That's fine, but I don't see how the superb sound I'm getting from the balanced Lightspeed is evidence of wasted resources. Thanks for the reading suggestion. However, I have read much of the literature on balanced operation -Whitlock's papers, Rane, Self, et al.
Folks..
My differential lightspeed is up for sale on ebay in the UK if anyone is interested.
"Lightspeed" LDR Differential Preamp on eBay (end time 18-Nov-09 09:01:34 GMT)
Works a treat. Happy to post world wide. This isn't a commercial venture, wanted to make my DIY project available to someone who would make more use of it than myself as I'm downsizing my hifi now :-(
Cheers..
My differential lightspeed is up for sale on ebay in the UK if anyone is interested.
"Lightspeed" LDR Differential Preamp on eBay (end time 18-Nov-09 09:01:34 GMT)
Works a treat. Happy to post world wide. This isn't a commercial venture, wanted to make my DIY project available to someone who would make more use of it than myself as I'm downsizing my hifi now :-(
Cheers..
Wow. You put a lot of work into that Lightspeed. Nice job.
Downsizing to what if you cant have this!?
Uriah
Downsizing to what if you cant have this!?
Uriah
Balanced Shunt with fixed resistor
AndrewT,
Your idea of using a fixed series resistor with a LDR shunt between the balanced signal phases, is an interesting way to get perfect matching for balanced. In this balanced configuration, would the range of resistance values needed in the shunt LDR be the same as those called out in a typical shunt calculator such as Neville?
50-Step Shunt Attenuator Resistor Calculator - Neville Roberts
Assuming the calculator is correct, if one were to build a 50K preamp volume control using a 50K series resistor, and target a useful operating range say from -60db to -10db, the shunt LDR would need to swing from 50R to 23K. Is this feasible? Any thoughts on how input Z of the volume control would vary throughout this range?
Thanks,
Dave
AndrewT,
Your idea of using a fixed series resistor with a LDR shunt between the balanced signal phases, is an interesting way to get perfect matching for balanced. In this balanced configuration, would the range of resistance values needed in the shunt LDR be the same as those called out in a typical shunt calculator such as Neville?
50-Step Shunt Attenuator Resistor Calculator - Neville Roberts
Assuming the calculator is correct, if one were to build a 50K preamp volume control using a 50K series resistor, and target a useful operating range say from -60db to -10db, the shunt LDR would need to swing from 50R to 23K. Is this feasible? Any thoughts on how input Z of the volume control would vary throughout this range?
Thanks,
Dave
the matched 50k series resistors can be switched.
A pair of 220k +-0.01% bypassed by a pair of 10k+-0.01% would usefully increase the range. If the source can really drive low Rin then the 10k bypass could be reduced to 1k to 2k.
A double pole single throw switch does this and introduces just one extra contact in parallel to the 220k of each signal route.
Using the calculator:- reduce the actual LDR resistance by half and enter that half value into the calculator.
Remember to allow for a load impedance (Rin) in parallel to the LDR.
A pair of 220k +-0.01% bypassed by a pair of 10k+-0.01% would usefully increase the range. If the source can really drive low Rin then the 10k bypass could be reduced to 1k to 2k.
A double pole single throw switch does this and introduces just one extra contact in parallel to the 220k of each signal route.
Using the calculator:- reduce the actual LDR resistance by half and enter that half value into the calculator.
Remember to allow for a load impedance (Rin) in parallel to the LDR.
Last edited:
I haven't been very active on diyaudo.com for about 18 months (I finally got "a real job" again, doing electronic warfare stuff.). But I am still subscribed to this thread and occasionally get an email about a reply here. Seeing AndrewT's name always piques my interest.
If you find my earlier posts in this thread, you can see that LTSpice can be used to plot any Z versus almost anything else in the circuit.
It's been a long time, but I think that back then I was investigating ways to control and improve the i/o Zs' excursions, etc. If I recall correctly, I thought it was interesting and significant but no one else seemed very interested. So maybe I was off base, at the time.
If I didn't post the spice models for the LDRs, etc, and someone wants them, let me know and I'll try to dig them out.
Tom Gootee
Edit: See my post, #1045 in this thread, to download the LTSpice models and circuits. But also read the next few pages of posts, for corrections etc.
Last edited:
The technique of controlling volume by a shunt resistor between balanced signals has been used in four designs that I know of. One was shown on Erno Borbely's site in a balanced line stage design which no longer appears there but which might be in one of the papers available. The other designs are vacuum tube line stages. Morgan Jones' design which appeared in Electronics World April 1996 used a single 6SN7, and he said of the variable shunt --'This has the disadvantage of a high output resistance when set for a sensible input resistance and will cause hf loss if ignored.' He obtains hf performance by using feedforward capacitance from grid to opposite cathode.
One of Allen Wright's vacuum tube line stage designs used a variable shunt volume control, but in his preamp cookbook he states -'Any system that changes the actual loading on the phono stage as the gain is changed will change the RIAA eq. -so tricky deals like the following are out!' The 'following' refers to a schematic of the variable shunt balanced volume control.
Richard Sears also used a variable shunt on the inputs of his balanced line stage which is a 6SN7 followed by an output transformer. PP Line Stage He says --'The choice of series resistance value needs a little thought. First, it must be high enough in value such that when the shunt is approaching zero, the load presented is acceptable to all sources which are to be used. Second, it must be small enough in value that the HF roll-off resulting by it acting in combination with the Miller capacitance of the 6SN7 is acceptable.'
In post #1513, rocekthebox stated 'I'm running a balanced circuit - LDR in series with each signal line & an LDR shunt between the two signals. I bought 25 LDR's weeded out 2 pairs that were very closely matched for the series LDRS & 1 pair matched for the shunt. Sound is very transparent, highly recommend this circuit.'
Perhaps it's the one he has for sale now. Earlier this year I asked here about this method but got no useful reply.
George stated more than once that his second version of Lightspeed which uses all LDRs sounds better than the first version which used LDRs in combination with fixed resistors.
Not being an engineer, and having no test gear or simulators, I took the simple route and built a balanced lightspeed using all LDRs just to find out how it sounds. The engineering explanations make precisely matched halves of a balanced circuit desirable, but in my listening experience the advantages of using all LDRs outweigh the potential disadvantages of mismatching.
-Terry C.
balanced is not there to make it sound better. Balanced is there to reject noise more effectively.
If the improved sound quality of the series/shunt LDR is more important then don't do balanced.
Using just slightly mismatched series LDRs will so degrade the rejection capabilities of balanced that all the expense of going to balanced is completely wasted.
U Shunt attenuator for balanced
My guess so far about LDRs for an interstage balanced volume control in an active preamp:
Fixed series resistor/LDR shunt arrangement could work well (that is, present a somewhat stable input Z to the preceding stage, provide perfect CMRR, and be flexibly built to typical 25K, 50K, or 100K impedance values), if operated strictly within a narrow range of attenuation. Outside this range fluctuations of input Z will affect preceding phono section.
LDR series/LDR shunt arrangement could work well across a wider range of attenuation-- provided that matched LDRs track within 5%. 5% is probably about the same tolerance as a typical series attenuator using 1% resistors-- whose resistors may collectively deviate up to 5%. The LDR/LDR volume control would probably be limited to about 25K impedance-- above which tracking deviates for even closely matched LDRs.
AndrewT, while it is nice to have perfect CMRR through a volume control, in practice in a balanced tube preamp deviations between even closely matched tube triode sections will be considerable-- probably comparable to tracking deviations in a balanced LDR/LDR volume control.
My guess so far about LDRs for an interstage balanced volume control in an active preamp:
Fixed series resistor/LDR shunt arrangement could work well (that is, present a somewhat stable input Z to the preceding stage, provide perfect CMRR, and be flexibly built to typical 25K, 50K, or 100K impedance values), if operated strictly within a narrow range of attenuation. Outside this range fluctuations of input Z will affect preceding phono section.
LDR series/LDR shunt arrangement could work well across a wider range of attenuation-- provided that matched LDRs track within 5%. 5% is probably about the same tolerance as a typical series attenuator using 1% resistors-- whose resistors may collectively deviate up to 5%. The LDR/LDR volume control would probably be limited to about 25K impedance-- above which tracking deviates for even closely matched LDRs.
AndrewT, while it is nice to have perfect CMRR through a volume control, in practice in a balanced tube preamp deviations between even closely matched tube triode sections will be considerable-- probably comparable to tracking deviations in a balanced LDR/LDR volume control.
Wish I could add valuable information to the balanced debate. It is an interesting read.
I would say this. I would not use a regular resistor along with a shunt LDR. I would use an LDR in series and one in shunt. I can tell that removing contacts from the circuit indeed improves the sound quality but the difference is NOTHING like the difference between regular resistors and LDRs. The sound is better in the Lightspeed because of the LDRs, IMHO. If you wanted to use a fixed series with a moving shunt why not just set your series LDR at 25k if thats the value you want and then use a pair of well matched LDRs for the moving shunt?
Uriah
I would say this. I would not use a regular resistor along with a shunt LDR. I would use an LDR in series and one in shunt. I can tell that removing contacts from the circuit indeed improves the sound quality but the difference is NOTHING like the difference between regular resistors and LDRs. The sound is better in the Lightspeed because of the LDRs, IMHO. If you wanted to use a fixed series with a moving shunt why not just set your series LDR at 25k if thats the value you want and then use a pair of well matched LDRs for the moving shunt?
Uriah
Can you please describe what the degradation sounds like? What can I expect to hear in my sound system as a result?
Hello. I have recently discovered this really interesting Lightspeed attenuator and I was wondering whether I could use it in my system or not. Reason of doubt is that I am reading the CD player should have an output impedance of <200 ohms. I've measured mine and it has 500 ohms output impedance. What would happen if I tried it with a Lightspeed:
a) and a future diy Firstwatt F2 (50 kohms input impedance), or
b) in conjunction with a future diy Pass B1 Salas version with symmetrical psu and either the abovementioned F2 or a pair of Mackie active monitors (10 kohms unbalanced input impedance)?
Is there any risk for the CD player? Should I expect bad sound?
Thank you in advance.
a) and a future diy Firstwatt F2 (50 kohms input impedance), or
b) in conjunction with a future diy Pass B1 Salas version with symmetrical psu and either the abovementioned F2 or a pair of Mackie active monitors (10 kohms unbalanced input impedance)?
Is there any risk for the CD player? Should I expect bad sound?
Thank you in advance.
I am not sure about the CDP.
I use mine with a Salas B1. Works great. I have driven an F5 with the original Lightspeed circuit minus the B1. That worked as well. I dont think it would be a good idea with your monitors without the B1. With the B1 it will be fine as the B1 has an output impedance of 260Ohms. The Salas version anyway, not sure about original Pass version.
Either way, the less than 200Ohms is probably based on the fact that most Lightspeeds end up with approx 7k impedance. What you could do is boost up the Lightspeed's impedance with a very small value series resistor placed between the power supply's positive output and the control pot for the Lightspeed. Pic a 100R or so trimmer. Start at 0R and slowly increase. Probably you wont need higher than 30R. While you are changing this you should be measuring the resistance of one of the Lightspeed channels. When it reaches a value you are comfortable with then you stop turning the trimmer and measure the trimmer. Go buy a resistor of that value and replace the trimmer. Now you have a Lightspeed that should work with your CDP. Now, over 25k each LDR will start to find its own path of resistance. Meaning its curve will probably start being pretty different from its other LDR matches. Sometimes you will get LDRs that will match well up to 35k and maybe 40k but higher than that is a dream that not often comes true. 25k, though, is quite possible and even more so if you are willing to use an extra pot for channel balance since you might find a match that is great up to 3/4 of the resistance you need and then starts to lose channel balance. A pot on the loud channel will allow you to ignore their poor matching at higher resistance.
Uriah
edit: I should add that I have matched thousands of LDRs and when I say its possible to get LDRs that will match well over 25k I only mean its possible, not that it happens often at all. I have had probably 5 sets that actually did. At about 25k, out of thousands of LDRs, I have probably had 10-15 sets. Lots of them will match up to 20k though and a huge amount of them up to 15k. I would not lose hope, you can tame your Lightspeed to be kind to both your CDP and your amp.
I use mine with a Salas B1. Works great. I have driven an F5 with the original Lightspeed circuit minus the B1. That worked as well. I dont think it would be a good idea with your monitors without the B1. With the B1 it will be fine as the B1 has an output impedance of 260Ohms. The Salas version anyway, not sure about original Pass version.
Either way, the less than 200Ohms is probably based on the fact that most Lightspeeds end up with approx 7k impedance. What you could do is boost up the Lightspeed's impedance with a very small value series resistor placed between the power supply's positive output and the control pot for the Lightspeed. Pic a 100R or so trimmer. Start at 0R and slowly increase. Probably you wont need higher than 30R. While you are changing this you should be measuring the resistance of one of the Lightspeed channels. When it reaches a value you are comfortable with then you stop turning the trimmer and measure the trimmer. Go buy a resistor of that value and replace the trimmer. Now you have a Lightspeed that should work with your CDP. Now, over 25k each LDR will start to find its own path of resistance. Meaning its curve will probably start being pretty different from its other LDR matches. Sometimes you will get LDRs that will match well up to 35k and maybe 40k but higher than that is a dream that not often comes true. 25k, though, is quite possible and even more so if you are willing to use an extra pot for channel balance since you might find a match that is great up to 3/4 of the resistance you need and then starts to lose channel balance. A pot on the loud channel will allow you to ignore their poor matching at higher resistance.
Uriah
edit: I should add that I have matched thousands of LDRs and when I say its possible to get LDRs that will match well over 25k I only mean its possible, not that it happens often at all. I have had probably 5 sets that actually did. At about 25k, out of thousands of LDRs, I have probably had 10-15 sets. Lots of them will match up to 20k though and a huge amount of them up to 15k. I would not lose hope, you can tame your Lightspeed to be kind to both your CDP and your amp.
Last edited:
Thank you both. I measured the CDP output impedance by the method described on the Sowter website here, scroll down to number 10. I got 500 ohms at 100 Hz, 1 kHz and 10 kHz, both channels measured the same on all frequencies.
Uriah, do you suggest getting a 25k+ ohms impedance on the Lightspeed in order to help the CDP, the monitors or both?
A detailed description of the adventures I had with a 50kohms pot in my system, wired normally and then in reverse position (!) (IN becoming OUT and OUT becoming IN) can be found here. Perhaps it would give you a clue on what should I expect from a Lightspeed in this system.
I'm asking all this because I'd like to know if :
1. the Lightspeed and my CDP are compatible at all
2. the B1 buffer is really mandatory ("try without it and see" is not such a good approach for me because I have to import almost all the components and I'd like to avoid paying the $35-40 shipping twice.
Thanks again.
Uriah, do you suggest getting a 25k+ ohms impedance on the Lightspeed in order to help the CDP, the monitors or both?
A detailed description of the adventures I had with a 50kohms pot in my system, wired normally and then in reverse position (!) (IN becoming OUT and OUT becoming IN) can be found here. Perhaps it would give you a clue on what should I expect from a Lightspeed in this system.
I'm asking all this because I'd like to know if :
1. the Lightspeed and my CDP are compatible at all
2. the B1 buffer is really mandatory ("try without it and see" is not such a good approach for me because I have to import almost all the components and I'd like to avoid paying the $35-40 shipping twice.
Thanks again.
Correct the output of your CDP was measured the right way, so you have 500ohms z out this I think will need a low output Z buffer, as I have found 200ohms to be much the max into the Lightspeed and poweramp.
You will not harm anything by playing it this way, you may loose a bit of dynamic swing is all, as your 500ohm output is now trying to drive both the z input of Lightspeed and the Z input of the power amp.
Say you have 7kohm Lightspeed and 47kohm power amp these two together with interconnects will represent a load to the CDP of 6kohm which is too low in my opinion for the 500ohm of the CDP to drive without loosing something.
Cheers George
Last edited:
No I am not recommending 25k I am just saying that you can increase the max resistance of the LDRs in your Lightspeed to about 25k before they will stop tracking in a similar fashion. Basically you will start to get channel imbalance and it might even start to get severe after around 25k. The reason is that the higher the resistance the more even a mV or less will very dramatically change the resistance and the LEDs in the Lightspeed all drop almost the same voltage but not exactly so at this point the difference in their voltage drop starts to make a very large impact on their resistance.
This isnt answering your question directly but what I am telling you is that there IS a resistance/impedance at which your CDP and amp will cooperate via the Lightspeed and you can dial it in. What point that is I dont know and you are not going to kill your equipment with a little Lightspeed experimentation. There is a formula for figuring your optimum impedance but I am sorry I dont know it.
Your powered speakers however I think will need that buffer.
Uriah
Thank you for your answers. Uriah, I've sent you an email for some LDRs.
George, by an additional buffer you mean I should build 2 B1 buffers and place them one before and one after the Lightspeed? Uriah's suggestion seems so much simpler and I'll certainly try it, although I understand there's no guarantee that <25kohms will be high enough for the CDP.
George, by an additional buffer you mean I should build 2 B1 buffers and place them one before and one after the Lightspeed? Uriah's suggestion seems so much simpler and I'll certainly try it, although I understand there's no guarantee that <25kohms will be high enough for the CDP.
Oh, I see, so I could use CDP --> very short ICs --> B1 buffer --> Lightspeed --> 2m or 0.5m ICs --> active monitors or Firstwatt F2, right? Great, then, thank you.
Edit: Oops, this really shows how little I know. I've just posted the above only to see that in in this case I'd run into another problem, the Lightspeed would not like the 10kohms input of my monitors, right? Then what's the solution, can you spell it for me? Sorry for being such a... hmmm... beginner!
Edit: Oops, this really shows how little I know. I've just posted the above only to see that in in this case I'd run into another problem, the Lightspeed would not like the 10kohms input of my monitors, right? Then what's the solution, can you spell it for me? Sorry for being such a... hmmm... beginner!
Last edited: