Lightspeed Attenuator

[24bit]

I posted it on diy forum as well - as diversification could help,
Hi,
A friend of mine built an attenuator based on LDR design.
I noticed that L+R channels are imbalanced.
when the pot knob is turned up, it starts from R channel +5dBFS louder then L channel (minimum),
but when knob turning up, the imbalance gets smaller and smaller, until around -20dbFS (around middle) it's almost perfect, then Left channel (when turning up) became louder and louder - up to -5dBFs, and it's almost perfect when in full 'open' position.
I was told it is just a problem with LDR and it possibly can't be resolved, but I remember reading some audio purist using LDR att. and I don't think they wouldn't notice a 5dB difference between channels, so I presume it's possibly to have it ideal (0.1dB - I can live with, 0.2dB - would be maximum)
Could you please help me - what could cause this sort of problem ?
How to resolve it ?
Is it simply swapping a pair of LDRs or sth else ?
I really appreciate your help, here - thank you in advance
If LDR design is not perfect and it's very difficult to find pair of LDR with 0.1dB difference - could you please advise me, what type of passive att. would be good ? - in a 'human - 99%' price range

[Puffin]

There are two ways to solve this. I am at work at the moment, but when I get home I will post a picture of where to add some variable resistors to balance the channels.

I have made several of these and in the one I currently use I have two additional pots that act in a way similar to a balance control. I made a diagram of how I wired it up (but will need to find it) or I will have to open up the pre-amp.

IMO the L/Speed is the best pre-amp I have ever heard. Mated with a Pass Labs B1 buffer it is very special for very silly money.

EDIT : here is the diagram. The variable resistors are the square blocks with a cross in (5k cermet trimmers)



Rob.

[Paul Hynes]

24bit,

I use a voltage controlled current source (VCCS) system to drive the LDR LEDs using George Stantscheff’s “Lightspeed” volume control configuration. This VCCS has a balance control and can also be set up with infra-red transmitter and receiver to give remote control of volume and balance. I have not measured the accuracy of balance but in practice neither I nor any other users have found any issues with balancing the signals when listening to music. I have used it with unmatched LDRs and the balance control is effective at levelling the two channels.

Regards
Paul

[24bit]

Thank you very much!,
I really appreciated your input,
waiting for more details from you - I'll have to pass it to constructor,
glad there's a way out...


here is a primitive video, sorry my camera is not a camera
I turn the volume knob up, and you can see first Right channel going up, then both channels ballance around the middle (-20dBFS) and then Left channels became louder up to 5dB and then level when the knob fully open

oops seems like the link doesn't work,
http://youtu.be/oEFS1cPpt6s

[24bit]

24bit,

I use a voltage controlled current source (VCCS) system to drive the LDR LEDs using George Stantscheff’s “Lightspeed” volume control configuration. This VCCS has a balance control and can also be set up with infra-red transmitter and receiver to give remote control of volume and balance. I have not measured the accuracy of balance but in practice neither I nor any other users have found any issues with balancing the signals when listening to music. I have used it with unmatched LDRs and the balance control is effective at levelling the two channels.

Regards
Paul

brilliant!
could you please give more details ?
I will pass it to constructor
thank you!

[24bit]

oh, and I forget to add
I fed foobar with 1kHz lopped file,

on this video - link works and pasted video not...sorry

[Puffin]

Filip you have a new PM.

[Reid Malenfant]

There are two ways to solve this. I am at work at the moment, but when I get home I will post a picture of where to add some variable resistors to balance the channels.

I have made several of these and in the one I currently use I have two additional pots that act in a way similar to a balance control. I made a diagram of how I wired it up (but will need to find it) or I will have to open up the pre-amp.

IMO the L/Speed is the best pre-amp I have ever heard. Mated with a Pass Labs B1 buffer it is very special for very silly money.

EDIT : here is the diagram. The variable resistors are the square blocks with a cross in (5k cermet trimmers)



Rob.

Hi Rob, apologies if I appear to be pi**ing on your fire here

This could be a good system, but there are more than a few things that can f*ck it up imo. For starters you need the variable resistor to track both channels accurately. Which is the equivalent of what you get in a normal stereo (or hifi) system.

Then you need the LEDs to give out the same brightness for the same current input....

Then you need the LDRs to be matched as well so they give the same resistance with 'X' amount of light input... Both input to output need to be matched & output to ground, I'd suggest that is impossible

On top of that you have the problem that LEDs don't all emit on the same wavelength even if they are from the same batch & LDRs respond in a different way to different wavelengths of light

To me that looks like at least four things that will make channel balance a pain in the ass. You might be able to balance it at one volume level, but turn that volume control pot & I can see it all falling to bits again balance wise

I'd suggest using an ADC with a single potentiometer & like the old ladder DACs setting up each increment to balance..

They certainly wont in that system as there are miles too many variables. In a potentiometer there is only that to blame, but the pictured system includes three more variables which will degrade channel balance in an unpredictable way.

Imho

[Puffin]

Mark. I understand all that you say. However I built my first L/S using unmatched LDRs. In fact it is the one I am still using 2.5 years on (roughly). I have not suffered any real problems with it.

Rob.

[Reid Malenfant]

Cool, there is of course the possibility that all the random differences between components can actually even things out. It has been known & now I think about it there may be some kind of explanation for it

I don't doubt you Rob, but in theory there is always the chance that these things can stack up against you.

I hate statistics I need to look into this, there may be something in it which can come out at some average as well


Hmmm, interesting