Lightspeed Attenuator a new passive preamp

The stepped way you are talking about I heartily agree with. Should work well but they still have to be moderately matched because when temp changes and they change resistance you want them to change in the same direction/rate. Also I wondered if people would put up with soldering 24 3 legged pots that cost about 84 cents each for each LDR. You might get away with 24 for each channel

Agree on still matching, to some level. Then only add pots as needed, to one side (L or R) as needed to get good balance.

I thought about a board, with mostly blank spaces for adjustment pots, which are in series with the normal shunt/series resistors. Also would have a jumper wire in parallel to the adjustment pot position to start. The jumper is cut when the pot is added.
Or use the alternate way suggested with parallel pots.

Probably excellent balance is only needed at louder volumes. Does anyone care about 2 dB mismatch for background/very low levels?
 
fit fixed resistors.
Then trim (by adding a parallel resistor) one or other of the pair to bring the LDRs to similar attenuation.

The trimming could be by using a variable resistor (1M0 or 2M2) and replace with a fixed once measured.

The big advantage of switched & fixed resistors, is the ability to choose the attenuation to suit the listening conditions.
You could have coarse steps from -1dB to -100dB or finer steps from -1dB to -30dB or again finer steps from -20dB to -50dB, or finally from -40dB to -80dB. Just get a ball park attenuation range that suit your equipment and range of listening levels. Then choose the fixed resistors to suit.

You could use a 3step attenuator:
background music / answering the doorbell.
listening attentively
party !

Good idea, that would work great.

I also thought about resistor selection based on LED/LDR measurements.

If you designed a program/spreadsheet, you could enter the LED/LDR measurements for resistance vs. current for series and shunt positions, and than have the program tell you what resistors to load for a perfect match on your stepped attenuator.

This way you'd know before you had to buy them, so you didn't need every possible value on hand.
You would need decent resolution on the measurements, at least 10-12 points perhaps, to get a good curve fit.

Last weekend I made such a 10 step attenuator to measure devices on hand, 3 at a time. It measures them from 5ma to 1ua, and works great. Figuring out the temp way to reliably hold the LED/LDRs was the hardest part. Those leads are pretty non-standard.
 
This is a shunt attenuator. LDRs in series, Dales in shunt. I get zero volume at min, customized curve, great channel matching, wonderful sound and there are zero contacts in the signal. Channel matching is best with matched resistors though, so still didnt get away from that part. THD is lower than LDR/LDR situation and channels are better matched. I can have infinite channels as well.
https://picasaweb.google.com/lh/photo/p2TYXQRALqF_Rp5kDG_2n7p-AhcJSuUnH3KCk0w1PKc?feat=directlink
 
Last weekend I made such a 10 step attenuator to measure devices on hand, 3 at a time. It measures them from 5ma to 1ua, and works great. Figuring out the temp way to reliably hold the LED/LDRs was the hardest part. Those leads are pretty non-standard.

For anyone interested, this may save you time to do the same. Here's what I used for the resistors in the step attenuator driving the 10 point LED/LDR measurement circuit.

Numbers below are resistor values, series / shunt. Output voltages are shown for a 5V DC voltage source to series side, and shunt to ground. The division ratio, output voltage, and resulting current across LED when used in the VCCS circuit where 5V = 5ma. That VCCS circuit is not for publication today, but not anything new or spectacular.

0R /10k, divide by 1, 5V, 5ma

10k / 10k, divide by 2, 2.5V, 2.5ma

40k / 10k, divide by 5, 1V, 1ma

9k / 1k, divide by 10, .5V, 500ua

50k / 1k, divide by 50, .1V, 100ua

10k / 100R, divide by 100, 50mV, 50ua

50k / 100R, divide by 500, 10mV, 10ua

10k / 10R, divide by 1000, 5mV, 5ua

20k / 10R, divide by 2000, 2.5mV, 2.5ua

50k / 10R, divide by 5000, 1mV, 1ua

Bob
 
Hallo,

I just finished and installed my lightspeed kit. It sounds wonderfull.

I have a problem though. The minimum level is just too loud to be acceptable.
I read in this thread, that it would be possible to add an additional series LDR to increase attenuation. How does one best manage this? Can someone give a short hint?

I guess the possible attenuation with second series LDR will be also "not so much different" than it is now, isn´t it?
I´m testing the lightspeed with a 1.8x-gain-preamp at the moment, an it is too loud (for quiet). Finally I would like to install it in my main system with 6x-gain-preamp. I guess this would probably compensate the additional LDR then? Can someone please convert the attenuation db-values into dummy-x-times gain?

Thanks,
Florian
 
Florian
Using gain when its already to loud really doesnt make sense unless your only reason to use the gain circuit is for better impedance matching. Take the gain out and try it.

Yes, i know. But the preamp gain can´t be changed, in fact, it is already reduced..
I will use the Lightspeed for a buffer circuit, that I am about to start, and which I am a little excited about. So this is no problem. I think this is maybe "where it belongs".