Hi Arne,
have you tried it so far? What optocoupler do you use?
I don't believe that it works (figure 14) cause there is no feedback of the "real" resistance! Furthermore the distortion curve is inferior to the configuration of figure 7 and figure 9!
have you tried it so far? What optocoupler do you use?
I don't believe that it works (figure 14) cause there is no feedback of the "real" resistance! Furthermore the distortion curve is inferior to the configuration of figure 7 and figure 9!
Hi all,
I am just confused about the unpredictable resistance behaviour of the SR2S with a constant voltage! Is it only a temperature effect? I think not cause the longer they run the more voltage I have to give to get the same resistance.
In order to built an attenuator with exact 1dB steps I need that perfect resistance behaviour. One idea was to mount the couplers onto a peltier element and heat them up to constant temperature of 40°C controlled by a µC. But I think the temperature issue isn't the only thing which is responsible for the resistance drift!
I am just confused about the unpredictable resistance behaviour of the SR2S with a constant voltage! Is it only a temperature effect? I think not cause the longer they run the more voltage I have to give to get the same resistance.
In order to built an attenuator with exact 1dB steps I need that perfect resistance behaviour. One idea was to mount the couplers onto a peltier element and heat them up to constant temperature of 40°C controlled by a µC. But I think the temperature issue isn't the only thing which is responsible for the resistance drift!
I connect a power resistor to a mounting plate (fets clipped down with TO92 clips) and feed it dc volts via a simple temp switch - I think my setup is about 52*C - the same idea for the "crystal oven" for the clock crystal (45*C) - available from farnells, etc.
Not sure it's necessary - the little LDRs wrm themselves up - just mount them together.
Not sure it's necessary - the little LDRs wrm themselves up - just mount them together.
jameshillj said:
do you have the circuit with the parts? is it a control circuit with constant temerature?
Do you have now constant voltage/resistance ratios of the optocouplers?
Sorry, Thomas, a misunderstanding.
The basic device as per Georges original design just sits and warms itself - I did mount the LDRs together and they don't change noticeably when listening to the balance or stereo image at different levels and various music. Very clean and dynamic.
The second one is Nelson's B1 unit - I'm poking around with some different variations of this - not sure if it'll improve anything, but changes the sound a bit and is rather fun to play with simple ccts like this.
The temp switch is a simple little bimetal button that simply screws/ glues onto the "heatsink" that's common to all the components - nothing complicated - just out of your junk box - it makes quite a difference with my simple Jboz "thingie", so tried it out here, but it's there for the fets, not the LDRs.
I'm a bit puzzled about this "unpredictable resistance behaviour" or "resistance drift". I've not seen anything like that (yet!)
I presume that you've built a good quality power supply - can't get consistent/good results without one. A tip here - don't use cheap, block bridges or ordinary diodes - get the" fastest" diodes you can find like BAV10s,1N5819, BD4148 (SMD), etc - it makes a HUGE difference - sounds ridiculous, but ... (see the" Trr" on spec sheets as a guide)
There is a rather good variation using constant current supply that has been said to get excellent results that may suit your particular control requirements without delving into complicated chip designs, and someone set up his "Twisted Pear" Joshua Tree style vol control to drive the LDRs and reported better than 1dB increments.
A simple 23 position rotary switch with a high/low second switch will give a 46dB range of voltage/current control - possibly, you may have to directly measure the increments but not very difficult, just tiresome.
There is/was quite a bit of work done by some enterprising guys from Italy/France on developing a processor controlled Lightspeed that may have more precise info about your query. It was mentioned a while back here in this thread.
Hope this may be of assistance .....
The basic device as per Georges original design just sits and warms itself - I did mount the LDRs together and they don't change noticeably when listening to the balance or stereo image at different levels and various music. Very clean and dynamic.
The second one is Nelson's B1 unit - I'm poking around with some different variations of this - not sure if it'll improve anything, but changes the sound a bit and is rather fun to play with simple ccts like this.
The temp switch is a simple little bimetal button that simply screws/ glues onto the "heatsink" that's common to all the components - nothing complicated - just out of your junk box - it makes quite a difference with my simple Jboz "thingie", so tried it out here, but it's there for the fets, not the LDRs.
I'm a bit puzzled about this "unpredictable resistance behaviour" or "resistance drift". I've not seen anything like that (yet!)
I presume that you've built a good quality power supply - can't get consistent/good results without one. A tip here - don't use cheap, block bridges or ordinary diodes - get the" fastest" diodes you can find like BAV10s,1N5819, BD4148 (SMD), etc - it makes a HUGE difference - sounds ridiculous, but ... (see the" Trr" on spec sheets as a guide)
There is a rather good variation using constant current supply that has been said to get excellent results that may suit your particular control requirements without delving into complicated chip designs, and someone set up his "Twisted Pear" Joshua Tree style vol control to drive the LDRs and reported better than 1dB increments.
A simple 23 position rotary switch with a high/low second switch will give a 46dB range of voltage/current control - possibly, you may have to directly measure the increments but not very difficult, just tiresome.
There is/was quite a bit of work done by some enterprising guys from Italy/France on developing a processor controlled Lightspeed that may have more precise info about your query. It was mentioned a while back here in this thread.
Hope this may be of assistance .....
jameshillj said:
This I also have only seen happen a couple of times in all the 35 years of building these things, and that was when we ran the LDR's up higher than their specs, in this case the 25mA max of NSL-32SR2S.
I can see the temptation to push them this hard when in mkI configuration, because of the lack of volume (min to max) range you get, when compared to the mkII.
Cheers George
I just measured my MKII. Resistance drift doesn't vary the absolut dB L/R difference. It stays the same.
The channel imbalance varies between 0.08 db at -35db and -30dB, 0.7db at -25dB, 0.64dB at -20dB and 0.1dB at -10dB. I think I can adjust the important area between -30 an -15dB with the 1k pot.
What I recognized: At -20dB the input impedance of left channel is about 10 kOhm and the right channel has 7.8 kOhm. That is not good I think!
I am just researching the dependancies between temperature drift and voltage drop.
P.S. Yesterday I killed an optocoupler with 5V instead of 2.5V! It works but the smallest resistance is now 70 Ohm and before it was 25 Ohm.
I have just one SR2S which is at 23 Ohm, the rest between 27 and 32! What is the advantage against SR3?
The channel imbalance varies between 0.08 db at -35db and -30dB, 0.7db at -25dB, 0.64dB at -20dB and 0.1dB at -10dB. I think I can adjust the important area between -30 an -15dB with the 1k pot.
What I recognized: At -20dB the input impedance of left channel is about 10 kOhm and the right channel has 7.8 kOhm. That is not good I think!
I am just researching the dependancies between temperature drift and voltage drop.
P.S. Yesterday I killed an optocoupler with 5V instead of 2.5V! It works but the smallest resistance is now 70 Ohm and before it was 25 Ohm.
I have just one SR2S which is at 23 Ohm, the rest between 27 and 32! What is the advantage against SR3?
Tolu said:
These small variations in measurements are really insignificant to sound quality, I never push my ldr's above 20mA, in MkII configuration this is fine for volume range, but I can see that in MkI configuration that won't be hence the need to push the ldr's harder and this is what can make them loose or change their electrical parameters.
The SR2 has a max of 1meg ohm, where the SR3 25meg ohm is much more, another reason to use the SR2's as this greatly helps matching up, and consistancy over time.
http://www.silonex.com/audiohm/index.html
Cheers George
Hi James, hi George,
thanks for your hints. I am just sitting in my living room and enjoying listening with MKII. It works almost perfect with my power amp that has 47k input impedance. I have 2 short cinch cables (1.5 ft) between CD player, LS and amp.
I am listening to a best of CD of Billy Joel and it sounds so real and life like, great! My 2k Euro preamp sees no sun against it.
As I wrote some posts ahead the input impedance between left and right channel varies between 7.5k and 10k.
I have spent many hours the last two weeks in order to understand the behaviour of the 32SR2S parts but I think I have to give up. I have 3 of C grade, 3 of D, and 2 F left from my order. The 4 best matching are in the MKII.
I made big tables with 16 bit values for my DAC to get specific resistance values of different optocouplers. But if room temperature rises or lowers a little bit the resistance curve has changed again and I didn't find any law how this happens.
My design goal is to make the LS remote controlled with -1 dB steps but isn't as easy as it seems. It wouldn't kick me if it is -1.36 or 0.89 dB, important is the channel tracking!
My last attempt will be the constant heating of the couplers with about 40°C. If this fails...
I have to stand up for volume control 
thanks for your hints. I am just sitting in my living room and enjoying listening with MKII. It works almost perfect with my power amp that has 47k input impedance. I have 2 short cinch cables (1.5 ft) between CD player, LS and amp.
I am listening to a best of CD of Billy Joel and it sounds so real and life like, great! My 2k Euro preamp sees no sun against it.
As I wrote some posts ahead the input impedance between left and right channel varies between 7.5k and 10k.
I have spent many hours the last two weeks in order to understand the behaviour of the 32SR2S parts but I think I have to give up. I have 3 of C grade, 3 of D, and 2 F left from my order. The 4 best matching are in the MKII.
I made big tables with 16 bit values for my DAC to get specific resistance values of different optocouplers. But if room temperature rises or lowers a little bit the resistance curve has changed again and I didn't find any law how this happens.
My design goal is to make the LS remote controlled with -1 dB steps but isn't as easy as it seems. It wouldn't kick me if it is -1.36 or 0.89 dB, important is the channel tracking!
My last attempt will be the constant heating of the couplers with about 40°C. If this fails...
I have to stand up for volume control Hey Thomas,
This is obviously a bit of a problem for you so just get a "crystal oven" from Farnells to heat all the LDRs to the same temp - they're available at different temp settings - they're built to keep the crystals in the clocks at same temp (and freq), cost little and easy to arrange - my master clock crystal runs at a constant 45*C, not sure if it makes much difference to operation but sounds good when talking about it (!) and does elimenate a possible problem - it's quite common in pro-gear.
Heating the fets(and especially input jFets) is well worth doing.
This is obviously a bit of a problem for you so just get a "crystal oven" from Farnells to heat all the LDRs to the same temp - they're available at different temp settings - they're built to keep the crystals in the clocks at same temp (and freq), cost little and easy to arrange - my master clock crystal runs at a constant 45*C, not sure if it makes much difference to operation but sounds good when talking about it (!) and does elimenate a possible problem - it's quite common in pro-gear.
Heating the fets(and especially input jFets) is well worth doing.
jameshillj said:
Yeah - it may help reduce dc drift but you could just place them close together and then glue them together - same with the LDR's . . . george pots them in wax, you could use epoxy resin, two part polyurethane or any number of any different methods that are much, much simpler than active temperature regulation.
just a thought . . .
remote?
Sorry Thomas,
Missed that last part about having to use manual vol control - I still use the "get -off your- a**" method but am looking at one of those remote controlled motorised pot kits and must also check out one of those universal control gadgets to see if they actually do work!!
"Twisted Pair" have a type of remote control sorted out, I think - might be worth a look ....
Sorry Thomas,
Missed that last part about having to use manual vol control - I still use the "get -off your- a**" method but am looking at one of those remote controlled motorised pot kits and must also check out one of those universal control gadgets to see if they actually do work!!
"Twisted Pair" have a type of remote control sorted out, I think - might be worth a look ....
Hi James,
RC with a 100k log motor pot would be the simplest and cheapest way. I did this with an 2$ atmel µC and one chip for an blue alps. But there is no difference in the matching process in comparison to the manual MKII version. My goal is digital controlled perfect channel matching.
THe crystal oven is a little bit oversized I think. A cheap 11W ceramic resistance, controlled with a temperature sensor and an atmel avr would do it for 5 bucks.
RC with a 100k log motor pot would be the simplest and cheapest way. I did this with an 2$ atmel µC and one chip for an blue alps. But there is no difference in the matching process in comparison to the manual MKII version. My goal is digital controlled perfect channel matching.
THe crystal oven is a little bit oversized I think. A cheap 11W ceramic resistance, controlled with a temperature sensor and an atmel avr would do it for 5 bucks.
Yeah Thomas,
I do tend to go overboard sometimes although the crystal oven was one of those simple things to try - the simple jfet heaters I use all the time, as they sound better when hot.
It's never really occured to me before that the L - R channel balance of the Lightspeed would be at all affected by input impedance variations caused by temp rise (or my buffer heating) but your systematic investigation is pretty conclusive and it's a bit of a surprise.
Did I get this right?
"there is perhaps over 1dB attenuation variation between the two channels at different points and this also varies with temp changes .... but it doesn't affect the sound of the device, just the balance?"
If so, then a controlled heater should keep the variation together, at least.
Incidently, your amp load is a bit low at 47kR, might increase this to maybe 70kR + - not sure, but might change sound a bit.
I do tend to go overboard sometimes although the crystal oven was one of those simple things to try - the simple jfet heaters I use all the time, as they sound better when hot.
It's never really occured to me before that the L - R channel balance of the Lightspeed would be at all affected by input impedance variations caused by temp rise (or my buffer heating) but your systematic investigation is pretty conclusive and it's a bit of a surprise.
Did I get this right?
"there is perhaps over 1dB attenuation variation between the two channels at different points and this also varies with temp changes .... but it doesn't affect the sound of the device, just the balance?"
If so, then a controlled heater should keep the variation together, at least.
Incidently, your amp load is a bit low at 47kR, might increase this to maybe 70kR + - not sure, but might change sound a bit.
jameshillj said:
Hi James
Mmmm - now that is interesting . . .
Do you think one would achieve this better sound in jfets by heating them simply by driving them a little harder so they get hot from inside themselves?
just guessing but I suspect that using internal heating due to power dissipation will massively increase the temperature swings in the junction temperature and that this will directly lead to dynamic changes in parameters adversely affecting the sound.mikelm said:
External heating, even from a resistor/diode/semi on chip, would be a far better way to assess the the temperature effect.
My opinion is to keep everything cool enough to operate correctly, but that is not quite the same as, keep everything cold.
Hi Andrew,
Yes, not too hot for these guys - I find about 50*C is about tops - other trannes, diodes, much cooler and chips even more so.
Mike,
As Andrew has said, this isn't the way to do this as Jfet in particular have a definite "sweet spot" for voltage, current and load, even when using them as buffers. Putting your best efforts towards top power supplies has a much more direct effect on the device - the temp thing is a bit of final detail really, but it does have an effect.
Over on the "Jfet Boz" thread, there is quite a bit of info about all this and currently, a rather good power supply from ZenMod that's worth a look - perhaps a bit over the top for the LDRs and buffers but a good supply DOES enhance them .
I like to see just how much you can get out of these simple designs and Georges little fella is astonishing - my F3 has 9kR inputZ and so needs a buffer, otherwise I wouldn't use one.
Yes, not too hot for these guys - I find about 50*C is about tops - other trannes, diodes, much cooler and chips even more so.
Mike,
As Andrew has said, this isn't the way to do this as Jfet in particular have a definite "sweet spot" for voltage, current and load, even when using them as buffers. Putting your best efforts towards top power supplies has a much more direct effect on the device - the temp thing is a bit of final detail really, but it does have an effect.
Over on the "Jfet Boz" thread, there is quite a bit of info about all this and currently, a rather good power supply from ZenMod that's worth a look - perhaps a bit over the top for the LDRs and buffers but a good supply DOES enhance them .
I like to see just how much you can get out of these simple designs and Georges little fella is astonishing - my F3 has 9kR inputZ and so needs a buffer, otherwise I wouldn't use one.
Hi James,
47k of my amp seems to be enough, because output impedance of MKII is max. 2.5k at -3db. The range between -6dB and -50db it is lower than 1k.
I'm not sure that heating of the optocoupler will be the solution at all because these little bastards act like they want. Temperature seems to be just one parameter of the whole behaviour to get one fixed resistance value. As you said power supply is one of the key success factors. For my test rig I just have a simple wall wart that feeds the STK500, the µC and the SR2s. Perhaps I should first build something more reliant and precious.
Beside my full time job and the family it's hard to act at the rare freetime as a audio developer! 😀
47k of my amp seems to be enough, because output impedance of MKII is max. 2.5k at -3db. The range between -6dB and -50db it is lower than 1k.
I'm not sure that heating of the optocoupler will be the solution at all because these little bastards act like they want. Temperature seems to be just one parameter of the whole behaviour to get one fixed resistance value. As you said power supply is one of the key success factors. For my test rig I just have a simple wall wart that feeds the STK500, the µC and the SR2s. Perhaps I should first build something more reliant and precious.
Beside my full time job and the family it's hard to act at the rare freetime as a audio developer! 😀
It's funny really Thomas that in the last 60 pages there are perobably all sorts of variations about power supplies, o/p buffers, etc -
I looked back to Feb this year to look up the voltage controlled current source design posted by Gooteee, I think, that might be worth another look -I think you were looking at some of these variations at that time, no?
John 65b's design with his Jfet Boz added as an o/p drive caught my eye - sort of "Serendipity" eh!!
With your MkII, I think you'll get good results with a simple discrete series reg'r supply (no 78xx, 317, etc) with led string, no noisy zeners, no 1N40xx diodes, etc - keep it simple and can then easily 'tweak it up'.
I think I just might go thru the thread again and seperate the cct variations, power supplies, etc - the thread is getting longer and lots of ideas and questions are getting repeated a bit.
Who would have thought that such a simple handful of gadjets could prove to be so interesting!!
I looked back to Feb this year to look up the voltage controlled current source design posted by Gooteee, I think, that might be worth another look -I think you were looking at some of these variations at that time, no?
John 65b's design with his Jfet Boz added as an o/p drive caught my eye - sort of "Serendipity" eh!!
With your MkII, I think you'll get good results with a simple discrete series reg'r supply (no 78xx, 317, etc) with led string, no noisy zeners, no 1N40xx diodes, etc - keep it simple and can then easily 'tweak it up'.
I think I just might go thru the thread again and seperate the cct variations, power supplies, etc - the thread is getting longer and lots of ideas and questions are getting repeated a bit.
Who would have thought that such a simple handful of gadjets could prove to be so interesting!!