Mark
You dont worry about adding distortion by running DC through the LDR? More voltage means more distortion and since the distortion is caused by the resistive element you wont remove the distortion with a cap, but I get where you are going with this. A guy at Maxim told me to do it this way a few years ago but the DC bit bothers me.
You could do it another way though and that would be with a high value resistor voltage divider coming from output of the attenuator to sample the voltage, then give it a gain of 100X or whatever, then rectify it and compare and adjust. Almost like correcting for DC with a servo I suppose but comparing the two channels not offset.
Problem here is that we get back to Georges complaint about not matching impedances between the two channels because we could have to wildly different impedance values between channels but still get the same output voltage. It would have to be done on both series and shunt and I wonder if it could happen smoothly so the correcting circuits dont fight each other back and forth. If you made it slow enough..
You dont worry about adding distortion by running DC through the LDR? More voltage means more distortion and since the distortion is caused by the resistive element you wont remove the distortion with a cap, but I get where you are going with this. A guy at Maxim told me to do it this way a few years ago but the DC bit bothers me.
You could do it another way though and that would be with a high value resistor voltage divider coming from output of the attenuator to sample the voltage, then give it a gain of 100X or whatever, then rectify it and compare and adjust. Almost like correcting for DC with a servo I suppose but comparing the two channels not offset.
Problem here is that we get back to Georges complaint about not matching impedances between the two channels because we could have to wildly different impedance values between channels but still get the same output voltage. It would have to be done on both series and shunt and I wonder if it could happen smoothly so the correcting circuits dont fight each other back and forth. If you made it slow enough..
Last edited:
I was thinking maybe RMS to DC conversion but then it forces both channels to exactly the same volume which means no balance control and also means that if listening to a great soundstage with a 3 piece band who happen to have 2 of the 3 on the left side... well soundstage would be ruined because the circuit would try to center them. Anyway, was looking at this http://www.analog.com/static/import...283758302176206322RMStoDC_Cover-Section-I.pdf and there is a section about a heater element that was on my mind but the balance thing makes it no good. I think this is exactly why you wanted to put DC through them of a known value so that we have something that IS equal between the two. Still the DC has to be high enough for the AC to not really make a difference and that will cause distortion. I put 6V of AC signal through an LDR and it was an evil terrible sound even though it was music. They dont like to be out of their <3V range.
Last edited:
What if the DC test and set up signal was chopped to only pass during gaps in the programme? Then the circuit remembers that new setting until the next gap in the programme.
So when AC = 0 then an opamp fires up and sends DC through the LDRs, on the other end we measure. Makes sense Andrew. If you want to change volume mid program then it auto mutes while this happens so there would have to be a second series LDR or series resistor before the mute circuit so we could still measure output voltage between the two series LDRs as the second one would always be at 40R but the first one would be the one doing the voltage division with shunt. Then we dont have to have a cap in the signal as the DC after the voltage division will be shunted to ground via the mute circuit.
BTW I just tested the rotary coded switch with a set up of 4 LDR values.
Values were 10k8, 8k2, 580R and 70R. I went through every combination that a hexidecimal 16 position switch would give us and ended up with
10k8
8k2
2k4
555
500
492
447
70
70
70
70
69
81?
81?
81?
must have jiggled something on the 81 measurements but anyway it was obvious before that that these numbers and any others would never work for values above a few hundred ohms of Rtot because its important for the hexadecimal thing to work that the lowest number to not be as influential as is actually is.
Values were 10k8, 8k2, 580R and 70R. I went through every combination that a hexidecimal 16 position switch would give us and ended up with
10k8
8k2
2k4
555
500
492
447
70
70
70
70
69
81?
81?
81?
must have jiggled something on the 81 measurements but anyway it was obvious before that that these numbers and any others would never work for values above a few hundred ohms of Rtot because its important for the hexadecimal thing to work that the lowest number to not be as influential as is actually is.
CPC1219 would work for detecting 0VAC. Also for mute, or relay for mute. CPC1219 is a optocoupler/transistor. When 0V through it then it goes to 5R resistance. When enough to fire a LED then it goes to megohms. So we could rectify a tiny bit of the music, through a 100k or more resistor off of the output. It would be amplified 1000X or something large like that via an opamp. Then the output to a zener so we get voltage that wont damage the LED in the CPC1219. The CPC1219 would go to megohms and this would be the signal for the calibration to start and since it would also be working on the same principle when the DC is flowing, the CPC1219 would continue doing its job for the whole calibration process. How to turn off calibration?? Same idea maybe? I suppose what you could do is increase the volume manually then hit the CAL button, it would calibrate for a few seconds and then you would be back in business. That way you can hear the volume change, then calibrate for perfect balance and impedance. If we went with 0VAC being the trigger for CAL then the CPC1219 would be good but if you use a CAL button then it would only be decent for a mute device in which case a relay would be just as good.
Last edited:
Well, if you guys or anyone wants to try a cal circuit off on another thread I will do what I can to help and learn. I am good at research, have some funds to work on circuits and obviously have the interest to keep with it. I also am getting pretty good at board layout so we could send proto boards to each other and have them be decent. My request would be that we try to do it with no programming so any DIYer can work on it, alter it, whatever they like without getting stuck in a rut where not many can help them.
For sure the 96 trimmer thing will work, but what a lot of mucking around. This idea would always be dialed in without the need for manually dialing in.
For sure the 96 trimmer thing will work, but what a lot of mucking around. This idea would always be dialed in without the need for manually dialing in.
Last edited:
Why bother adjusting the balance for a new volume setting mid programme. Balance was OK during the last session and will be re-calibrated during the next silence, 2minutes away or 20minutes away.
For that short while the balance could be out, hopefully only slightly. I would put up with that.
For that short while the balance could be out, hopefully only slightly. I would put up with that.
I dont think that auto cal on 0VAC will work very well. Theoretically yeah. However, it has to work for more than just a few people. For instance I use JRiver. Each song melds into the next so there is never 0VAC. Okay thats just for me and a handful of JRiver users. The other problem would be that I think the calibration would take a minimum of 5 to 10 seconds to be decent and maybe a bit longer to be good enough for balanced if it were a balanced circuit. Take an LDR up to 10k and watch it waiver and thats why it will take a while. If the circuit averaged the values it got for a few seconds then acted on that value.. This wouldnt matter for single ended very much but for balanced it would matter a lot.
While I was eating lunch I thought about this.. what if we have a reference channel. So either L or R or one line of one balanced channel is hard set, meaning that we dial those values in or match those 2 LDRs real well, like with the previous trimmer idea. Then everything else must conform to the voltages we get out of those 2 LDRs...?
So we could either match or go by a 12-24 step volume increase for one shunt and one series LDR. Then the circuit could try to force the other channel, if single ended, or the other 3 if balanced... to match the reference LDRs.
OR perhaps even easier we could do the real great matching on one pair of LDRs for series and create this circuit to assume that the series are matched and it will then force the shunts to be 'value X' that will output a voltage equal on both channels!!
Still, for balanced you could do the trimmered 12-24 step volume adjustment for the series LDRs and then this circuit to force shunts to output the same voltage on each.
While I was eating lunch I thought about this.. what if we have a reference channel. So either L or R or one line of one balanced channel is hard set, meaning that we dial those values in or match those 2 LDRs real well, like with the previous trimmer idea. Then everything else must conform to the voltages we get out of those 2 LDRs...?
So we could either match or go by a 12-24 step volume increase for one shunt and one series LDR. Then the circuit could try to force the other channel, if single ended, or the other 3 if balanced... to match the reference LDRs.
OR perhaps even easier we could do the real great matching on one pair of LDRs for series and create this circuit to assume that the series are matched and it will then force the shunts to be 'value X' that will output a voltage equal on both channels!!
Still, for balanced you could do the trimmered 12-24 step volume adjustment for the series LDRs and then this circuit to force shunts to output the same voltage on each.
I commissioned an expert/friend on this way back, one thing that can't be done without introducing extra distortions is an auto calibrating system that monitors the resistances of all NSL's and keeps them equal for balance/ channel balance purposes.
What can be done without introducing anything, is a semi auto calibration arrangement where at the beginning of listening at a pre set volume a cal set button is pressed and this calibrates all the NSL's for correct balance, trouble with this is it has to be done every time the volume is changed.
1: You start the track pick your volume you want listen at.
2: Stop the track press the auto cal button.
3: Then start the track again and then listen.
As you can see this only hooks up the auto cal sensing circuit while not listening and the resistances, then removes it when you are ready to listen.
But who would want to live with this, best off just quad matching your NSL's and KISS
Cheers George
What can be done without introducing anything, is a semi auto calibration arrangement where at the beginning of listening at a pre set volume a cal set button is pressed and this calibrates all the NSL's for correct balance, trouble with this is it has to be done every time the volume is changed.
1: You start the track pick your volume you want listen at.
2: Stop the track press the auto cal button.
3: Then start the track again and then listen.
As you can see this only hooks up the auto cal sensing circuit while not listening and the resistances, then removes it when you are ready to listen.
But who would want to live with this, best off just quad matching your NSL's and KISS
Cheers George
Been lurking as usual!
I still use my Maximus' lightspeed. Still working fine and still not in a box!
This balance issue has been w/ my lightspeed since day 1. As I turn up the volume, the image shifts to the right slightly.
I had an idea; a more passive approach (and I like the use of old meters), using an old balance meter (pointer rests in a vertical state). Feed voltages (decoupled somehow) from the Dallas pot chips to the meter.
Meter starts in center position at 0 volume, no deflection.
Play source white noise or music. Then turn up the volume and adjust balance to center image. Take note on meter deflection (either by it's scale or mark my own scale). Repeat for different steps of volume. Sort of crude, but could work and look rather cool!
I have no idea how to implement this yet. Just a thought.. and my 2c!
Cheers- Kent
I still use my Maximus' lightspeed. Still working fine and still not in a box!
This balance issue has been w/ my lightspeed since day 1. As I turn up the volume, the image shifts to the right slightly.
I had an idea; a more passive approach (and I like the use of old meters), using an old balance meter (pointer rests in a vertical state). Feed voltages (decoupled somehow) from the Dallas pot chips to the meter.
Meter starts in center position at 0 volume, no deflection.
Play source white noise or music. Then turn up the volume and adjust balance to center image. Take note on meter deflection (either by it's scale or mark my own scale). Repeat for different steps of volume. Sort of crude, but could work and look rather cool!
I have no idea how to implement this yet. Just a thought.. and my 2c!
Cheers- Kent
the switched 4way could have 400r for both series LDRs elements and 200r for both shunt LDR elements.
But I would not connect that equivalent 600r load to my sources.
__________________
regards Andrew T.
The above situation is a result of hasty reply and will never occur with a quad pot.
All u need is a preset across each pot except one, with which the LDRs will be calibrated.
Enjoy
Gajanan Phadte
Note:in this configuration, both the volume controls will have same value of Z and NOT balanced output for both channels.
But I would not connect that equivalent 600r load to my sources.
__________________
regards Andrew T.
The above situation is a result of hasty reply and will never occur with a quad pot.
All u need is a preset across each pot except one, with which the LDRs will be calibrated.
Enjoy
Gajanan Phadte
Note:in this configuration, both the volume controls will have same value of Z and NOT balanced output for both channels.
Last edited:
Your preset idea works at a predetermined level, but when that's changed so too will all the presets have to be re-aligned, as the 4 x NSL's are all different in their mA to resistance curves no 2 are the same.
It is still best to quad match as I've outlined for a KISS approach, why complicate things without having any advancement in sound quality.
Cheers George
It is still best to quad match as I've outlined for a KISS approach, why complicate things without having any advancement in sound quality.
Cheers George
With you all the way here George !
Young people these days, never heard of Occam or his beard removal device.