Tortuga Audio LDR3
- Distortion in Volume Controls
- Brief Overview of the Tortuga Audio LDR3
- Frequency Response
- Harmonic Distortion
- Harmonic Spectrum
- Intermodulation Distortion (IMD) - SMPTE
- Intermodulation Distortion (IMD) - DFD
- Multi-Tone Intermodulation Distortion
- Residual Noise & Hum
- Inter-Channel Crosstalk
Volume controls based on LDRs (Light Dependent Resistors) are often hailed as the holy grail in the DIY community. While LDRs make great light sensors, I must admit that I am rather skeptical of their use in audio applications, except for effects circuits, such as compressors and tremolo circuits. I suspect they owe their reputation in the DIY community to the fact that they are are not commonly used by commercial equipment manufacturers, thus somehow "special", "exotic", or simply different.
At the schematic level, a "passive preamp" is about the simplest circuit available. It is some form of variable attenuator, usually implemented as a potentiometer or switched resistive attenuator. Simple, right? For commonly used resistor types, such as potentiometers, metal film resistors, and select thin film resistors that certainly is the case. At least I have yet to be able to measure any distortion from an ALPS pot (RK271 "Blue Velvet" or RK097 series), leaded metal film resistors, or Susumu thin film SMD resistors. However, not all resistors are created equal. For example, some thick film SMD resistors cause measurable distortion due to the voltage coefficient of the resistors.
Some resistive materials change resistance slightly dependent on the voltage applied across the resistor. This change is known as the voltage coefficient of the resistor. In a volume control application, this means that the attenuation of the volume control changes slightly as function of the instantaneous voltage across the attenuator. This introduces a signal-dependant artifact, which can be measured as an increase in harmonic distortion. LDRs are commonly made from cadmium compounds, such as cadmium sulphide (CdS) or cadmium selenide (CdSe) and are optimized for light sensitivity – not for a low voltage coefficient. Some LDRs, such as the Silonex (now part of Luna Optoelectronics) NSL32SR3S used by Tortuga, are supposedly optimized for low THD, though the data sheet is rather minimal and their application note is incredibly vague on the topic of distortion with terminology such as "medium THD" and the like. To their credit, Silonex do show some data for the expected distortion performance of an attenuator built using their LDRs, which indicate that an LDR based attenuator should be able to achieve pretty decent performance, at least at low attenuation settings.
The Tortuga Audio LDR3 consists of several circuit boards. Two display boards indicate the volume setting on the left and right channel, respectively. An encoder board takes user input and routes it to a micro controller board, which controls the LDRs. The kit comes together pretty easily and is fairly intuitive to use.
The volume setting ranges from 00 to 70 (arbitrary units). In my sample, a setting of 60 corresponded to -5 dB gain; 46 to -20 dB gain; 26 to -40 dB gain; and 04 = -60 dB gain. The settings 00 (mute) and 70 (0 dB gain) cause the displays to blink, which I found rather distracting in use, though one could argue that the volume control is rarely used at its extreme settings. The encoder has detents and the volume control setting increments by two (00, 02, 04...) at each detent. The volume control setting will increment by one if set between detents, and this will cause an off-by-one error if the rotation of the encoder is changed between detents. This results in the selection of odd numbered volume control settings at the detents (01, 03, 05 .... 65, 67, 69), thereby making the extreme settings impossible to reach. This is clearly a bug in the micro controller firmware.
One nice feature of the Tortuga Audio LDR3 is that the micro controller board has an on-board switching regulator to provide the voltages needed for all the circuitry. Thus, the board requires only a single +12 V supply to operate. While the LDR3 is marketed as a "passive preamp", it should be noted that any noise on the power supply used to power the LEDs which control the LDRs will have an impact on the audio quality. This is because any perturbations on the LED power supply will modulate the light output from the LED, thereby the resistance of the LDR. Thus, while the audio signal does not go through active circuitry, an LDR volume control is as sensitive to the power supply as any active preamp – if not more sensitive. To eliminate any power supply related concerns for the measurements, I used an HP6237B laboratory supply to power the LDR3.
The frequency response of the Tortuga Audio LDR3 is shown below. This measurement shows the worst case gain variation of the LDR3, which occurs at the -5 dB gain setting. As seen below, the volume control does show some rolloff towards the high end of the audio band, which is rather surprising. I see two possible explanations for this: Either the LDRs are rather capacitive and form a lowpass filter with the 20 Ω output impedance of the APx525 audio analyzer source or the LDRs have high output impedance and form a lowpass filter with the APx525 audio analyzer input. The latter is the most likely scenario.
Interestingly, some gain settings show a slight bass boost in the gain flatness plot below. In general, the LDR3 shows considerably more gain variation versus frequency than a common volume potentiometer.
Harmonic Distortion & Noise
The THD+N vs output level is shown below. At 0 dB (max. volume) the LDR3 shows pretty decent performance. It's not stellar, but reminiscent of the performance of the simple transistor-based preamps of yesteryear. However, as soon as the volume control is turned down from its maximum setting, the distortion starts to skyrocket with distortion figures reaching 10 %. Granted, this is with 10 V RMS applied to the input of the volume control, but even at the common 2.0 V RMS line level, the distortion is still within 0.1 % to 1.0 % at all the tested gain settings except max. volume. This would barely pass the 1960ies era DIN 45500 HiFi standard.
The THD+N does appear to be relatively flat versus frequency as shown below (-5 dB gain, 2 V RMS).
Few buy a preamp only to crank the volume to the max, however. With the common 26-32 dB of gain in power amplifiers, the majority of listening takes place with preamp gains on the order of -50 dB to -30 dB. Thus, I performed a series of measurements of the THD+N vs volume setting at two different input voltages. The measurements were performed at 1 kHz with 20 kHz measurement bandwidth and are plotted below.
Harmonic Spectrum (1 kHz)
The harmonic distortion is predominantly odd-order as shown in the distortion spectra below for various gain settings. Note how the even-order harmonics start to increase as the gain is decreased. Also note the changes to the noise floor as the gain is changed. This is likely due to noise on the LED power supply as mentioned earlier.
Intermodulation distortion (IMD) - SMPTE (60 Hz + 7 kHz @ 4:1)
The intermodulation distortion of the LDR3 is mediocre at best as shown in the plots below. The IMD percentages are tabulated below.
Intermodulation distortion (IMD) - DFD (18 + 19 kHz @ 1:1)
The IMD percentages are tabulated below.
|Gain||SMPTE (60 Hz + 7 kHz @ 4:1)||DFD (18 kHz + 19 kHz @ 1:1)|
|0 dB||0.0353 %||0.0068 %|
|-5 dB||0.565 %||0.104 %|
|-20 dB||3.420 %||0.609 %|
|-40 dB||4.780 %||0.873 %|
The multi-tone IMD measurements uses 32 tones which are logarithmically spaced in frequency. This signal provides a way to measure a deterministic signal which is very close to an actual music signal. It sounds a bit like an out-of-tune pipe organ. The peak voltage of this signal is 2.82 V, corresponding to a 2 V RMS line level.
Note all the "grass" between the 32 tones. Those are intermodulation distortion products. A good preamp design will show few and low level IMD products.
Residual Noise & Hum
The noise floor, especially at 0 dB (best case), is very low. Sadly, it degrades significantly as soon as the volume control is turned away from its maximum position. Some degradation is common, and -6 dB gain is the worst case setting for a variable resistor volume control, as it results in the highest output impedance of the volume control. Still the 15.5 µV RMS noise measured at the output of the LDR3 seems excessive for a "passive preamp". That is the equivalent noise of a 1.5 MΩ volume pot.
Also note how various mains-related hum components and their IMD products creep up at the -5 dB gain setting.
The integrated hum + noise voltages (20 Hz - 20 kHz bandwidth) are tabulated below.
|0 dB||1.20 µV||0.83 µV|
|-5 dB||15.5 µV||5.70 µV|
|-20 dB||6.50 µV||2.72 µV|
If there is a bright spot for the LDR3, it would be the inter-channel crosstalk, which is very low. The crosstalk does increase (i.e. the channel separation degrades) at higher frequencies, likely because the two channels share the same ground. That said, the channel separation is quite good.
I considered not publishing these results as I generally avoid speaking negatively about my competitors. However, I have received quite a few questions about LDR based volume controls and rather than having to repeat myself, I would rather point people to this page so they can draw their own conclusions from the data.
I approached the measurements of the Tortuga Audio LDR3 with an open mind. I did my absolute best to make the LDR3 shine and I was more than willing to have my skepticism of LDR based volume controls proven wrong. Fortunately measurements tell the truth and the measurements of the LDR3 are not impressive.