted fashion. There are the usual trite back and forths about objective measurements and subjective evaluations, but there's also some questions about what's going on/whether it's audible/understanding listener preferences.
And there's value in enjoying the system you have because of the investment you've made in it, whether or not it would hold up to blinded evaluation. It's often hard to separate the meat from the "secret mojo" effects, as Scott puts it.
Please reference Pano's thread http://www.diyaudio.com/forums/mult...much-voltage-power-do-your-speakers-need.html and view the Poll Results diyAudio - View Poll Results. Out of the 479 who tested a total of almost 71% answered "2V or Less" (185 members) and "Between 2V to 5V" (155 members).
Two years before the poll I posted this #28 http://www.diyaudio.com/forums/vend...optocoupler-volume-control-3.html#post2386926
Referencing Mr. Pass's LDR distortion vs. attenuator output voltage graphs here http://www.diyaudio.com/forums/anal...uator-new-passive-preamp-134.html#post1520243 and here http://www.diyaudio.com/forums/anal...uator-new-passive-preamp-134.html#post1520244, while keeping in mind that power amp voltage gains are 15x to 30x...
1. What will be the LDR attenuator output voltages at the power amp's input for peaks of 2 volts? Peaks of 5V?
2. Where do these output voltage ranges fall on Mr. Pass's graphs?
3. What will be the approximate distortion percentage? And in dB?
4. Will the values from #3 be audible?
Blues, I looked at Nelson's distortion graphs, am I interpreting it correctly:
If we go to a specific input voltage on the X-axis, say 1 or 2V, we see that the distortion goes up with attenuation. The curve stops at -10dB (att of 3 times or gain 1/3) which seems not in the normal use range of a 'preamp'. As we listen mostly in the 1W range, the necessary average attenuation with a 1V or 2V source would be much higher, maybe -20dB or -30dB.
Extrapolating Nelson's curves (not sure that is allowed) would indicate that in normal use the distortion from these things would be rather high, especially when compared to transparent means like a good stepped attenuator or a good pot.
I would expect that to be clearly audible.
Jan
These things ! ... why guess Jan, when its a simple matter of listening to one yourself, to reach conclusion that distortion is NOT clearly audible,
Look forward to your report having listened.
Cheers / Chris
Either way it's distortion compared to conventional resistor attenuation.
I have a feeling that some here think of "distortion" as the horrible noise that a broken or clipping amplifier/speaker makes.
But (non-linear) distortion is anything that adds or removes frequency components from the original waveform.
At the risk of repeating myself: Having done many blind comparisons I can say that distortion in audio reproduction generally degrades sound quality for me, so when someone has different preferences then so be it.
I care only as much as knowing what the person prefers.
I have a feeling that some here think of "distortion" as the horrible noise that a broken or clipping amplifier/speaker makes.
But (non-linear) distortion is anything that adds or removes frequency components from the original waveform.
At the risk of repeating myself: Having done many blind comparisons I can say that distortion in audio reproduction generally degrades sound quality for me, so when someone has different preferences then so be it.
I care only as much as knowing what the person prefers.
Last edited:
Mr. Wurcer, you seem to be the most open minded here. Could you please kindly answer the questions on #183 http://www.diyaudio.com/forums/analog-line-level/284078-ldr-attenuator-impressions-19.html.
Well thank you, if it's true that normal listening is in the -20 or so attenuation range I would go on to ask why design the attenuator to mimic a pot, where the performance is best at the wrong end? Unless I read the Silonix data sheets wrong it should be possible to do much better than the numbers given at those levels of attenuation. Just like sometimes I wonder why all those resistors to make a 23 position stepped attenuator when only 3 or 4 clicks end up being used.