[mgalusha]

After reading much of the long threads on diyaudio and the shorter ones here I picked up some of the LDR's that Uriah D. was matching. This was 6-8 months ago. Fast forward to early June and I finally got around to building one up and installing it in my diy tube preamp. I was pretty amazed with the results though I'd made some other mods at the same time so did not realize just how much the volume control was contributing.

A couple of weeks ago I decided to rework the preamp a bit more and swap in some new tube sockets as the old ones were getting noisy. While I was in there I decided I needed to rebuild the LDR on better board as I'd used some Radio Shack phenolic board the first time as it was all I had at the time. I also wanted to see if I could work in a balance control as the LDR's don't track perfectly, even matched ones are a little different at various levels.

So I put in a regular Alps log pot and buttoned it up. I have formed the habit of doing some quick measurements on gear before it leaves the bench, I don't want any big surprises when it goes into the system. It measured fine so I put it back in the system. This was a rude surprise, somehow much of the magic had been removed from the system. Not that it sounded crappy by any means, but it wasn't very close to what I had become accustomed to.

The new FR4 board arrived this week so I built a new one this morning and came up with a working balance control as well. I have not put it in the preamp yet as I am out of time for today but figured I would take a photo and post this bit.



Schematic. The only real difference between this and GeorgeHifi's (the Lightspeed guy) drawing is the addition of the dual 20K linear pot. One of the beautiful things about this design is none of the pots carry the audio signal or need to track well, so cheap ones are just fine.







You might notice the resistors across the balance control. That is because I didn't have a 20K dual pot on hand but I had several 100K models. I just added 2 10K resistors across each leg of the 100K pot, not exactly 20K but good enough for this, it's really not too critical. Since none of the wiring carries the signal I used CAT-5 network cable to wire the balance pot. I needed 6 conductors, it had 8 and I have lots of it. Perfect.

I built it using perf board and turret posts with some wire wrap wire handing the point to point connections. The little 500K log pot is supported in back by a little U shaped piece of wire soldered to the board and the pot body. This is because the pot will be the only support for the board. It was a bit fragile with just the pot leads, this makes it nice and stiff.

A clean 5VDC is required, total of about 50mA. It needs to be clean as any noise and ripple will modulate the LED's, which will of course modulate the photo resistors.

Seek out the LDR threads for more info that you could ever want to know.

I'm just hoping when I put this back in tomorrow that the missing magic returns with it.

mike

[JoshK]

Very cool.  I still haven't gotten around to trying mine out yet.  One of these days.

[jkeny]

Looking good, Mike (i'm now your official stalker; following you around between threads - it just seems I have the same interests as you).
Interesting note on the loss of magic - hmm
I suppose you caught the whole debate concerning MkI Vs MkII configurations (for those not up-to-speed on these devices, Mikes config is a MKII using an LDR for series & shunt duties; MKI uses a series RESISTOR & shunt LDR). I have only heard the MKI & thought it sounded excellent. Will be hoping to hear a MKII soon side-by-side.

Keep us posted!

[mgalusha]

I did read some of the posts about using the LDR in shunt mode only vs series/shunt. I also read the app notes from Silonex and like the attenuator and distortion curves of the series/shut configuration. They also show a series/shunt configuration with an additional 2k2 resistor before the series element. This is how I wired the first one as they show a little better attenuation. I'm going to try ii with and without the additional resistor and see how it sounds.

[jkeny]

The series R/shunt LDR has the lower distortion at around the voltages that it will be used in normal listening conditions 100mV to 500mV but that, of course, doesn't mean it will necessarily sound better.

It's an interesting phenomena that nobody has quiet explained the good sound out of this thing - two theories around: one is that it avoids the fine contact area of potentiometers which tend to act like a capacitor at high frequencies; second is that the resistive element of the LDR is CdS (Cadmium Sulfide?) which is acting like a very high quality resistor, equivalent to a Vishay Bulk nude resistor. Who knows?

[TomS]

Very nice Mike.  Glad to hear there is magic there as it seems so deceptively simple.  I'd still like to try the ZenMod balanced version mated to the Paul Hynes remote VCCS setup.  One of these days...

Oh, and I love those little turret posts.  I used those years ago and much preferred them for the PC boards I designed with hardwire connections.  I definitely need go back and peruse the Digikey Keystone pages and pick up a staking tool and some parts.

[Scott F.]

Mike,

Your findings are interesting. Jon VerHalen pointed me to to the Lightspeed thread a while back. Its an interesting design.

Can you elaborate on your 'lost its magic' comment a bit? Is it that the LSA is just a little too 'clean' sounding?

[mgalusha]

I need to amend this thread with an updated schematic. What works good on the bench is not always so good in reality...

I tried the LDR both ways, as a series/shunt and as shunt only with a fixed series resistor. In my system I preferred the fixed resistor with the LDR as the shunt element. I used a 10K series resistor.

I also had to revise the balance control. 20K is too high and prevents the LDR from getting enough current when the volume pot is all the way down, it just didn't go low enough in volume. Since I'd decided on just using one LDR for the shunt element I only needed a single deck balance pot and had a 10K on hand. That was still too much so I put a 500R resistor on each leg to the wiper to make what is more or less a 1K pot. This did the trick nicely and now it sounds very good again and the magic is back. One of those things where had I not heard the difference between a standard carbon pot and the LDR it would be tough to believe.

Scott, to answer your question, the magic is with the LDR in the circuit. It is clean but not in a clinical way, just seems to remove a layer of haze that I didn't know I had. More low level details such as spatial cues but not shoved at you, they just seem to be there, like they were already there but obscured.

Mike

edit for stupid typo's...

[jkeny]

Hey Mike,
That's a result - saves you 2 LDRs & all the matching, etc, - simplifies the circuit - no more complexity than is needed.