Lightspeed Attenuator a new passive preamp

To add to this "powering the led's" mix, try a pure DC source such as a 12vdc Li-Ion (Lithium) rechargeable battery. See if you can reliably pick the difference between it and a 12vdc (linear) wall wart. Nobody I know can.
Even Sam Tellig from Stereophile says with his Lightspeed he has to look to see what's powering it at the time, because he switches between the two a bit and cannot tell the difference between the wall wart or battery power by just listening.

Cheers George
 
wapo54001,

Your argument is logical and well thought out. However, build the attenuator using the simple circuit. Then solder 150uf OSCON SEP caps right on the power supply leads of each optocoupler. Listen again. I guarantee you will leave the caps in place.

To your argument however, perhaps it will sound even better if the Oscons are replaced with very low leakage PPE smt film caps. You could never get 150u. But a few 1uf smt caps might be interesting.

I won't question your assertion because I don't have knowledge of the circuit as you have built it (nor the extremely critical power supply), but I would say that if such capacitors do have a beneficial effect you have a power supply issue and the capacitors belong on the power supply side of the volume control pot.
 
To add to this "powering the led's" mix, try a pure DC source such as a 12vdc Li-Ion (Lithium) rechargeable battery. See if you can reliably pick the difference between it and a 12vdc (linear) wall wart. Nobody I know can.
Even Sam Tellig from Stereophile says with his Lightspeed he has to look to see what's powering it at the time, because he switches between the two a bit and cannot tell the difference between the wall wart or battery power by just listening.

Cheers George

My comment about minimizing the current leaks in the LED path particularly apply to a situation where one is trying to hit a specific target current which is the case with my precision LDR project. In my case, I'm striving for an absolute current value flowing through the LED at a given volume control rotation, not a relative current as is the case with the Lightspeed. But I stand by my statement that eliminating extraneous current paths from the circuit after the control device -- the pot in the case of the Lightspeed, the active control circuitry in a PIC-based system -- would be advantageous in any circumstance.
 
To add to this "powering the led's" mix, try a pure DC source such as a 12vdc Li-Ion (Lithium) rechargeable battery. See if you can reliably pick the difference between it and a 12vdc (linear) wall wart. Nobody I know can.
Even Sam Tellig from Stereophile says with his Lightspeed he has to look to see what's powering it at the time, because he switches between the two a bit and cannot tell the difference between the wall wart or battery power by just listening.

Cheers George

That doesn't surprise me.

I simultaneously watch both the control input (current) and the resulting LDR resistance through my computer terminal, with refresh happening several times a second. The control current does vary plus or minus a bit, but the output resistance stays quite steady but not perfect -- when I'm looking at a 50 ohm output, the resistance is steady +/- one one-hundredth of an ohm. At 3000 ohms, it's within about +/- 75 ohms. I believe that the Lightspeed's LDR stability would be similar.

If you think of the resistance of a pot and it's effect on gain, a hundred ohms at 3K ohms of a 10K pot isn't going to be audible -- it's less than .2dB. I have wondered off and on whether the continuous slight variation in resistance characteristic of LDRs might not be the basis for mysteriously exciting quality of the sound it produces . . .
 
I have wondered off and on whether the continuous slight variation in resistance characteristic of LDRs might not be the basis for mysteriously exciting quality of the sound it produces . . .

I believe it's a combination of two things:

1: "Diode Effect" Having no touch contacts of disimilar metals in the signal path. Ever done a cleanup with contact cleaner of your rca's and sockets, "hear the difference" or hard wire (solder) eveything and hear an even bigger difference.


2: It could be that cadmium sulphide (CdS) sounds better than metal film, carbon or conductive plastic.

Cheers George
 
I won't question your assertion because I don't have knowledge of the circuit as you have built it (nor the extremely critical power supply), but I would say that if such capacitors do have a beneficial effect you have a power supply issue and the capacitors belong on the power supply side of the volume control pot.

I realize it is counter intuitive. After all the LED should isolate the PS. But it does not. It is great just as George builds it. But is can be improved and the difference is not subtle.
The lightspeed circuit I use is more or less per the diagram posted back in spring 2008 when a bunch of people experimented with a variety of caps on the pins of the optocoupler.
I have played around with various power supplies. Walwort, battery and linear. I believe a low noise diy voltage regulator and the caps on the pins are the most important elements. I still prefer linear power supply in front, but with the caps and the low noise reg, not sure if the supply is all that important. For the record I use a 12v toroid, bridge rectifier, CLCLC filter. The last couple of filter caps are Black Gate F. Toroid and first filter element are in a separate chassis.
Sound? Sublime!
 

Attachments

  • cct diagram with caps on leds.jpg
    cct diagram with caps on leds.jpg
    30 KB · Views: 622
Adding caps and getting better sound does not indicate a problem with the power supply. The caps ARE part of the power supply, as is everything before the LDRs. He now has a better power supply. Caps directly on the pins of an opamp and chipamps are well known ways to get better sound and it works for LDRs as well. Recently experimented with film vs ceramic .1uf's. No difference. I use this parallel with a 10uf-100uf electrolytic. I doubt brand has anything to do with the sound here but I use Nichicon FW, KW and those green bipolars depending on whats in stock. I'm sure any old electrolytic would sound the same but I think people expect a higher quality cap in their kits. I would NOT encourage higher values of caps after the potentiometer as you can hurt the wiper by forcing it to pass higher currents. If you have that 100R current limiting resistor in place you are much more safe using higher values of caps but I dont think you will benefit from higher than 100uf.
The NSL32SR2 used to have larger legs. Maybe 4-5 years ago. Then they started getting thinner. Now they are a little thicker, not as thick as before and have less spring to them. Softer. I am betting that the Sorted are older stock. Nothing wrong with that, just saying they probably sell far fewer of them.
If you have low efficiency speakers I would definitely follow AndrewT's advice. 40 Ohms is pushing them and even a few ohms lower is really a bad thing as you need multiples of the current to get much lower. 50 Ohms as a minimum would be an wonderful break for the LDRs and I run series LDRs in the 250R range. My speakers are 96db efficient and my amp is a gainclone so I really need that 40R low resistance on the shunts. I am willing to replace with a new pair if I have a problem. Been running the same LDRs for 3 years now. If they were to break tomorrow and I was a customer rather than a guy who matches LDRs I might be a bit upset, especially when the datasheets indicate no problem running LDRs at this current. I use LM334's to run my LDRs and they are only capable of 10mA and I still get into the 30's when adjusting a new pair of LDRs in someone's new attenuator.
 
By the way I just reread the note above about Oscons. I built a pre using LME49710. Awesome opamp, great sound, bass was so so good. Anyway, I built it and used tantalum caps. Very rock n roll sound. Good stuff but then removed and replaced with Silmic II. After burn in it was just a great sounding pre. Nothing remarkable but really really good nonetheless. Removed Silmics and replaced with Oscons. Now THAT was classy. Extremely smooth and wonderful. Just a few steps above either other cap I used and I liked this a lot. Very expensive caps at 150uf and 20V though. Its off topic I know and I would not use this comparison as a way to determine which caps to use in an LDR circuit. Just something to bounce off of wlowes comment.
 
You find a tighter tolerance for the NSL32SR2S (selected), this is why it's a bit more expensive but worth it, it gives typical 40ohm min to 5mohm max.
Where the NSL32SR2 they won't give the typical low figure it can be anywhere with a max to 5mohm.
And the NSL32SR3 is even worse at typical 150ohm min to 25mohm max.

Cheers George

I was talking about signal amplitude into the LDR attenuators. For example, has any tried feeding a 0.5V signal vs a 10V signal and see if the gain is the same. I know with most resistors it is the same. Just have not tested with LDRs.
 
I believe it's a combination of two things:

1: "Diode Effect" Having no touch contacts of disimilar metals in the signal path. Ever done a cleanup with contact cleaner of your rca's and sockets, "hear the difference" or hard wire (solder) eveything and hear an even bigger difference.


2: It could be that cadmium sulphide (CdS) sounds better than metal film, carbon or conductive plastic.

Cheers George
True, these can all make a difference. Higher CdS density might be the cause of it sounding better than other materiel.
 
I was talking about signal amplitude into the LDR attenuators. For example, has any tried feeding a 0.5V signal vs a 10V signal and see if the gain is the same. ..................
10Vac is a high signal level.
What is the dissipation in the attenuator, at various volume settings?

Total dissipation and individual dissipations must be low enough not to affect the short term capability and for reliability, the long term dissipation limits may be quite a bit lower.
 
I've avoided to push the LEDs really hard. When you push them hard, they become really slow and will die soon. The lowest volume I can still change is low enough, even for my wife (and it's an excuse if she wants inaudible level). And when I lower volume further, I have an automatic mute relay kicking in, bringing volume to zero.
I tought of putting in parallel a relais with an extra resistor to reduce by 12 or more dB. In fact it would become a 2-range volume (low volume range with resistor, and high volume without), with a processor behind, everything is possible. But I haven't implemented it as the range is OK now. I couldn't live without the mute though.
 
I dont recall looking for different 'gain' but the distortion is way way different. 200mV least distortion of all in a series/shunt LDR arrangement. 2.5V was as high as I would ever go. 10V distortion was at very high levels but no I dont remember the numbers. I would bet it was in the 8-10% range. Sorry I didnt write down numbers to remember. I just looked at the V vs % and made up some rules of thumb for myself.
 
I've avoided to push the LEDs really hard. When you push them hard, they become really slow and will die soon. The lowest volume I can still change is low enough, even for my wife (and it's an excuse if she wants inaudible level). And when I lower volume further, I have an automatic mute relay kicking in, bringing volume to zero.
I tought of putting in parallel a relais with an extra resistor to reduce by 12 or more dB. In fact it would become a 2-range volume (low volume range with resistor, and high volume without), with a processor behind, everything is possible. But I haven't implemented it as the range is OK now. I couldn't live without the mute though.

I have now tested with three LDRs on the shunt side with LDR devices that meet my 50 ohms at 10ma or less criteria (3/4 of the LDRs I get from udaily meet this) I'm getting 17 ohms or less with no loss in control in a circuit that draws less than 10ma total (LDR LEDs in series) so the LDR devices cannot be affected even if you keep the 10K pot set at -56dB continuously. Actually, the control down there is so good that I can control to within 1/100th of an ohm.

I think in future for people wanting substantial attentuation (54~60+ db) paralleled devices will be the way to go. A device costs maybe $3, so a multi-LDR path is reasonable, and it doesn't complicate the circuit unreasonably. The series side requires only one LDR for any configuration.

In my circuit, increasing to eight LDRs for stereo will require value changes for one resistor and two LEDs, otherwise everything remains the same.
 
I think in future for people wanting substantial attentuation (54~60+ db) paralleled devices will be the way to go. QUOTE]

I have tried 2 and 3 x paralleled ldr's in the past, and to my ear 1 x ldr sounded better, two seemed to soften/smear the sound a touch, not by much but it was detectable. More so than the difference between pure battery power vs linear wall wart, which is hard to pick the difference.

KISS

Cheers George
 
I dont recall looking for different 'gain' but the distortion is way way different. 200mV least distortion of all in a series/shunt LDR arrangement. 2.5V was as high as I would ever go. 10V distortion was at very high levels but no I dont remember the numbers. I would bet it was in the 8-10% range. Sorry I didnt write down numbers to remember. I just looked at the V vs % and made up some rules of thumb for myself.

I think this is good indication that gain testing would prove to be interesting. If the trend indicates higher gain at higher voltages, then the LDRs actually compensate BL curves of speaker drivers. In this case, the LDR should be showing higher distortion at higher input levels as well because it is compensating nonlinear driver with nonlinear characteristics.
 
I think in future for people wanting substantial attentuation (54~60+ db) paralleled devices will be the way to go. QUOTE]

I have tried 2 and 3 x paralleled ldr's in the past, and to my ear 1 x ldr sounded better, two seemed to soften/smear the sound a touch, not by much but it was detectable. More so than the difference between pure battery power vs linear wall wart, which is hard to pick the difference.

KISS

Cheers George

I wondered about that (and agree that KISS is good) and felt it would not be a problem in my system because the paralleled devices are only in the shunt resistor. Overall, my system is quite complex, but the audio signal path is no different from the Lightspeed except for the paralleled LDRs on the shunt side.
 
I wondered about that (and agree that KISS is good) and felt it would not be a problem in my system because the paralleled devices are only in the shunt resistor. Overall, my system is quite complex, but the audio signal path is no different from the Lightspeed except for the paralleled LDRs on the shunt side.

This is where I tried it as well as on the series side. Even the quality of the shunt component in speaker xovers make a difference, good caps sound better than cheap caps.

Cheers George
 
My LDR based DCB1 Build

I just finished my DCB1 build with LDR attenuation. It still needs a chassis and I haven't had the chance to listen yet - that comes tonight!

DCB1.jpg

The LDRs are mounted directly to the DCB1 board and are powered by Salas' 5V shunt regulator.

LDR Reg.jpg

I built George's LDR control on perf board and have the 100k log pot on a tether for some flexibility.

LDR Board.jpg

The power supply to the LDR control board is a perfect 5.00VDC. With 100ohm resistors and a 100uF caps followed by 0.47uF caps, I get a voltage range of 1.62v to 2.18v for the LDRs. This allows the audio resistance provided by the LDRs to vary from 22R ohms to about 6k7 ohms.

It still needs a nice chassis and bigger sinks so I can push the bias on the DCB1. I can't wait to give it a listen tonight with my AudioGD DAC and Pass a40 amp.

Thanks to George and Uriah for making this available!