Lightspeed Attenuator a new passive preamp

[gootee]

(DF96: )
<snipped>
As I said, I have offered as much evidence as those who like LDR attenuators. However, a symmetric non-linear resistor (e.g. LDR) will generate odd-order distortion if there is no DC bias, and if the non-linearity is smooth then the distortion will be dominated by third-order for most reasonable signal levels. The only part of that sentence which is a "guess" is the word 'symmetric'; as this is the best case you would be wise not to argue with it, because an asymmetric non-linear resistor would almost certainly generate more distortion. The remainder is just standard knowledge.
<snipped>

DF96,

As seems to often be the case, the question is probably, "How MUCH distortion?".

I gather that you meant that a curve in the resistor's V-I (voltage-vs-current) plot would cause the distortion you mentioned. Please correct me if that assumption is wrong.

Intuitively, it seems almost certain that it would cause some amount of some kind of distortion, if the V-I characteristic were not linear, around a fixed resistance setpoint. But, to be fair, if the distortion level was, for example, 0.00000000001%, then it would be utterly negligible.

Without first determining at least the order of magnitude's neighborhood, it almost seems wrong to even mention it in the presence of non-EEs. <grin>

Anyway, I don't know either. But I also wasn't aware that a CdS (Cadmium Sulfide) cell's V-I characteristic is non-linear, when the illumination is not varying.

Remember that we CAN'T consider the LED-LDR or R-vs-lux curves:

After the volume knob has been set for a fixed attenuation level, the LED's I-vs-V and the LDR's R-vs-I and the CdS's R-vs-lux response curves are totally out of the equation.

At that point, it's just a CdS cell with a fixed resistance setpoint, with non-varying illumination.

Are you saying that the CdS cell's V-I plot is curved, when the illumination is NOT varying? (Sorry, I guess I could look it up, myself...) Interesting.

Edit: See Figure 4 I on page 198 or so of

downloads.hindawi.com/journals/apec/1981/372454.pdf

Cheers,

Tom

[gootee]

Here's a diyaudio post/thread where they say the NSL-32SR2 can do -130 dB distortion:

Jim Williams designs a new Super Oscillator

It looks like it will do down to a hair over 0.003%, if the signal level is at -15 dBu.

[AndrewT]

How does -130dB for distortion from post37 come from the data shown in post35?
Jim Williams designs a new Super Oscillator

[DF96]

Figures I have seen for LDR attenuators are in the region of 0.01-0.1% distortion for signals around 1V. Small, but not necessarily entirely negligible. Perhaps many orders of magnitude larger than the distortion of a normal pot? Why should we drop something with almost immeasurably small distortion for something with much more distortion?

Google 'LDR attenuator distortion'. I understand some optoisolator manufacturers publish their distortion curves?

[georgehifi]

Typically the NS32SR2S will do .01-.04% with a 2v input and this is 2nd harmonic, and that's the nice sounding one. When SE amps are giving out percents and speakers even more this isn't worth the breath you guys are using to argue the point.

Cheers George

[mike_mcf]

Originally Posted by georgehifi

When SE amps are giving out percents and speakers even more this isn't worth the breath you guys are using to argue the point.

Cheers George

It always amazes me when people get all concerned about hundredths of a percent distortion in electronics, when speakers have 1,000 times more distortion (or more). There was a time when .1% was considered distortionless. Then is became .05%, then .001%. I won't even bother to comment on single ended tube equipment.

[gootee]

Originally Posted by DF96

Figures I have seen for LDR attenuators are in the region of 0.01-0.1% distortion for signals around 1V. Small, but not necessarily entirely negligible. Perhaps many orders of magnitude larger than the distortion of a normal pot? Why should we drop something with almost immeasurably small distortion for something with much more distortion?

Google 'LDR attenuator distortion'. I understand some optoisolator manufacturers publish their distortion curves?

The distortion plot was at the link I gave for the "Jim Williams..." thread. It shows about 0.003% as a minimum, for a -20 dBu signal level (I think it was).

Anyway, the reason that someone would "drop something with almost immeasurably small distortion for something with much more distortion" is because the distortion doesn't generate sufficient weighting in their optimization metric to prevent them from doing that. i.e. They want to.

If I recall correctly, one of George's concerns was mechanical contacts, and some of the sound-quality-threatening effects associated with them. And to be fair I can easily imagine an old dirty pot causing a lot of distortion, and worse.

But my main concern was that you might have been confusing the non-linearities associated with changing the resistance setpoint with a non-linear V-I characteristic after the setpoint was established, when none of the LDR-response non-linearities are still present.

I haven't seen any data that suggests that the V-I characteristic of a CdS photocell is non-linear, once it is at a particular resistance value.

It might be non-linear, and probably is, slightly, based on the distortion plot I saw. But I haven't found any good data on the non-linearity itself. And the one plot I found shows NO non-linearity.

Of course, everything is probably non-linear at some point, to some degree. Plain resistors are non-linear due to heating effects interacting with their temperature coefficient when the current level changes with an AC signal. That effect is said to actually be significant-enough for audio that some designers use over-sized resistors where it might matter. The same type of effect has been noted with in-line fuses for speakers.

But George is probably right that the distortion level is so low that it's not significant.

Cheers,

Tom

[georgehifi]

Originally Posted by gootee

If I recall correctly, one of George's concerns was mechanical contacts, and some of the sound-quality-threatening effects associated with them. And to be fair I can easily imagine an old dirty pot causing a lot of distortion, and worse.
Tom

This is the first article I found, I'm sure there are many others.
In this article near the bottom of the first page, the author talks about the diode effect (non-linear effect) between dissimilar contacts in plugs and sockets, which have a far better contact pressure and area than the very lightweight tiny contact area of an audio volume pot (potentiometer).

Passive Intermodulation, PIM Basics :: Radio-Electronics.Com

Cheers George

[soongsc]

Originally Posted by georgehifi

This is the first article I found, I'm sure there are many others.
In this article near the bottom of the first page, the author talks about the diode effect (non-linear effect) between dissimilar contacts in plugs and sockets, which have a far better contact pressure and area than the very lightweight tiny contact area of an audio volume pot (potentiometer).

Passive Intermodulation, PIM Basics :: Radio-Electronics.Com

Cheers George

Very interesting! Almost all gold plated connectors have 90% of nickel. The this is also true with vast gold plated PCBs that use the so called "chemical gold".

[DF96]

Originally Posted by georgehifi

Typically the NS32SR2S will do .01-.04% with a 2v input and this is 2nd harmonic, and that's the nice sounding one. When SE amps are giving out percents and speakers even more this isn't worth the breath you guys are using to argue the point.

If 0.04% can be ignored, why is that the much much smaller distortion from a pot is best avoided?

Originally Posted by gootee

But my main concern was that you might have been confusing the non-linearities associated with changing the resistance setpoint with a non-linear V-I characteristic after the setpoint was established, when none of the LDR-response non-linearities are still present.

I haven't seen any data that suggests that the V-I characteristic of a CdS photocell is non-linear, once it is at a particular resistance value.

No, I am not confusing setting nonlinearity with signal nonlinearity. CdS photocells are a little non-linear, that is why they generate distortion. Not much, but a heck of a lot more than a pot.

Originally Posted by georgehifi

This is the first article I found, I'm sure there are many others.

That is talking about the well-known 'rusty bolt effect'. Applies to dirty or loose contacts passing appreciable signal current levels in lowish impedance circuits, but even then the IM is usually quite small. Irrelevant to audio pots, unless very dirty and feeding a low impedance. Classic FUD technique: find an article which says something different and hope people won't spot the difference.