With ANY LDR attenuator you can change the Rtot to whatever you want. Mine is easier to do but you can even do it with the Lightspeed.
Want a lower Rtot with the Lightspeed? Use a lower value pot to control it.
Want higher Rtot so that its quieter at min volume? Put a high value (100k+) pot from series V+ to 0V and dial it so that it drains a tiny bit of power away. This tiny bit wont make a difference ever until you hit near and at min volume and then it will drive the LDR to very high resistance because this tiny bit being drained away to 0V is very much larger relative to the power the LDR is trying to use.
NO LDR attenuator only using one LDR per shunt will go min volume as in 'really quiet' with a super efficient or high gain system without doing something like this (getting series to high resistance) because it is NOT POSSIBLE to drive a shunt LDR to below 40R without suffering long term or even immediate damage to the LDR. We have here a voltage divider. A min shunt of 40R means you cant go to 0V for signal. So volume cant go quiet.
Want a lower Rtot with the Lightspeed? Use a lower value pot to control it.
Want higher Rtot so that its quieter at min volume? Put a high value (100k+) pot from series V+ to 0V and dial it so that it drains a tiny bit of power away. This tiny bit wont make a difference ever until you hit near and at min volume and then it will drive the LDR to very high resistance because this tiny bit being drained away to 0V is very much larger relative to the power the LDR is trying to use.
NO LDR attenuator only using one LDR per shunt will go min volume as in 'really quiet' with a super efficient or high gain system without doing something like this (getting series to high resistance) because it is NOT POSSIBLE to drive a shunt LDR to below 40R without suffering long term or even immediate damage to the LDR. We have here a voltage divider. A min shunt of 40R means you cant go to 0V for signal. So volume cant go quiet.
I hate to disagree but designed properly:
-One audio channel needs, one series LDR and one shunt LDR voltage divider to go from dead silence to very very loud in a very smooth, increasing sound level;
-Will not burn out at any volume control position;
-One pot for stereo control (SE or Bal);
-NOT IMPOSSIBLE to go below 40R without going over the LDR's 50mW rating, but not even needed to go completely silent
-One audio channel needs, one series LDR and one shunt LDR voltage divider to go from dead silence to very very loud in a very smooth, increasing sound level;
-Will not burn out at any volume control position;
-One pot for stereo control (SE or Bal);
-NOT IMPOSSIBLE to go below 40R without going over the LDR's 50mW rating, but not even needed to go completely silent
Carlmart, I do not know who posted that and have not tried it.
Blues, if you believe the Silonex datasheet..... The datasheet is wrong on nearly every point. Do you know how many times they have changed manufacturing locations in the last three years? The LDRs have changed with every move. They will lose their matching at less than 40R.
If you turn off the series LDRs obviously you will get dead silence. How about 96db speakers, 30x gain and low volume listening while the kids sleep in the next room.. with just one shunt LDR?
Blues, if you believe the Silonex datasheet..... The datasheet is wrong on nearly every point. Do you know how many times they have changed manufacturing locations in the last three years? The LDRs have changed with every move. They will lose their matching at less than 40R.
If you turn off the series LDRs obviously you will get dead silence. How about 96db speakers, 30x gain and low volume listening while the kids sleep in the next room.. with just one shunt LDR?
Carlmart,
Guessing that this guy posted.. well didnt post but came up with it anyway.
http://www.diyaudio.com/forums/members/peter-roodzant.html
Guessing that this guy posted.. well didnt post but came up with it anyway.
http://www.diyaudio.com/forums/members/peter-roodzant.html
Blues, if you believe the Silonex datasheet..... The datasheet is wrong on nearly every point. Do you know how many times they have changed manufacturing locations in the last three years? The LDRs have changed with every move. They will lose their matching at less than 40R.
If you turn off the series LDRs obviously you will get dead silence. How about 96db speakers, 30x gain and low volume listening while the kids sleep in the next room.. with just one shunt LDR?
The Silonex LDR datasheet is correct but its specs and useability can be expanded...I have found majority of the devices to be consistent and of good quality too. They are actually more stable at the lower R values.
Like I've been telling you all this time...the series LDR does NOT have to turn OFF(no voltage/current) to go dead silent, the right combo of series/shunt values will do it regardless of amp gain/speakers sensitivity.
Your kids will have their 10-hour sleep and you can have a long romantic night with your wife...with just one shunt LDR.
<--this smilie is taken
Carl,
That was one of the first designs using a CCS - Peitjers designed this one a few years ago and you can find details about the operation on post #1000. I built one of the Series LDR/shunt LDR ones (George's MkII) complete with the caps across the Leds and one of those cheap china stepped SMD vol controls instead of a pots - works like a charm.
ZenMod developed a similar version with some refinements in the Poor Man's Serbian .... with the k170/j74 jfet buffers, not the B1 cct.
Maximus, Uriah, Blues, also use CCS's to drive the Leds in different ways - different results, but all very good sounding devices, just like the original.
A timely reminder for anyone soldering these devices that they are quite sensitive to over heating so either be quick, or use some of those cheap heat-sink clips - this has been mentioned quite a bit, and is worth repeating.
That was one of the first designs using a CCS - Peitjers designed this one a few years ago and you can find details about the operation on post #1000. I built one of the Series LDR/shunt LDR ones (George's MkII) complete with the caps across the Leds and one of those cheap china stepped SMD vol controls instead of a pots - works like a charm.
ZenMod developed a similar version with some refinements in the Poor Man's Serbian .... with the k170/j74 jfet buffers, not the B1 cct.
Maximus, Uriah, Blues, also use CCS's to drive the Leds in different ways - different results, but all very good sounding devices, just like the original.
A timely reminder for anyone soldering these devices that they are quite sensitive to over heating so either be quick, or use some of those cheap heat-sink clips - this has been mentioned quite a bit, and is worth repeating.
This is why they can go haywire impedance wise or fail completely when hammered at or below 40ohms for any length of time, they heat up just like soldering for them for too long.
I'm fed up with the inquiries I get with how to fix DIY Lightspeeds because they have been pushed too far, the reply is, well, "you can imagine".
As you can see I'm getting less tolerant of the cowboys trying to make a quick buck by pushing or modifying a simple and worked out design. Hell I had guys powering them with 13 plate car batteries. Go figure, Cukoo!!!!!
Cheers Grumpy George
Up above on post #4093, I indicated that Uriah's Lighter note wouldn't drive low impedance loads like my F3 at 9kR.
Well, I hasten to correct this, as you can adjust the LN to any particular load that you want - it's designed to do just that very thing.
Now, I do use a complimentary buffer (2sk170/2sj74 + Shunt Reg Supply) in front of the F3 amp as it allows more flexibility in it's adjustment for the particular sound that I want.
Hello Grumpy!
They ask you 'cos "you be da man"!
You're not going to be too happy with me either, as I'm going to compare a car battery to some cheap China NiMh and Li-ion batteries - only have to recharge it once a year!
Too cold for the surf, yet?
Well, I hasten to correct this, as you can adjust the LN to any particular load that you want - it's designed to do just that very thing.
Now, I do use a complimentary buffer (2sk170/2sj74 + Shunt Reg Supply) in front of the F3 amp as it allows more flexibility in it's adjustment for the particular sound that I want.
Hello Grumpy!
They ask you 'cos "you be da man"!
You're not going to be too happy with me either, as I'm going to compare a car battery to some cheap China NiMh and Li-ion batteries - only have to recharge it once a year!
Too cold for the surf, yet?
6V through an LDR will give a min of 15% distortion. Incredibly audible. LDRs are best used for low signal voltages. Silonex used to post similar figures but are now hard or impossible to find.
If you plan to add active circuitry to bolster an LDR attenuator you may have the decision already made for you by the equipment you are using. For instance a high output impedance source means active circuits before the LDRs and the other way around as well. However if you simply have a need for gain then I have found that its best to put a small amount of gain AFTER the LDRs. To much gain and that 40R shunt comes into play as a higher min volume. To much gain before the LDRs and you get significant distortion. This is LDR dependent, not circuit dependent. The resistive material is simply not meant for high voltages even though it can take them, it will distort.
Normally balanced lines run around 2.5V each. Fine for LDRs and for 2.5 the THD is low. I used to sell balanced sets for people to make balanced Lightspeeds until I had a better understanding of balanced inputs and outputs and how they worked. AndrewT gives us an education several times in this thread. The unfortunate truth is that a circuit that uses only LDRs in balanced line before an amp will not have impressive CMRR. AndrewT was telling us we should be around .01% to .1% matching. He is backed up by Douglas Self. In Self's book Small Signal Audio Design on page 349 we begin to see why. Here is a link to Figure 14.3 which shows the two series resistors (LDRs in our case) on one channel in a balanced line.
Self proves that even a 1% difference in LDRs matching will cause huge drops in CMRR quality.
Self assumes 100R as output of your source and 10k, 20k, 100k, and 1M as possible values for the series LDRs (plain resistors in his case).
He shows in his table an Rout deviation and its results as well and most importantly as showing us the R1 and R2 (series LDRs) deviation and its results on CMRR.
Here is table 14.2 https://picasaweb.google.com/lh/photo/sUzyO5qIdeJ6s-dbfOr6204ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self similates in SPICE and finds that perfect matching of all resistors gives CMRR of -400dB. If one of the input OR output resistors is changed by 1% then CMRR drop to -80dB and a 10% (very easy to achieve with LDRs and much better is easy to, but 1% is a fairy tale) yields 20dB worse at -60dB CMRR. You can see that a 1% deviation (very hard to achieve with LDRs across a resistance curve of infinite points and liable to change significantly if pushed to below 40R) will give you CMRR worse than a properly implemented LM317 (just as an example since we wouldnt use that in an audio signal) and far far far worse than that of NE5532 which has a typical CMRR of -100dB.
Lets imagine using the Lightspeed for a balanced circuit. Lets give ourselves a 100R output impedance because we have a newer CDP or other relatively new source and because it jives nicely with Self's examples. We have a Rin/Rout ratio of about 70 since the Rtot of the Lightspeed is generally about 7k. The result with 1% matching would be even worse than with the 10k series LDRs. Why? Self explains that if we have Rout of 100R and R1 and R2 of 100k with a ratio of 1000 that CMRR improves significantly with 1% deviation of R1 and R2. Unfortunately we can not right now control LDRs well enough to have any kind of matching at 100k or even 50k. So with 100k for R1 and R2 Self finds that CMRR (R1 and R2 deviate by 1%) is -100dB and 80dB with 10% deviation. You can start to follow the table 14.2 now and see where we are going. So what if we had an input resistor value (series LDR) of 1k? I think we can all figure that the CMRR is awful and completely removes any reason to use balanced.
I know a lot of you are proponents of balanced or at least believe that balanced has its virtues, and it DOES. However it has its demons as well. With LDRs, single ended is the way to go. We all only have about 2 or 3 feet max to go from source to pre and pre to amp anyway. So the advantages in the balanced cable are not important to us.
Now I am going to talk about what I have learned about balanced that does NOT relate directly to LDRs but dispels some misconceptions about balanced. What self is explaining at the beginning of this chapter is that an example single ended circuit with a cheap NE5532 has -119dBu noise output over 22Hz - 22kHz. He compares this with what he calls the 'basic balanced input amplifier using a single opamp. With the NE5532 this balanced circuit yields -94dB CMRR. We are not comparing resistors tolerance anymore or their effect on balanced. What Self is telling us is that the single ended circuit vs the balanced circuit wins in low noise. Its the structure of the circuits themselves that create noise. Balanced circuits create more noise than similar single ended circuits. He goes on to step by step change the balanced circuit for improvement after improvement trying to reach the output noise of the single ended circuit.
So here is the single ended circuit he uses as a foundation https://picasaweb.google.com/lh/photo/Avcg_OQoMQ589GmhCIjKHE4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Here is the balanced circuit he begins with
https://picasaweb.google.com/lh/photo/Z3MX2DmtbVsRqAinlxTK8k4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self shows us that we must parallel balanced circuits and add buffers (remember to have that high input impedance for high ratio???) to finally get the balanced circuit to perform at nearly the level of the single ended. Here is that link.
https://picasaweb.google.com/lh/photo/n02N4jVDqv4nC4gsleL9P04ZlfyPjbqFXExOpR4aRG4?feat=directlink
Here is a table, table 14.10 that shows the progression of his experiments and the results. Notice that the first line of data is the single ended implementation.
https://picasaweb.google.com/lh/photo/ucvECoAVxlhuW2akcuAm7E4ZlfyPjbqFXExOpR4aRG4?feat=directlink
So how does this all tie together? Well, your amp (single ended or balanced) is represented by the opamps in these circuits but I guarantee you your amp has nowhere near the low noise or CMRR of an opamp. Not a chance. So the problems you see in these circuits are amplified in your amps. This goes to show that a single ended Lightspeed is THE BEST low noise solution for your home audio. Balanced is truly meant for pro audio. Pro doesnt mean better, pro means used by audio professionals and by that they mean people that make music for a living on a stage that is long enough to require balanced signals to run across it. You have a quieter system at home with single ended. Its a fact, not an audiophile hunch. It not your ears vs someone else' ears. Single ended smokes balanced and LDRs are not a good resistor to stick in a balanced circuit because not only is it near impossible to match at 1% across the entire resistance curve but also they cant be relied upon to stay the same resistance given the same voltage and current. WHY? The LEDs change over time. With a single ended circuit all you need is a balance pot in the LED supply. In a balanced circuit you are screwed unless you can measure the LDRs in circuit and adjust them at every volume step to be exactly equal.
Want an equation? CMRR=20 log ((1+R2+R1)/(4T/100)) where T= the tolerance of the resistor in %. This equation applies to figure 14.5 which is shown at the bottom here
https://picasaweb.google.com/lh/photo/Z3MX2DmtbVsRqAinlxTK8k4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self says that this equation is a worst case scenario because not all resistors will always be at the max end of tolerance. A 1% 1k resistor might be 990 and another might be 998 or 1002 for example.
Want more infomation? Get the Douglas Self book "Small Signal Audio Design" printed by Focal Press in 2010. This is my version of a reference for all of the pictures I linked in this post. Read at least chapter 14. There is similar discussion throughout the book and even I can understand most of it. He really speaks to us DIYers in this book. Its easy to understand.
If you plan to add active circuitry to bolster an LDR attenuator you may have the decision already made for you by the equipment you are using. For instance a high output impedance source means active circuits before the LDRs and the other way around as well. However if you simply have a need for gain then I have found that its best to put a small amount of gain AFTER the LDRs. To much gain and that 40R shunt comes into play as a higher min volume. To much gain before the LDRs and you get significant distortion. This is LDR dependent, not circuit dependent. The resistive material is simply not meant for high voltages even though it can take them, it will distort.
Normally balanced lines run around 2.5V each. Fine for LDRs and for 2.5 the THD is low. I used to sell balanced sets for people to make balanced Lightspeeds until I had a better understanding of balanced inputs and outputs and how they worked. AndrewT gives us an education several times in this thread. The unfortunate truth is that a circuit that uses only LDRs in balanced line before an amp will not have impressive CMRR. AndrewT was telling us we should be around .01% to .1% matching. He is backed up by Douglas Self. In Self's book Small Signal Audio Design on page 349 we begin to see why. Here is a link to Figure 14.3 which shows the two series resistors (LDRs in our case) on one channel in a balanced line.
Self proves that even a 1% difference in LDRs matching will cause huge drops in CMRR quality.
Self assumes 100R as output of your source and 10k, 20k, 100k, and 1M as possible values for the series LDRs (plain resistors in his case).
He shows in his table an Rout deviation and its results as well and most importantly as showing us the R1 and R2 (series LDRs) deviation and its results on CMRR.
Here is table 14.2 https://picasaweb.google.com/lh/photo/sUzyO5qIdeJ6s-dbfOr6204ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self similates in SPICE and finds that perfect matching of all resistors gives CMRR of -400dB. If one of the input OR output resistors is changed by 1% then CMRR drop to -80dB and a 10% (very easy to achieve with LDRs and much better is easy to, but 1% is a fairy tale) yields 20dB worse at -60dB CMRR. You can see that a 1% deviation (very hard to achieve with LDRs across a resistance curve of infinite points and liable to change significantly if pushed to below 40R) will give you CMRR worse than a properly implemented LM317 (just as an example since we wouldnt use that in an audio signal) and far far far worse than that of NE5532 which has a typical CMRR of -100dB.
Lets imagine using the Lightspeed for a balanced circuit. Lets give ourselves a 100R output impedance because we have a newer CDP or other relatively new source and because it jives nicely with Self's examples. We have a Rin/Rout ratio of about 70 since the Rtot of the Lightspeed is generally about 7k. The result with 1% matching would be even worse than with the 10k series LDRs. Why? Self explains that if we have Rout of 100R and R1 and R2 of 100k with a ratio of 1000 that CMRR improves significantly with 1% deviation of R1 and R2. Unfortunately we can not right now control LDRs well enough to have any kind of matching at 100k or even 50k. So with 100k for R1 and R2 Self finds that CMRR (R1 and R2 deviate by 1%) is -100dB and 80dB with 10% deviation. You can start to follow the table 14.2 now and see where we are going. So what if we had an input resistor value (series LDR) of 1k? I think we can all figure that the CMRR is awful and completely removes any reason to use balanced.
I know a lot of you are proponents of balanced or at least believe that balanced has its virtues, and it DOES. However it has its demons as well. With LDRs, single ended is the way to go. We all only have about 2 or 3 feet max to go from source to pre and pre to amp anyway. So the advantages in the balanced cable are not important to us.
Now I am going to talk about what I have learned about balanced that does NOT relate directly to LDRs but dispels some misconceptions about balanced. What self is explaining at the beginning of this chapter is that an example single ended circuit with a cheap NE5532 has -119dBu noise output over 22Hz - 22kHz. He compares this with what he calls the 'basic balanced input amplifier using a single opamp. With the NE5532 this balanced circuit yields -94dB CMRR. We are not comparing resistors tolerance anymore or their effect on balanced. What Self is telling us is that the single ended circuit vs the balanced circuit wins in low noise. Its the structure of the circuits themselves that create noise. Balanced circuits create more noise than similar single ended circuits. He goes on to step by step change the balanced circuit for improvement after improvement trying to reach the output noise of the single ended circuit.
So here is the single ended circuit he uses as a foundation https://picasaweb.google.com/lh/photo/Avcg_OQoMQ589GmhCIjKHE4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Here is the balanced circuit he begins with
https://picasaweb.google.com/lh/photo/Z3MX2DmtbVsRqAinlxTK8k4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self shows us that we must parallel balanced circuits and add buffers (remember to have that high input impedance for high ratio???) to finally get the balanced circuit to perform at nearly the level of the single ended. Here is that link.
https://picasaweb.google.com/lh/photo/n02N4jVDqv4nC4gsleL9P04ZlfyPjbqFXExOpR4aRG4?feat=directlink
Here is a table, table 14.10 that shows the progression of his experiments and the results. Notice that the first line of data is the single ended implementation.
https://picasaweb.google.com/lh/photo/ucvECoAVxlhuW2akcuAm7E4ZlfyPjbqFXExOpR4aRG4?feat=directlink
So how does this all tie together? Well, your amp (single ended or balanced) is represented by the opamps in these circuits but I guarantee you your amp has nowhere near the low noise or CMRR of an opamp. Not a chance. So the problems you see in these circuits are amplified in your amps. This goes to show that a single ended Lightspeed is THE BEST low noise solution for your home audio. Balanced is truly meant for pro audio. Pro doesnt mean better, pro means used by audio professionals and by that they mean people that make music for a living on a stage that is long enough to require balanced signals to run across it. You have a quieter system at home with single ended. Its a fact, not an audiophile hunch. It not your ears vs someone else' ears. Single ended smokes balanced and LDRs are not a good resistor to stick in a balanced circuit because not only is it near impossible to match at 1% across the entire resistance curve but also they cant be relied upon to stay the same resistance given the same voltage and current. WHY? The LEDs change over time. With a single ended circuit all you need is a balance pot in the LED supply. In a balanced circuit you are screwed unless you can measure the LDRs in circuit and adjust them at every volume step to be exactly equal.
Want an equation? CMRR=20 log ((1+R2+R1)/(4T/100)) where T= the tolerance of the resistor in %. This equation applies to figure 14.5 which is shown at the bottom here
https://picasaweb.google.com/lh/photo/Z3MX2DmtbVsRqAinlxTK8k4ZlfyPjbqFXExOpR4aRG4?feat=directlink
Self says that this equation is a worst case scenario because not all resistors will always be at the max end of tolerance. A 1% 1k resistor might be 990 and another might be 998 or 1002 for example.
Want more infomation? Get the Douglas Self book "Small Signal Audio Design" printed by Focal Press in 2010. This is my version of a reference for all of the pictures I linked in this post. Read at least chapter 14. There is similar discussion throughout the book and even I can understand most of it. He really speaks to us DIYers in this book. Its easy to understand.
Last edited:
obviously a poster picked the wrong thread, unintentionally
its been deleted, along with the related response posts
using on topic rules I suggest to leave it at that
moderation
I will have to respectfully disagree with you based on pure experience. Using all the same equipment with the exception of cables, I find single ended noiser.
Stop the music, turn the volume up full and see who much noise you hear between single ended and balanced.
My theory is this. CMRR is great to start with.
1. However the benefits of balanced, even if slightly unbalance, have a chance of cancelling out the noise.
2. Balanced signal is the difference between the 2 signals. If the ground is noise, it does not affect the signal like in single ended.
3. In Single ended systems, both ground and the signal line can introduce different noise for various reasons.
4. In a balance system noise that is introduced to both can be reduced based on CMRR. In a single ended system if the ground holds it own, then the noise introduced on the hot signal line is fed into the system as noise.
Don't know if I make sense to everyone. Since systems are complex, I would suggest 1 ratio does not make a system.
Edit:
I did not change the feed from the pre-amp to the power amp - still balanced. All other things equal, I found less noise with balanced circuits versus unbalanced. This included the source as well on LDR and non-LDR systems. I may have a noisy house.... spooky?
Last edited:
This is no hypothesis. Its proven in not only Doug Self's books but many others.
1: No. This is wrong. It will cancel only to the point that they signals are 180 out of phase with each other and then removing voltages it finds that are equal. If two attenuators, as in balanced, are sending different voltages VL=((r2/r1+r2)*Vs), then whatever is the same will be cancelled. How will the noise end up being presented to the balanced to single ended converter as being the same voltage on both + and - of a balanced line if the attenuators present it as offset? It cant be cancelled if its different, only if its the same. Slightly unbalanced means not cancelled.
The rest of your post really has nothing to do with using LDRs in a balanced signal. I am talking about the inherent noise in the balanced circuit itself and the decrease of noise reduction ability when resistors in the input to a balanced to single ended circuit are not equal.
Assume the balanced to single ended circuit is being fed no signal at all and assume there are no long or short cables attached. There is still more noise in a balanced circuit. The circuit itself is noisier than a single ended circuit. Then add LDRs to the equation and we have an imbalance in our balanced circuit, further reducing the cancellation of noise. Then add a signal and it gets screwed up.
Imagine the single ended signal being split into - and +. 2.5V and -2.5V, lets assume. Also, lets assume that its a sine wave that has a static frequency. Easier to visualize. Then it hits an attenuator on both + and -. 2.5V goes now down to 1V and -2.5V goes to -.95V. So we have 1V and -.95V sine waves. Now in your mind invert that 1V to a -1V sine wave. It overlaps the -.95V sinewave, but not really. Really it doesnt touch that -.95V sinewave. Now, lets inject some noise into the signal, equally on both lines and before the attenuators. The noise is .0XV so we get .0XV and -((.0XV)*.95). The same thing happens to that noise as happens to the signal. It doesnt get removed because it is not identical.
Lets imagine that we inject noise into the ground instead of into the signal. Ground doesnt get inverted. The signal gets inverted. Lets assume .01V of noise. Probably quite a lot there but its for the sake of argument. Since ground is always 0 because the circuit assumes ground to be 0 as ground is really just a point that everything is measured in relation to then all noise injected to ground will still be there later because it is not removed when it is considered to be 0. It is the yardstick by which noise is removed from the signal. So when your balanced to single ended converter converts to single ended the ground will still be noisy. Noise in ground is bad in both single ended and balanced circuits.
I feel pretty confident in my assertions about LDRs unbalancing a balanced signal and making the balanced part of it essentially useless. I will say that its the first time I have considered a noisy ground this much and I am not positive I am right about the noisy ground but it makes sense to me right now.
Another point I certainly debate is your assertion that systems are complex and 1 ratio does not make a system. This is a blanket statement asserting nothing and is in no way any kind of conclusion or even argument. We are, in fact, talking about one part of a system that has an effect on the sound coming out of the system. The rest of the system trusts this small part of the system without prejudice. Everything the rest of the system does is based on what happens right here in the balanced to single ended conversion. In a balanced system the balance between - and + is critical.
In saying that one ratio doesnt make a system you are essentially saying to not take this point to seriously and that is simply, imho, because you dont understand it well enough to take a strong stance on it and defend your argument further. In essence you are saying dont take a balanced attenuator to seriously as it does not make the entire system, yet at the same time you are putting a balanced LDR attenuator on a pedestal and proclaiming it as the pinnacle of attenuation for a system.
In a single ended system at least we can say that an LDR attenuator is not doing something negative to the signal. In fact it works better than all wipered attenuators and doesnt smear like a TVC. In a balanced situation a set of LDRs that are not perfectly matched will 100% of the time degrade the signal more than a single ended. Its math and physics and its not up for debate whether LDRs that are matched to 98% will provide noise reduction. They cant.
1: No. This is wrong. It will cancel only to the point that they signals are 180 out of phase with each other and then removing voltages it finds that are equal. If two attenuators, as in balanced, are sending different voltages VL=((r2/r1+r2)*Vs), then whatever is the same will be cancelled. How will the noise end up being presented to the balanced to single ended converter as being the same voltage on both + and - of a balanced line if the attenuators present it as offset? It cant be cancelled if its different, only if its the same. Slightly unbalanced means not cancelled.
The rest of your post really has nothing to do with using LDRs in a balanced signal. I am talking about the inherent noise in the balanced circuit itself and the decrease of noise reduction ability when resistors in the input to a balanced to single ended circuit are not equal.
Assume the balanced to single ended circuit is being fed no signal at all and assume there are no long or short cables attached. There is still more noise in a balanced circuit. The circuit itself is noisier than a single ended circuit. Then add LDRs to the equation and we have an imbalance in our balanced circuit, further reducing the cancellation of noise. Then add a signal and it gets screwed up.
Imagine the single ended signal being split into - and +. 2.5V and -2.5V, lets assume. Also, lets assume that its a sine wave that has a static frequency. Easier to visualize. Then it hits an attenuator on both + and -. 2.5V goes now down to 1V and -2.5V goes to -.95V. So we have 1V and -.95V sine waves. Now in your mind invert that 1V to a -1V sine wave. It overlaps the -.95V sinewave, but not really. Really it doesnt touch that -.95V sinewave. Now, lets inject some noise into the signal, equally on both lines and before the attenuators. The noise is .0XV so we get .0XV and -((.0XV)*.95). The same thing happens to that noise as happens to the signal. It doesnt get removed because it is not identical.
Lets imagine that we inject noise into the ground instead of into the signal. Ground doesnt get inverted. The signal gets inverted. Lets assume .01V of noise. Probably quite a lot there but its for the sake of argument. Since ground is always 0 because the circuit assumes ground to be 0 as ground is really just a point that everything is measured in relation to then all noise injected to ground will still be there later because it is not removed when it is considered to be 0. It is the yardstick by which noise is removed from the signal. So when your balanced to single ended converter converts to single ended the ground will still be noisy. Noise in ground is bad in both single ended and balanced circuits.
I feel pretty confident in my assertions about LDRs unbalancing a balanced signal and making the balanced part of it essentially useless. I will say that its the first time I have considered a noisy ground this much and I am not positive I am right about the noisy ground but it makes sense to me right now.
Another point I certainly debate is your assertion that systems are complex and 1 ratio does not make a system. This is a blanket statement asserting nothing and is in no way any kind of conclusion or even argument. We are, in fact, talking about one part of a system that has an effect on the sound coming out of the system. The rest of the system trusts this small part of the system without prejudice. Everything the rest of the system does is based on what happens right here in the balanced to single ended conversion. In a balanced system the balance between - and + is critical.
In saying that one ratio doesnt make a system you are essentially saying to not take this point to seriously and that is simply, imho, because you dont understand it well enough to take a strong stance on it and defend your argument further. In essence you are saying dont take a balanced attenuator to seriously as it does not make the entire system, yet at the same time you are putting a balanced LDR attenuator on a pedestal and proclaiming it as the pinnacle of attenuation for a system.
In a single ended system at least we can say that an LDR attenuator is not doing something negative to the signal. In fact it works better than all wipered attenuators and doesnt smear like a TVC. In a balanced situation a set of LDRs that are not perfectly matched will 100% of the time degrade the signal more than a single ended. Its math and physics and its not up for debate whether LDRs that are matched to 98% will provide noise reduction. They cant.
So far the talk is about noise with balanced (XLR) vs Single Ended, no one has mentioned the sound quality difference.
1: As in nearly all cases of balanced output cd sources I have seen, it has been achieved by introducing another (usually cheap NE5532)opamp into the signal path.
2: You may gain s/n ratio with 5mt or more interconnects, but lose out because you are listening to another opamp in the signal path.
3: Also from what I've seen with many high end poweramps the balanced input sometimes adds an opamp also into the signal path then converts it to single ended, before the signal is fed to the original rca single ended discrete input inside the amp. The Classe DR250 does this, so to many others, and it defiantly sounds better using the single ended input because it is all discrete from RCA input to the speaker output, whereas the balanced is going through a cheap NE5532 opamp first.
4: So in the above cases you are listening to an extra 2 x cheap opamps per channel in the signal path in the the system
5: So what would you rather listen to 5mt plus balanced interconnects and cheap opamps in the signal path, or 1mt signal ended interconnects and no opamps, in which no noise can be heard anyway.
KISS KISS KISS
Cheers George
1: As in nearly all cases of balanced output cd sources I have seen, it has been achieved by introducing another (usually cheap NE5532)opamp into the signal path.
2: You may gain s/n ratio with 5mt or more interconnects, but lose out because you are listening to another opamp in the signal path.
3: Also from what I've seen with many high end poweramps the balanced input sometimes adds an opamp also into the signal path then converts it to single ended, before the signal is fed to the original rca single ended discrete input inside the amp. The Classe DR250 does this, so to many others, and it defiantly sounds better using the single ended input because it is all discrete from RCA input to the speaker output, whereas the balanced is going through a cheap NE5532 opamp first.
4: So in the above cases you are listening to an extra 2 x cheap opamps per channel in the signal path in the the system
5: So what would you rather listen to 5mt plus balanced interconnects and cheap opamps in the signal path, or 1mt signal ended interconnects and no opamps, in which no noise can be heard anyway.
KISS KISS KISS
Cheers George
Last edited:
Additional comments about this balanced/unbalanced questions.
Balanced is mainly justified in long connections or in high RFI situations.
You certainly need a differential driver and a differential receiver to balance a connection, and that adds a lot of parts in the middle, particularly capacitors (which I personally try to avoid).
One way to solve that and stay unbalanced is to use high level buffers, usually video types, to drive long interconnects. This is usually not the case in most noisy unbalanced situations, or the earthing is poorly implemented.
In any case, most balanced amplifiers I've seen go back to single-ended before amplifying, so why not add the LDR attenuation there?
Balanced is mainly justified in long connections or in high RFI situations.
You certainly need a differential driver and a differential receiver to balance a connection, and that adds a lot of parts in the middle, particularly capacitors (which I personally try to avoid).
One way to solve that and stay unbalanced is to use high level buffers, usually video types, to drive long interconnects. This is usually not the case in most noisy unbalanced situations, or the earthing is poorly implemented.
In any case, most balanced amplifiers I've seen go back to single-ended before amplifying, so why not add the LDR attenuation there?
You hit the nail on the head Carl. If you must use balanced then you use LDRs after the balanced to single ended conversion has happened. Then we have no issues with LDRs not being in ultra close tolerance with each other.
ps I have to brag a little. My friend Vlad is giving me a pair of his Eargasm Horn speakers. I am beside myself with his generosity and cant wait to complete building them. He has them about halfway finished. Here are some pics. I cant believe my good luck. Hopefully I can get them looking like these pics http://forum.audiogon.com/cgi-bin/frr.pl?rspkr&1132019363&read&3&4&
Uriah
ps I have to brag a little. My friend Vlad is giving me a pair of his Eargasm Horn speakers. I am beside myself with his generosity and cant wait to complete building them. He has them about halfway finished. Here are some pics. I cant believe my good luck. Hopefully I can get them looking like these pics http://forum.audiogon.com/cgi-bin/frr.pl?rspkr&1132019363&read&3&4&
Uriah
Last edited:
George,
Several pages ago you proposed a system with several LS (Lightspeed) attenuators to control several inputs. As I will be beginning a project to work as 6-channel control system, I wonder how would you handle the remote level setting and input switching.
I already know some Alps and Panasonic (there might be others) 6-gang motorized pots that should be remoteable, but I have no exact clue on how to do that for the signal switching. Of course I do not want the audio to go through any switching chip, and your relay based signal switching might be the answer.
Is there anyone presently working on how to remote a control system such as the one I am thinking on?
Carlos
Several pages ago you proposed a system with several LS (Lightspeed) attenuators to control several inputs. As I will be beginning a project to work as 6-channel control system, I wonder how would you handle the remote level setting and input switching.
I already know some Alps and Panasonic (there might be others) 6-gang motorized pots that should be remoteable, but I have no exact clue on how to do that for the signal switching. Of course I do not want the audio to go through any switching chip, and your relay based signal switching might be the answer.
Is there anyone presently working on how to remote a control system such as the one I am thinking on?
Carlos
You mean for selecting the LDRs?
OK.
I would not think of producing anything like this to sell.