Lightspeed Attenuator a new passive preamp

[wlowes]

Originally Posted by AndrewT

There is no argument about yield.

It's whether the S is different production components.

I am only guessing, but is it not customary to run one production line.. some output from the line are not tested and marked and sold as unsorted. Some subset are 100% tested and then graded as sorted. These would also be the ones that yield better specs.

This was the process with TDA1541a chips in the day as the best yeild got a crown and now sell for 10x the regular. Same with all intel processors. One production line produces a full range of processor speeds. The expensive fast ones are simply the ones that happened to come out better. I have to expect it is the same for these LDRs. I recall that when I ordered mine, I ordered 10 sorted ones hoping for a couple of pairs that match. They came back all ten from the C grade and all yielded 4 pairs with very tight matches. The original problem was that the distributor was not keeping them sorted. You could order 10 sorted and get 10 all with different grades at random. Complaints from this forum corrected that if I recall.

[jackinnj]

You could mount the LDR's on a transistor array, depletion MOSFET etc and heat it to a level above ambient and use one of the transistors to monitor the temperature -- just like a temperature stabilized frequency reference.

[bambadoo]

Her er foresten app note fra silonex regarding optcoupler volume control.. (It has probably been linked to before)

Silonex Inc.: Technical Reference: Audio Level Control with Resistive Optocouplers

[Scheff]

I use 2 LDRs and one stereo pot for the left channel, and the same for the right channel, and a mono switch . Then I regulate the balance by ear. If the center signal comes out right in the middle of the speakers, it´s fine. Back to stereo and voila. It´s a little more work, but I like it. And I wot go back to Pots anymore. And I only match LDR for the lowest Level, I dont match the upper LDRs at all. I use 4 DC sources , one for every LDR. You don´t have to like that, is just an idea I´m using for a couple of years... but the sound...

[georgehifi]

Originally Posted by wlowes

The original problem was that the distributor was not keeping them sorted. You could order 10 sorted and get 10 all with different grades at random. Complaints from this forum corrected that if I recall.

Correct, this was Allied Electronics where I first used to get them years ago, they were getting the NSL32SR2S in 1k lots which were baged and batched from Silonex.
But what was happening Allied were topping up their bins not yet empty with with the new stock (not of the same batch).
I got onto this and complained and Silonex sorted Allied out with their problem, not to top up bins with new stock unless they were empty.

Cheers George

[bambadoo]

Originally Posted by bambadoo

Her er foresten app note fra silonex regarding optcoupler volume control.. (It has probably been linked to before)

Silonex Inc.: Technical Reference: Audio Level Control with Resistive Optocouplers

WHAT
Well mixed up Norwegian with English and unable to edit the post now.

It says :
"Here is app note from ..........."

[pwgtang]

Originally Posted by udailey

Yes they are just that sensitive, especially when they are at higher resistances. At 250R or so you cant budge them but at 10k a 1 degree difference can mean hundreds of ohms. The motion of air from a door opening in another room can make a difference.
I have tested with two LDRs being monitored by one DMM each and have found that some LDRs will move more than others even though they get the same power and are initially set to the same resistance. So this is why I wait about an hour before I start testing and every time I change to a higher resistance I wait about 15 minutes before beginning the next test. I have a LDR attenuator running now that has been running for 2 years with the top off. My channel balance is pretty nice but its in a huge room and there are no hotspots around it. I have never built a LDR attenuator INTO an amp chassis. Not saying I wouldnt but I havent so I cant speak from experience. I can only say that table top testing of individual LDRs makes me think that, yes, temp changes will get them moving and the fact that they might move at different rates could show as a channel imbalance.

I recently was sent a DIY implementation of a Lightspeed for a look. The gentleman had issues with series LDRs not changing value and the balance being off to boot. Here he had done a point to point on breadboard using largish gauge wires as signal and power wires. The problem he had was that it took so much heat to get the wires to accept solder that the LDRs were damaged and in fact one gang of the control pot was damaged so that it no longer worked, while the other gang was okay. Heat kills.

I build LDR attenuators from time to time and sometimes I will find that the LDRs match according to my datasheets but past the max value of my datasheets they do NOT match. I never assume that they will but we can always hope.
With this build I have been able to solve the problem. Usually its one LDR of the 4 that is misbehaving at higher values. So lets say that at max resistance its sitting at 14k and its mate is around 12k and lets say this permeates a bit lower so that at 2k they are a great match and up to 6k they are a decent match and beyond they seem to be hundreds of ohms away from each other. I can fix this with a high value resistor in parallel with the misbehaving LDR. Oh LORD! You cant be serious! A real resistor in the signal?! Well, yes its a very high value that usually fixes the problem and its in parallel with a relatively low value LDR so the amount of signal passing the resistor is quite small in comparison to the offending LDR and to my ears and my Audio Precision the sound is not compromised subjectively or objectively. I start with a 249k and work my way to two 249k in parallel then try a 100k or a 75k. This is usually all it takes.. one of these options will make the problem disappear.
This brings us to gootee's suggestion about parallel LDRs. I made a batch of 10 of these boards a few years ago. I still have one of them. I goofed them because I designed it using the default pin sizes in my pcb layout software and never expanded them. SO I had to drill out several of them and then never rebuilt them. One reason only. It was just beginning to be to much for the builders. I was considering 8 LDRs where 4 are dynamic and 4 stay the same value during operation for fine tuning balance and for limiting max resistance. It works and its nice.
Back to matching: I match up to an average of 6k for a few reasons. 1: 6k sounds fantastic to me compared to other values so thats a subjective reason. 2: LDRs are relatively stable at 6k. They start to get a little squirrely around 5k but 6k is not to bad. Squirrely meaning affected to a greater degree by heat. So if I match to an average of 6k I can be sure I am selling the customer a reality up to that point. When I say an average of 6k I mean that I choose a voltage and current that results in most of the LDRs on the test boards to be grouping around 6k. Still I will have many up into 12-15k but most at 6k.
In no way should anyone assume that a set of LDRs that are tested to, lets say 8k, should be a great match at 15k. They might be but I dont know if they are.
So some things can be done here. If its a Lightspeed implementation a trimmer of a few hundred K can be placed in rheostat mode across each of the gangs of the 100k control pot. Max resistance of the LDRs can be brought a bit lower. Lower means more stable. So you can now control your Lightspeed's total resistance.
Lost in all these posts is a way to increase your Lightspeeds total resistance. A 100R multiturn trimmer in series with the 5V fed TO the volume pot. Around 10R will make dramatic changes in max resistance and you will need the multiturn capability as its very dramatic changes with just a touch of the dial on the trimmer.

Yes, the LDR are so sensitive to the temperature changed, after many tries, I had built one with temperature compensation, I can used any LDR as it can go to 50 ohm or below, I used four pots to adjusted each LDR, I can calibrated it from 5k to 50k ohm ranged. I had put the attenuator on a hot plate, heat up very slow from room temperature to 36.4C and the channel balance changed less than 1db. Set up as: input signal 1 khz 1.5 vrms attenuator=20k
volume set at 12 O'clock. At 20.7C L channel = -7.42db R= -7.84db
at 31.1C L= -6.59db R= -6.80 at 36.4C L= -6.83db R= -7.40db

[udailey]

Originally Posted by pwgtang

Yes, the LDR are so sensitive to the temperature changed, after many tries, I had built one with temperature compensation, I can used any LDR as it can go to 50 ohm or below, I used four pots to adjusted each LDR, I can calibrated it from 5k to 50k ohm ranged. I had put the attenuator on a hot plate, heat up very slow from room temperature to 36.4C and the channel balance changed less than 1db. Set up as: input signal 1 khz 1.5 vrms attenuator=20k
volume set at 12 O'clock. At 20.7C L channel = -7.42db R= -7.84db
at 31.1C L= -6.59db R= -6.80 at 36.4C L= -6.83db R= -7.40db

Thats pretty "COOL"! How do you compensate for temp? Are you using something similar to the temp comp on a LM334 or are you using thermistors in parallel with the control pot? I suppose that would only work in higher resistances.

[pwgtang]

Originally Posted by udailey

Thats pretty "COOL"! How do you compensate for temp? Are you using something similar to the temp comp on a LM334 or are you using thermistors in parallel with the control pot? I suppose that would only work in higher resistances.

I am not used ccs or fix 5 volt, my designed was variable the voltage to
the LDR led, I used thermistors to compensated the voltage to the led.
It only need to compensated uv/degree, with variable voltage I can used a switch to changed the output voltage to the LDR's and the resistance will changed, this design the attenuator can had multi resistance value.but I found two is good enough high or low resistance, high/20k for my tube amp,low/10k
for solid state amp.

[udailey]

Great innovation! I really like the thermistor idea.