LDR photoresistor attenuator preamp, in general

[gootee]

Originally Posted by romana

You're right about that. Apparently, it was developed by Jim Strickland (formerly of Acoustat), after David Hafler sold the company.

John Atkinson's review in the June, 1989, Stereophile discusses the Iris LDR attenuator in some detail.

His description of it as based on "a Cadmium Sulfide (CdS) photosensitive resistor, as used in countless photographic lightmeters," suggests that Ed, of "The Truth" preamp, may be playing a semantic game with his technology (though it's not worth getting too twisted about it).

Not quite sure how Ed meant that, but I thought that most or all LDR (Light-Dependent Resistor?) devices were made by encapsulating a photocell (typically a CdS [Cadmium Sulfide] photocell) and an LED together in a light-tight enclosure. The Silonex devices used in the Lightspeed work that way, as do the Perkin-Elmer VTL-series "Vactrol" LDRs. Or am I missing the point?

[gootee]

Originally Posted by romana

Granted, this thread revolves around George's LS LDR circuit, but here are some Big Picture Questions:

1. Is the LDR technology inherently superior in and of itself, such that sonically, the associated circuit is not really critical to the end result or do different LDR circuit designs clearly sound different from each other?

2. A related question is, will different brands of caps or resistors in the same LDR circuit change the sound, as happens in most audio designs?

3. Has anyone here built or used two or more of the DIY or drop-in circuits and compared them from a sonic point of view?

4. Or has anyone built-up two or more of the same boards, using different passive components, then compared the sound? Have any of the DIY developers done this or have the component choices been based on availability, price, or fit, rather than direct comparison? Let's assume that practical measures, like staying with the Silonex LDRs, are a given, regardless of the design.

In part, I posted the summary lists in order to sort out the different approaches. If one can end up with better sonics through an alternate approach, then starting a new thread with this in mind, may make sense. However, if the end result is just a more complex route to the same destination, then it's only an exercise with no practical benefit, other than experimenting for the sake of experimenting.

I know this sounds somewhat obsessive-compulsive, but little things often make a difference.

Ciao for now,
~R

One concern that I have with the Lightspeed-type LDR-based attenuators is that the input and output impedances change significantly, depending on the attenuation level. When combined with any capacitance or inductance present in the source or load or interconnects, this could make the overall frequency response depend on the attenuation-level setting. Variations in input impedance could also change how the source responds while driving it, while variations in output impedance could change how the load responds.

I am not sure whether or not the magnitude of any resulting effects is significant-enough to worry about.

Back when I was thinking about it, I developed circuit modifications to try to "linearize", or, actually, make more-constant, the impedances, which I posted in the Lightspeed thread. But no one seemed too interested, at the time.

Below is a link to my post # 1095, in that thread, showing how the input impedance could be kept nearly constant. But it also includes a simulation plot of the variation of the input impedance of the original Lightspeed circuit.

Lightspeed Attenuator a new passive preamp

Similarly, the Lightspeed's simulated output impedance varies from about 37 Ohms to about 14.6 kOhms, as the attenuation level is varied from maximum to minimum.

With a 50kOhm load, the input impedance varies from about 14.9k to 9.9k and back up to about 11.3k, as the attenuation level is changed from min to max. With a 100kOhm load, the input impedance goes from 14.9k to 10.3k to 12.8k. With a 22kOhm load, the input impedance goes from 14.9k to 8.7k minimum then back up to 8.8k.

Someone might also be interested in my posts #1071, 1064, 1056, and 1045, in that thread, and some of the others. My LTspice simulation files download link is in post #1045, and also below. Please note that a 50 Ohm resistance should probably be added in series with the current source. The Ltspice circuit simulator, itself, is free to download, from linear.com.

Lightspeed attenuator LTspice simulation files:

http://www.fullnet.com/~tomg/OptoAtten.zip

I also have a spice model for the Perkin Elmer model VTL5C2 Vactrol LDR, at:

http://www.fullnet.com/~tomg/gooteesp.htm

Cheers,

Tom

[tinitus]

Thread title changed on request
Can we move on now please

[barmycodger]

Originally Posted by gootee

Lightspeed's simulated output impedance varies from about 37 Ohms to about 14.6 kOhms, as the attenuation level is varied from maximum to minimum.

With the Lightspeed I made I haven't noticed variations in frequency response at minimum and maximum settings because I don't listen to music at those extremes.

[anubisgrau]

it's an interesting question that's overlooked when talking about LDR attenuators - output impedance in non-bufferred models. normally, without a prior test, i would have a problem to get assembled or bought something that states 7-10k impedance. i was struggling so much when driving 3k source via transformer volume control. however i don't think buffers are viable solution because they MUST have an effect on some LDR qualities, in particular transparency and overall sound character (i yet have to hear one that doesn't dominate).

anyone know any non-buffered models, DIY-ed or non-DIY-ed that guarantees lower output impedance. OK, it doesn't have to be 100ohm, but at least under a 1k. despite what some people write or say, i found preamp/amp ratio needs to be at least 100x to get a good drive (in my room at least).

[Bas Horneman]

it's an interesting question that's overlooked when talking about LDR attenuators

I don't think it has been overlooked. Output inmpedance is always something that needs to be taken into consideration.

[anubisgrau]

Originally Posted by Bas Horneman

I don't think it has been overlooked. Output inmpedance is always something that needs to be taken into consideration.

as far as i could see, it's always buffers which is offerred as an instant solution. has anyone come with any solution for LDR attenuation with lower impedance - i can't recall seeing anyting like that. i may check with EVA manufacturer for their data, LS and similar are i think around 7k.

[AndrewT]

The manufacturer of the Lightspeed addresses exactly this topic on our Forum.

It's a while since I read his viewpoint so I'll paraphrase:
A buffered Lightspeed does not sound as good as a Lightspeed, even though it has a higher output impedance that is usual for a buffered source.

A Lightspeed sounds better than an unbuffered pot which often has an output impedance many times the Lightspeed's.

eg. a 100k pot has an output impedance about four times the Lightspeed's output impedance.

[anubisgrau]

Originally Posted by AndrewT

The manufacturer of the Lightspeed addresses exactly this topic on our Forum.

It's a while since I read his viewpoint so I'll paraphrase:
A buffered Lightspeed does not sound as good as a Lightspeed, even though it has a higher output impedance that is usual for a buffered source.

A Lightspeed sounds better than an unbuffered pot which often has an output impedance many times the Lightspeed's.

eg. a 100k pot has an output impedance about four times the Lightspeed's output impedance.

well i'm not going to argue with writing of any manufacturer but it needs quite lots of confidence to claim 28k of any pot out there (7k output imp of the Lightspeed impedance times four). i don't think this is true.

OTOH, i agree on remarks that buffers are not perfect. but it's a really great question, to go commando with 7k, or to go bufferred with a low low imp (say 5ohm). before i try i can't tell.

[tomchr]

My concerns with LDR-based volume controls:

Linearity (--> distortion).
Tracking between channels.
Modulation by the light source.
Noise.

And my concern with passive preamps in general: Output impedance.

The LDR-based solution, while I guess you can call it a passive preamp, still needs a power supply to drive the light source. That really takes away the only advantage I can see of a passive preamp.

In my opinion, a quality pot preceded and followed by a quality buffer is a better solution from a performance stand point. If you want remote control, use a motorized pot or a volume control IC, such as the PGA2320. Though, the noise floor of the PGA2320 is about 10 dB higher than that of a quality buffer (I prefer LME49710).

~Tom