With a log pot, the *most* control and resolution is at the full counterclockwise position. More than a linear pot. So the log pot works better than linear in this case, at least in my tests. The matching at at low resistance is fairly easy in comparison to matching at 1-10Mohm. BTW, this is WAY off the standard datasheet Silonex current - resistance curve, involving current control at levels in the single digit microamp range.
Pot drives a high impedance voltage controlled current source. So no worries on power through pot. Current then goes to the LEDs, then to gnd. 4 independent closed loop current sources, one for each LED in a stereo unbalanced volume control. Makes dual mono or balance control easy, plus lets you potentially make series and shunt LED control different from each other.
If you do this in a way where voltage from pot is directly proportional to current, you can then play with reducing voltage on the input to pot. This will let you move the LED operation to the left hand side of the curve, for instance, operating at min. of 0, maximum of 2.5 ma, or even lower. You could do a max. of 0.1 ma if you wanted to.
You can even install a 2nd pot, or step attenuator switch to make the voltage to the pot input variable, giving different ranges of control for different conditions. Not pie in the sky, tested and working nice. Required some extensive component testing and evaluation to find the right ones that can do it.
No I have none of these problems with my design, voltage at the UAR LDR's anode to cathode ranges from 1.35v to 1.66v, and LAR anode to cathode from 1.46 to 1.7v, current is regulated to 24ma , 1.25/26+1 (UAR comprising two LDR's ) and current measured at the anode of UAR one LDR ranges between 16.9ua and 24.1ma, and is shared evenly to the other LDR two also 16.9ua to 24.1ma. Signal side the connection with UAR and LAR is standard L pad, and it provides a very pleasant log response, and the best audio myself and a friend have ever heard with an attenuator.
Cheers / Chris
Last edited:
Yes, lots of different ways to do the same thing.
Point is, at 20 mA you can get 40 ohms, at 4mA 100ohms, and at 1mA 140 ohms.
These numbers are nothing compared to changes like going from 100 ohms to 1K, 10K, 100K, and 1 Mohms. All the big changes happen below 1mA.
I like to use the pot resolution to control that action, not waste like 1/3 of the total pot rotation by going from 40 to 140 ohms.
Not sure what you referred to in your response as a "problem".
What is your minimum input impedance?
hi BFNY The problem I should qualify with reference to your earlier post that does not occur, is load on upstream electronics. Yes you are right, the different ways all lead to perfect audio , and all the changes occur below 1ma which we are I think in total agreement with. Lowering voltage and providing current regulation is the key to excellent audio with LDR's Input impedance is very high. I sent Andrew T a PM, so will await his reply to me... before I respond further, possibly with a new post.
Cheers / Chris
Having measured LDRs to at least 1M ohm, these are not easy to match even up to 50KOhm. I question the log pot curve tolerance because I have not seen any specs on these.
Chris;
Fully understand. No need to go any further. Looking back, I now see from voltage ranges given above you are utilizing a unique control method to solve the input impedance issue.
To control the resistance at 1M ohm range, you need to be able to control current, *rock solid*, at = or less than 1uA. I don't think one can do this with voltage control.
You are working the lower end of I-V curve for an LED. I think most people know what that looks like, if not, look it up.
At the low end you have incredibly small changes in V for large changes in I.
I believe we are talking about matching levels L and R channels here, right ?
If you drive both L and R channels from the *same pot*, why is pot accuracy a problem? Both channels get the exact same voltage control, so are matched, IF the LDRs are matched.
This assumes you use a dual section pot, and one section drives both the L,R series LDRs, the other pot section drives the L,R shunt LDRs.
If you are talking about sorting LDRs to match them, no, don't use a pot to set current. Use low noise resistors in a switched attenuator to get constant current points for sorting. And yes, let them stabilize before taking a reading. It's not just temperature.
If you use a simple voltage divider to match LDRs, I think you will not get constant current points -- you get constant voltage points, and that's OK if you're using a voltage-controlled method as in the Lightspeed. A more accurate method would be to use a precision current source both for testing and for control of the four devices in a stereo setup during operation.
When I tried to design a voltage-controlled setup, I got nowhere because the resistances varied quite wildly even with precision voltage control. Now that I'm looking at current control I can achieve stable resistances all the way up to 10K and beyond, but it still takes quite a while for the LDR resistances to settle down that last little bit after the 'pot' has been moved.
You are absolutely right about letting the LDR stabilize -- at higher resistance values around 8~10K with a precision current source, it will take an LDR up to two minutes for the output resistance to stabilize. And that is with vanishingly small current, so it's not heating that's causing the change.
We are in agreement on all points about using current control. As per my posts a few above this one, I advocate using a voltage controlled current source that can control accurately and repeatedly down to 1 micro amp.
The point was for matching LDRs, to get repeat ability from unit to unit, use a resistor based divider to create the precision voltages that THEN drives the voltage to current source.
LDRs do not behave, in my experience, above 22k and generally a bit lower. Luckily they sound best around 6k total resistance so this doesnt bother me. Subjective of course.
Distortion is not an effect of the voltage fed the diode but it is a result of the signal voltage. Lowest distortion is at 200mV of signal. Again, beside the point because they smoke everything out there (yes, yes subjectively) even when distortion is hitting higher than 1%.
2.5V is no magic number for good sound. I have to disagree with you. My best subjective results and best results measured on my AP are with voltages between 10-13V (I have not tested higher as at 22V I have had an LDR explode even though its rated to 40V).
The magic is in an extremely quiet supply, current sources and caps across the LED. Its pretty simply and anyone with a basic Lightspeed circuit can make obvious incremental improvements testing for better sonics by improving their voltage source (make it quiet) and adding caps on the LED.
Going current source takes it to another level and probably takes it to another thread. Its the Lightspeed thread, not the "Launch my new product on George's shoulders thread."
Distortion is not an effect of the voltage fed the diode but it is a result of the signal voltage. Lowest distortion is at 200mV of signal. Again, beside the point because they smoke everything out there (yes, yes subjectively) even when distortion is hitting higher than 1%.
2.5V is no magic number for good sound. I have to disagree with you. My best subjective results and best results measured on my AP are with voltages between 10-13V (I have not tested higher as at 22V I have had an LDR explode even though its rated to 40V).
The magic is in an extremely quiet supply, current sources and caps across the LED. Its pretty simply and anyone with a basic Lightspeed circuit can make obvious incremental improvements testing for better sonics by improving their voltage source (make it quiet) and adding caps on the LED.
Going current source takes it to another level and probably takes it to another thread. Its the Lightspeed thread, not the "Launch my new product on George's shoulders thread."
Dave
Definitely use caps then. You should be looking at junk box caps from 10-100uf, I usually use 47 or 100 and then place a small film on it as well. You can hear these differences immediately. Also put 1uf tants on each leg of your regulator to ground/0V and place some decent capacitance before and after. A few thousand before and 100 after the reg. Experimenting with Earth/0V/Signal Ground with a wire and alligator clips while listening will often show a weakness/room for improvement.
Uriah
Definitely use caps then. You should be looking at junk box caps from 10-100uf, I usually use 47 or 100 and then place a small film on it as well. You can hear these differences immediately. Also put 1uf tants on each leg of your regulator to ground/0V and place some decent capacitance before and after. A few thousand before and 100 after the reg. Experimenting with Earth/0V/Signal Ground with a wire and alligator clips while listening will often show a weakness/room for improvement.
Uriah
Chris I read your posts and found you to be very knowledgeable and helpful. I am sorry for my snarky remark about launching a product on Georges shoulders. It doesnt look like thats your style. I am fairly sure its not your intention to wind up a bunch of customers using George's popular thread. Its happened so many times in the past and I get a bit worked up. I should have waited a while til I posted as I usually do when I get a little bent.
I am sure these guys would love to know what you think is a great solution but if you know the moderators and George you know that they like to stick to the Lightspeed circuit here, not alternatives. Alternatives require their own thread. Let us know when Silonex allows.
Uriah
I am sure these guys would love to know what you think is a great solution but if you know the moderators and George you know that they like to stick to the Lightspeed circuit here, not alternatives. Alternatives require their own thread. Let us know when Silonex allows.
Uriah
Hi Uriah
No that is not my style, I note your use of the words " bunch of customers " and that the forum is analog line level, not vendors forums, which correct me if I am wrong enables comment and discussion about analog line level topics, of which this is one.
"bunch of customers" to me sounds like reference to vendors forums. I am also mindful and respectful of creating a new topic when required.
Cheers / Chris
Well I tried waiting for a while but still have the urge to post.
While its in the analog line level forum its not in the "any old circuit" thread. Its the Lightspeed thread. George obviously sells the Lightspeed but he doesnt try to sell it here, only describe it to DIYers and help them build their own.
You might note that Nelson Pass has many threads (even his own forum), for instance the F5 and B1. These are commercial products as well (and not in the vendor forum) but you wont find people talking about circuits that do not resemble the F5 or B1. The inclusion of a IRFP9240, like that used in the F5, in the circuit does not make a different amp circuit 'on topic' in the F5 thread. Simply because you are talking about LDRs and attenuation does not mean that your circuit is the topic of this discussion.
If you were helping people build their Lightspeed, and there has been plenty of opportunity in the last few days.....but you arent. You are saying there are better ways.. but you cant divulge. Tres mysterious!
"Threadjacking is the practice of taking over a thread by posting off-topic replies such that the original topic becomes diluted or lost. Off-topic posts, and replies to off-topic posts, can be a positive outcome of discussion, but must either be brief or be moved to another thread. If something interesting does arise that warrants extensive discussion -- then start a new thread and link to it."
While its in the analog line level forum its not in the "any old circuit" thread. Its the Lightspeed thread. George obviously sells the Lightspeed but he doesnt try to sell it here, only describe it to DIYers and help them build their own.
You might note that Nelson Pass has many threads (even his own forum), for instance the F5 and B1. These are commercial products as well (and not in the vendor forum) but you wont find people talking about circuits that do not resemble the F5 or B1. The inclusion of a IRFP9240, like that used in the F5, in the circuit does not make a different amp circuit 'on topic' in the F5 thread. Simply because you are talking about LDRs and attenuation does not mean that your circuit is the topic of this discussion.
If you were helping people build their Lightspeed, and there has been plenty of opportunity in the last few days.....but you arent. You are saying there are better ways.. but you cant divulge. Tres mysterious!
"Threadjacking is the practice of taking over a thread by posting off-topic replies such that the original topic becomes diluted or lost. Off-topic posts, and replies to off-topic posts, can be a positive outcome of discussion, but must either be brief or be moved to another thread. If something interesting does arise that warrants extensive discussion -- then start a new thread and link to it."