I would first try inserting a series resistor of 5K on each of the Lightspeed inputs, before the series LED/LDR. Or play with other values larger and smaller.
Many commercial "off the shelf" sources have a source impedance in the 1K to 2K range, and yours appear to be about 100 ohms. This should help with the control issue.
This is my suggestion only, but, yes, I have built and fully tested the Lightspeed and other LDR designs. 3 posts on raw LDR measurements and comments are here
http://www.diyaudio.com/forums/anal...-resistor-current-control-24.html#post2983854
With a differential input like this, I don't think you can use the normal Lightspeed configuration. How do you have the LDRs connected in relation with this differential input?
This is just the way it turned out. I agree that ones system would have to match the Lightspeed, no doubt, but I wasnt reporting on testing systems and The guy asking the question seemed to think there was on odd frequency response attributed to the Lightspeed. I dont think thats the case with his reported output and input impedances, do you?
I solved the mistery with odd freq change. By mistake L and R input where shorted somehow on the pcb and what i was hearing was a mono sound. Sorry 
If would put a logaritmic pot i would have a logaritmic variation of sound ? Now i have one linear and i hate linear pots but i dont have a 100k logaritmic one.
If would put a logaritmic pot i would have a logaritmic variation of sound ? Now i have one linear and i hate linear pots but i dont have a 100k logaritmic one.
What dac output stage is this? it looks to need another buffer after it? Is it a commercial offering ?
Cheers George
Last edited:
I was wondering the same thing. The LT1028 does not drive much load to my understanding. This might be a problem.
In the current Lightspeed configuration, you will need a log pot to make the gain variation more linear. High input impedance does shift the gain curve up at the lower volume positions, thus the log pot helps makes the gain control smoother.I solved the mistery with odd freq change. By mistake L and R input where shorted somehow on the pcb and what i was hearing was a mono sound. Sorry
If would put a logaritmic pot i would have a logaritmic variation of sound ? Now i have one linear and i hate linear pots but i dont have a 100k logaritmic one.
I solved the mistery with odd freq change. By mistake L and R input where shorted somehow on the pcb and what i was hearing was a mono sound. Sorry
If would put a logaritmic pot i would have a logaritmic variation of sound ? Now i have one linear and i hate linear pots but i dont have a 100k logaritmic one.
A log pot 100k will give better low level sweep than a lin.
With match quad set of ldrs you will get more low level control in the first half of rotation. With matched pairs anything can happen depends which is in series and which is in shunt.
Cheers George
Just looked at the specs it can only typicaly give 7mA drive max, this needs another buffer after it I think, even without the Lightspeed attached to it. Output opamp buffers are usually over 50mA drive or more output current, correct me if I'm wrong but this to me is just the I/V stage and it needs another buffer opamp after it.
Cheers George
My Dac is not a commercial one it's a DIY kit based on PCM1794 wich had a very bad implementation of the analog (output) stage and i reworked it using the datasheet reccomendation of Nat semi. For the buffer i have used LME49720 not the LT like in the picture.
Before the buffer i have 4xLm49713 for I/U convesion as PCM1794 is a full diferential dac.
Before the buffer i have 4xLm49713 for I/U convesion as PCM1794 is a full diferential dac.
you can use a linear dual track vol pot and add a log law faking resistor.
The usual value for the resistor is 10% to 20% of the pot value.
If your next stage has a low Rin then part of that faking resistor is already in place.
eg. 100k lin vol pot feeding a 50k input impedance.
The 15% value of 100k is 15k.
An additional parallel resistor of 22k gives ~15k of load for the vol pot to feed.
The first ¾ of the vol pot rotation is roughly the output shape one requires of an audio vol pot. The last ¼ rotation (the loudest part) deviates markedly from audio taper.
There is a big advantage to using the faking resistor + Lin. The matching of attenuation between channels can be much tighter than using an audio dual track taper.
The LDR total resistance is around 8K, At low volume, using a faking resistor does not create a problem, but the input impedance of the volume control will reduce as the volume is increased.
