• Mark

Are all LDR attenuators the same? The long read

If you look at the LDR attenuators' proof of concept, it seems that build one is straightforward, and, since it is simple, they perform all the same.

At the end of the day, what do you need? Two LDRs per channel, one load, the other shunt to mimic a potentiometer and.. presto!

And, if you really want to be 'exotic' and deliver a quality product, then a calibration system is required, some cheap MCUs, few lines of software and you are in business ready to compete with the very best!


Nothing farther from the truth!

Time ago, I came across a fascinating post from Tom Christiansen. Tom tested the Tortuga Audio LDR3 attenuator and the results fo his tests were very much disappointing, you can read about the test here. The test seems very well made, thought a bit too harsh toward the poor Tortuga (but for that price tag a bit of 'harshness' is required), what the test enlighten is that it's easy to screw up a simple concept.

I'm personally engaged in building an LDR attenuator, after the Lightstream Mk I, Lightstream Mk II, and now evolved into the 'Loch Eite Mk I' that's at the stage of pre-production, I can call myself an expert in trials and errors with LDRs.

Now, if the reader want to read the test results for the 'Loch Eite Mk I' after the horror of the Tortuga, then, please, click on this link.


Now, what makes build a suitable LDR attenuator so tricky? All in all, are just two optocouplers connected together! I'll try to give simple answers to a simple question, but the matter is not so simple!

First: calibration. LDRs are not linear in response to the voltage you supply to them, and this is NOT the worse. LDRs parameters wildly vary from one to another, just have a look at this screen from our 'Loch Eite' calibration process (ok, this picture is from the older LightStream Mk II; however, the concept is the same):

Do you see the PWM value for R2 left and right channel? In human language that means that the shunt LDR for the left channel needs 0.027 Volts for reach a resistance of 4415 Ω while the shunt LDR for the right channel needs 0.038 Volts for achieving almost the same value as LDR at the left channel. Nice.

Second: MCU resolution. Guys, 8 bits resolution is far too little, 14 or, better, 16 bits is the way to go. I know that manage a 16 bits PWM is much more complicated than 8, but, if you want linear response between channels, you need to put a more expensive MCU and work few hundreds of more hours on the software. Especially taking into consideration the problem laid out in the previous point.

Third: Smooth that PWM! Don't trust the latency data from the producer! Do you know what happens when you do a lousy job in controlling the LDR? This:

And this:

And this:

All the previous pictures (and the lousy video too, I'm sorry for that) are from a 1 kHz sine wave shot into LDR attenuators (our prototypes) that where not perfectly tuned-up. You need and high resolution oscilloscope to see that, at the end of the day the waveform must look like this:

Fourth: Keep the damn noise low! Is it passive right? Then no problem with noise! Wrong!

LDR Attenuators are, under a point of view, an active device. This means that the signal ground must connect with the driver's ground at some point. The driver is, most of the times, an MCU unit with another bunch of active electronics connected to it. In the case of the 'Loch Eite Mk I' we have an MCU, touchscreen, clock, relays and LDR drivers, voltage regulator and power supply. All these devices generate noise, and the unit must be designed to limit the noise at the minimum. If you do a lousy job thinking that noise won't be a problem, then, think twice! Star and bus grounding is a must in LDR attenuators together with a solid ground line!

And some screening will help too:

Loch Eite Mk I - Pre-production unit - internal view


Now, if you want to buy an LDR attenuator bear in mind that, despite the conceptual simplicity, the device is not simple, we all buy the LDRs from the same source, the Silonex Nsl-32sr2 or Nsl-32sr3, they are almost identical, but is the implementation that makes the difference between a lousy sounding LDR, and average sounding LDR and a great sounding LDR!

Got questions? Contact me.

234 views0 comments

Recent Posts

See All