Lightspeed Attenuator a new passive preamp

George: the middle fets both measured 7.55mA and the outer fets measured 7.57 and 7.59mA. The other six were random and measured between 9.27 and 10.54mA. I did not use any coupling caps at all. I am not sure how to measure the output impedance, though I would be happy to if you could provide some instructions.

Uriah: I was wondering if the caps should have been closer to the LDRs... I can certainly move the little Wima ones there on the next iteration.

Now, I have a problem to solve. The minimum volume that I can achieve from the LDRs is quite high and a bit distorted. I am using an Audio GD DAC which has an output amplitude of 2.5V and my a40 amp needs only about 1.5v to reach max output power. As built, I can't turn it down low enough to reach a comfortable volume- I am somewhere north of 90dB at 1m from the speaker.

I tried inserting a 4k75 resistor in series with the 5v to the pot to reduce the voltage going to the LDRs. The new min voltage that goes to the LDRs is 1.375v and the max is 1.622v. At min volume setting, the sound is very thin and hollow - mostly treble and no mid/bass. The LDRs seems to need a minimum power of about 1.390v before the bass and treble comes back.

Any ideas for a solution?
 
Last edited:
George: the middle fets both measured 7.55mA and the outer fets measured 7.57 and 7.59mA. The other six were random and measured between 9.27 and 10.54mA. I did not use any coupling caps at all. I am not sure how to measure the output impedance, though I would be happy to if you could provide some instructions.

You can measure the output impedance with a load resistor, whatever load resistor value drops a 2v or 1v sine wave by half, is the same resistance as what the output impedance is, as all you are creating is a voltage divider, so if the load resistor is 200ohms that drops the output by half then the output resistance of the B1 is 200ohms.



Uriah: I was wondering if the caps should have been closer to the LDRs... I can certainly move the little Wima ones there on the next iteration.

Now, I have a problem to solve. The minimum volume that I can achieve from the LDRs is quite high and a bit distorted. I am using an Audio GD DAC which has an output amplitude of 2.5V and my a40 amp needs only about 1.5v to reach max output power. As built, I can't turn it down low enough to reach a comfortable volume- I am somewhere north of 90dB at 1m from the speaker.

If the min volume is too high and distorted it can be one of two things, either you have the input in the output and the output in the input.
Or for the minimum volume problem if the ldr's are not quad matched you may have put the series in the shunt and the shunt in the series positions. Happy Uriah?
Cheers George


I tried inserting a 4k75 resistor in series with the 5v to the pot to reduce the voltage going to the LDRs. The new min voltage that goes to the LDRs is 1.375v and the max is 1.622v. At min volume setting, the sound is very thin and hollow - mostly treble and no mid/bass. The LDRs seems to need a minimum power of about 1.390v before the bass and treble comes back.

Any ideas for a solution?
 
Last edited:
In what universe do quad matched LDRs produce less distortion than any other LDRs that are otherwise in the same situation and dividing the input voltage into the same output voltage?
Anyway, The problem is the 4k75 resistor in front of the 100k pot. If you want to use a resistor there its a fine idea but you need to be using a multiturn trimmer with a final value of maybe 10 ohms. Just ten or so.
What you could do instead is take a 100k trimmer and put it in a shunt position on the output of the 100k gang that controls series LDRs. This trimmer goes from the output of the pot gang to ground. It divides the voltage down a little bit to increase the max value of the series pair. Now, 100k is what I used in the Lighter Note and I used higher than 5V so I am unsure if 100k is the right value for this postion. It may need higher or lower but if lower then you for sure have it with a 100k since it'll cover that value. If you have a 500k I would start there since it'll give you either a great value or a good idea of where you need to be. Since it increases max value of the series then the output voltage will be less, lower volume, even when the shunts stay the same. Right now your 4k75 is making the series WAAAAAY to high resulting in the thin sound.
 
If the LDRs are quad matched I would assume they would be identical. All four of them. In which case put any LDR anywhere you want. There should be no Shunt or Series LDR.
I was looking at the numbers of measured LDRs today. Here is a set of 2 pair that I think would go well together
193 246 1207 3103 7516
192 242 1189 3098 7693

185 233 1117 2885 7132
190 241 1130 2821 7042
Certainly they are not a quad matched pair and I would normally try to find pairs that are closer matched. Like these..
236 302 1495 3912 9644
239 305 1502 3924 9652

225 289 1459 3834 9473
227 290 1466 3867 9550
Those are great matched LDRs and the pairs are even pretty close to each other but I dont have the desire to try to sell them as a quad. I've never really known what I should define as a quad except 4 that almost exactly match. I still wouldnt sell them as a quad.
But it doesnt matter. This is what gets me about calling anything a quad. Calling them a quad is misleading and in this thread seems to be a leading statement. Meaning that folks are left with an impression that the resistance of the series will be equal to the total resistance of the voltage divider at max or min volume minus the value of the shunt at ALL volume positions which just isnt the case no matter how perfect the matching is. Matching LDRs perfectly has no influence on the resistance curve of those LDRs when controlled by a 100k dual log or even linear pot. These guys that are using lookup tables are the only ones who will ever be able to control the total resistance at all volume settings. I'll show you what I mean.
I made these curves when I started matching LDRs and i used massive quantities of pots all dialed into the exact same value to test LDRs. I used values from 1k up to 100k to replicate being controlled as they are in a Lightspeed. So we get this slope. This is how a shunt LDR resistance would look as you turn up the volume on your 100k dual pot. Now simply reverse that picture and think of whats the total resistance of this voltage divider I've made with series and shunt LDRs controlled by a 100k pot? Its not static. Its dynamic. Its the shape of a smile and it sounds freaking fantastic and the total resistance is not at all the same at mid volume as it is at min or max. Quad matched isnt going to change this. If you use a set of LDRs that are hitting about 12k at 100k you can bet your total resistance at half volume is going to be around 70% of that and in fact the easiest way to make the total resistance to stay about the same in a usable working range is to use a handpicked pair that nowhere near resembles the other pair when being controlled with identical voltage and current. This way you can slide the resistance curves around til you find a pair that actually give you a near static total resistance in a band of useable volume. I have done that but I certainly dont make an effort to do this as I dont mind max and min volumes having much higher total resistance than my main listening volume which I like to be more around 6k. Thats what sounds good to my ears in my system.
The Audio Precision tells me that the lower the total resistance then the lower the THD+N and also that the lower the input voltage the lower the distortion.
I dont recommend over a 2.5V input voltage because the THD does start to get silly only a bit higher than 2.5V. I love my AudioGD DAC and it gives me 2.5V just like our buddy a few posts up. So I think his problem has nothing to do with quad matching and fairly little to do with his input voltage.
The other two pics are how I used to do matching. Now I use another method because its just to fiddly to adjust dozens or hundreds of pots for each and every measurement.
 

Attachments

  • ldrRange.jpg
    ldrRange.jpg
    57 KB · Views: 516
  • DSCF3241.JPG
    DSCF3241.JPG
    108 KB · Views: 516
  • DSCF3240.JPG
    DSCF3240.JPG
    174 KB · Views: 515
Last edited:
I took some voltage readings from the LED side of my LDRs as installed on the DCB1 board - perhaps this will provide some insights. I also double checked the input and output wiring. They don't appear to be mixed up...

On the DCB1 board, the holes on the left of where the pot goes lead to ground, the middle holes are the signal out, and the holes on the right are the signal in.

From the two sets of matched LDRs, the first matched pair connects across the In and Out mounting holes. The second matched pair connects across the In and Ground mounting holes.

With no resistor in series with the pot (as per George's diagram), when the 100k log pot is rotated for minimum volume, I get 1.48v on the legs of the LED that sits across In to Out and 2.00v on the legs of the LED that sits across In to Ground. When I rotate the pot for maximum volume, I get 1.98v on the LED that sits across In to Out and 1.48v on the LED that sits across In to Ground.

As I rotate the pot from min to max volume, the voltage across the In to Ground LED falls more quickly than the voltage across the In to Out LED increases, until I hit the half way mark of the pot's rotation. Then the behavior reverses.

Also, the change in volume exhibits a several second "delay" from when I adjust the pot. I suspect this is a result of the 100uF caps charging and discharging as the pot changes. Is this normal?

Have I connected the LDRs in the proper manner?
 

Attachments

  • Mez_pot_top.png
    Mez_pot_top.png
    12.6 KB · Views: 493
Last edited:
The LDRs that I used exhibit the following readings:

Matched Set #1:
LDR#043: 208 1230 4321 9162
LDR#207: 198 1183 4287 9262

Matched Set#2
LDR#045: 183 1123 4252 9465
LDR#041: 184 1190 4395 9512

One channel uses LDR #043 from In to Out and #041 from In to Ground. The other channel uses LDR#207 from In to Out and #045 from In to Ground.
 
Nice work Eric, I'm interested did you manage to build the B1 without using any output coupling caps by matching the fets very tightly. Because from my experiences a coupling cap will mask all the good that comes before it.
Also can/did you measure the output impedance of the B1, as I don't believe it's very low (for a buffer).

Cheers George
the B1 has an output offset of roughly half the supply rail voltage. This must be blocked with a suitable capacitor before it enters a power amplifier.
There is no option to delete the DC blocking capacitor.

Did you mean the DCB1? It has no DC blocking capacitor since the output offset is virtually eliminated by using selected Idss jFETs.
 
George: The numbers in post #5128 are the matching numbers provided by Uriah - I am not sure exactly what he measures when he matches them. I tried to take readings from the legs of the LDRs while installed in the circuit, but I was having a great deal of difficulty - lots of inconsistency for some reason. Perhaps this is the problem... So, I removed the LDRs from the circuit and read the actual resistance as I varied the 100k volume pot.

At minimum volume, my series resistance is about 22k ohm and shunt is about 18 ohms. At maximum volume, series is about 24 ohms and shunt is about 22k ohm.

So, it appears to me that the LDRs and their control circuit are functioning properly. The problem must be somewhere else...

Andrew & George: This is a DCB1 build with no output caps. My plan is to put this in front of a new F4 that I am building.
 
Last edited:
I was thinking it was my wiring of the LDRs that was in error... I'll make the switch and let you know how it turns out. I suspect we already know the outcome!

Uriah, thanks for you kind compliments on my theater and projects! They span about 11 or 12 years now and despite little mistakes like this mixup, they provide tons of fun and entertainment.
 
Success! Thanks for the help - I appreciate it very much! Over the past few weeks, I've been making some upgrades. Each one has been noticeably better and bringing them all together just brings a smile to my face!

I started with an Audio GD DAC. This was a nice improvement over the very nice DAC that is in my Adcom GCD-700 CD player. More 3D imaging, greater separation of vocals and instruments. Now, I have a DCB1 instead of using "Pure direct" mode of my Marantz AV8003 preamp for music. The Marantz preamp is a nice piece, but the DCB1 with LDRs brings a new level of clarity, focus, and what I can only describe as "crispness" to vocals and music. A few weeks ago I replaced all of the series resistors in the mid and treble networks of my speakers with Caddock MP821 resistors and added some extra bracing to the rear of the open baffle. The cumulative impact of all these changes is really quite nice!
 
Success! Thanks for the help - I appreciate it very much! Over the past few weeks, I've been making some upgrades. Each one has been noticeably better and bringing them all together just brings a smile to my face!

I started with an Audio GD DAC. This was a nice improvement over the very nice DAC that is in my Adcom GCD-700 CD player. More 3D imaging, greater separation of vocals and instruments. Now, I have a DCB1 instead of using "Pure direct" mode of my Marantz AV8003 preamp for music. The Marantz preamp is a nice piece, but the DCB1 with LDRs brings a new level of clarity, focus, and what I can only describe as "crispness" to vocals and music. A few weeks ago I replaced all of the series resistors in the mid and treble networks of my speakers with Caddock MP821 resistors and added some extra bracing to the rear of the open baffle. The cumulative impact of all these changes is really quite nice!

Good to hear you sorted out your problem Eric.
What amp/s are you driving the Lightspeed with?
And what is your source?

Cheers George
 
Hi George,

My thanks to you and Uriah for helping me solve my silly problem and obviously for sharing your great work with us!

My plan for this system is my office at work. Over the weekend I was using a Pass Labs a40 amp that I built about a decade ago. I have a Pass F4 in the works right now that is just a few parts shy of being completed - this is my target amp for my new preamp. I have three primary sources in mind: CDs, satellite radio, and streaming from Pandora, etc. And before the whole adventure is done, I'll probably "need" to build a new pair of speakers, too ;)
 
Over the weekend I was using a Pass Labs a40 amp that I built about a decade ago. I have a Pass F4 in the works right now that is just a few parts shy of being completed - this is my target amp for my new preamp.

Eric with the A40 you need the DCB1 as it is bi-polar 12.5kohm input and this may not able to be increased, you could ask Nelson to see if you could take it to 47kohm as the the 7kohm of the Lightspeed will be the resistance seen by the bi-polars which is almost 10kohm anyway.

With the F4 it has 47kohm input so try the Lightspeed into it without the DCB1, and seeing it's fet input you could take this say to 68kohm or even 100kohm no problems, I think you will be in for a very nice surprise.

Cheers George
 
Last edited: