Re: Re: Re: TDA5141's I/V with lightspeed
KK this is the simple way, send a 1k sine wave through whatever devices output impedance you are trying to measure, then start loading that output to ground with varing resistances till the 1k signal drops by half, then that is the output impedance of the device. What you have done is formed a voltage divider with the output impedance of whatever you are tying to measure. you can do this with a crow or a dmm.
Cheers George
daredevil_kk said:
KK this is the simple way, send a 1k sine wave through whatever devices output impedance you are trying to measure, then start loading that output to ground with varing resistances till the 1k signal drops by half, then that is the output impedance of the device. What you have done is formed a voltage divider with the output impedance of whatever you are tying to measure. you can do this with a crow or a dmm.
Cheers George
Re: Re: Re: Re: TDA5141's I/V with lightspeed
I did some read up and I think I made some mistakes. The turn ratio is 600:25000. My calculation is 6.45 dB signal gain and the impedance ratio is 1:1736. Given that the DC resistance is 32ohm, the estimate impedance is 40ohm. So reflected impedance is 69.44k!!! Is my calculation correct???
I will try your method out, but give the above calculation I may need to redo the I/V and out stage.
AndrewT said:
I did some read up and I think I made some mistakes. The turn ratio is 600:25000. My calculation is 6.45 dB signal gain and the impedance ratio is 1:1736. Given that the DC resistance is 32ohm, the estimate impedance is 40ohm. So reflected impedance is 69.44k!!! Is my calculation correct???
georgehifi said:
I will try your method out, but give the above calculation I may need to redo the I/V and out stage.
600:25000 looks more like a 600ohm impedance transformer.
A turns ratio of 1:41 seems unlikely.
Using sqrt {600ohms:25000ohms} gives a turns ratio of 1:6.45 that is much more reasonable and tallies with your 6.45 figure.
6.45times is +16.2dB.
It looks like your transformer is a 600ohm to 25kohm impedance converter. Is it for a microphone?
A turns ratio of 1:41 seems unlikely.
Using sqrt {600ohms:25000ohms} gives a turns ratio of 1:6.45 that is much more reasonable and tallies with your 6.45 figure.
6.45times is +16.2dB.
It looks like your transformer is a 600ohm to 25kohm impedance converter. Is it for a microphone?
The audio isn't going through a potentiometer in this circuit. Which means as previously mentioned, no potentially dodgy contact problems etc etc in the audio path, which is good.
But the control voltage IS going through several of them, so the emphasis is now shifted towards those pots, with their limitations etc, as above. DC on potentiometers ( can ) lead to unreliability etc.
I wonder if eliminating one source of potential problems, ( might ) be transferring them in other directions ?
Also having examined the data sheets on all the LDR's, there is no doubt that distortion is a factor which has to be taken into consideration.
But the control voltage IS going through several of them, so the emphasis is now shifted towards those pots, with their limitations etc, as above. DC on potentiometers ( can ) lead to unreliability etc.
I wonder if eliminating one source of potential problems, ( might ) be transferring them in other directions ?
Also having examined the data sheets on all the LDR's, there is no doubt that distortion is a factor which has to be taken into consideration.
ZeroD,
The LED is NOT the LDR its in the same capsule. This is an optocoupler. There is no DC on the signal unless your source is sending out DC. That would not make good sound.
The distortion has nothing to do with the voltage the LED is driven with. Zero, zip, nada. Its got to do with the voltage your CDP or other source is driving through the resistive cell. So find out the voltage your source outputs and you will know the distortion level you can expect. Most likely your source outputs something around 2-3V.
See this page
http://www1.silonex.com/audiohm/constants.html
After reading that page reference this pic that shows distortion for the Lightspeed http://www1.silonex.com/img/audio/level/fig11_sm.gif
Their distortion numbers are all based on DC. I have no idea what they would be with AC.
Uriah
The LED is NOT the LDR its in the same capsule. This is an optocoupler. There is no DC on the signal unless your source is sending out DC. That would not make good sound.
The distortion has nothing to do with the voltage the LED is driven with. Zero, zip, nada. Its got to do with the voltage your CDP or other source is driving through the resistive cell. So find out the voltage your source outputs and you will know the distortion level you can expect. Most likely your source outputs something around 2-3V.
See this page
http://www1.silonex.com/audiohm/constants.html
After reading that page reference this pic that shows distortion for the Lightspeed http://www1.silonex.com/img/audio/level/fig11_sm.gif
Their distortion numbers are all based on DC. I have no idea what they would be with AC.
Uriah
udailey
Hi,
You appear to have misinterpreted what i Actually wrote !
I know it's an optocoupler and how they work.
The points i raised however were about the transfer of potential dodgy contact etc problems, that have been mentioned by others, being moved away from a normal potentiometer with audio running through it, to one/s with DC on them. Whatever contact etc problems may arise with audio running through potentiometers, still apply with DC. In fact as i noted, running DC through potentiometers can cause other problems that don't appear when AC coupled.
I did say, i'd examined the data sheets on all the LDR's, which also included the links you gave. And i also said, there is no doubt that distortion is a factor which has to be taken into consideration. Which is true of course.
http://www1.silonex.com/audiohm/constants.html " The applied voltage across the cell: the higher the voltage, the worse the distortion. Thus, when the device is turned hard ON and cell voltage is small, THD will be very low. "
This means that the current drawn by the LED, together with the audio AC voltage arcoss the LDR, have a direct relationship on distortion. At various points between maximum/minimum wiper travel, the distortion will be less or more, but never zero, and more than what goes in. It would be interesting to see the results of an ultra low distortion oscillator with very low output impedance, being fed into the LDR at various voltage levels, whilst also varying the DC potentiometers. I'm convinced there would be a difference between input and output distortion levels.
I want to make it perfectly clear, that i'm NOT saying this circuit is awful etc at all, because it isn't. I myself have designed and used other opto devices myself with great success. But what comes out is not 100% only what goes in !
A number of statements have been made throughout this thread which i just felt needed addressing.
Hi,
You appear to have misinterpreted what i Actually wrote !
I know it's an optocoupler and how they work.
The points i raised however were about the transfer of potential dodgy contact etc problems, that have been mentioned by others, being moved away from a normal potentiometer with audio running through it, to one/s with DC on them. Whatever contact etc problems may arise with audio running through potentiometers, still apply with DC. In fact as i noted, running DC through potentiometers can cause other problems that don't appear when AC coupled.
I did say, i'd examined the data sheets on all the LDR's, which also included the links you gave. And i also said, there is no doubt that distortion is a factor which has to be taken into consideration. Which is true of course.
http://www1.silonex.com/audiohm/constants.html " The applied voltage across the cell: the higher the voltage, the worse the distortion. Thus, when the device is turned hard ON and cell voltage is small, THD will be very low. "
This means that the current drawn by the LED, together with the audio AC voltage arcoss the LDR, have a direct relationship on distortion. At various points between maximum/minimum wiper travel, the distortion will be less or more, but never zero, and more than what goes in. It would be interesting to see the results of an ultra low distortion oscillator with very low output impedance, being fed into the LDR at various voltage levels, whilst also varying the DC potentiometers. I'm convinced there would be a difference between input and output distortion levels.
I want to make it perfectly clear, that i'm NOT saying this circuit is awful etc at all, because it isn't. I myself have designed and used other opto devices myself with great success. But what comes out is not 100% only what goes in !
A number of statements have been made throughout this thread which i just felt needed addressing.
I am not being argumentative or jumping for the defense of the Lightspeed as I think its sound speaks for itself anyway, but trying to get where you are coming from.
So when you say that the current through the LED changes distortion level how do you mean? How is the current or voltage or anything coming through the LED effecting distortion?
Uriah
So when you say that the current through the LED changes distortion level how do you mean? How is the current or voltage or anything coming through the LED effecting distortion?
Uriah
Not sure if LM336 volt ref has better thermo resistence than LM334, I am using VCCS and it's very stable.
Have you compared the sound for LDR in higher resistence vs lower?
Thanks
Have you compared the sound for LDR in higher resistence vs lower?
Thanks
udailey said:I dont think the double switch is necessary as the resistor will absorb still not allow any extra mA through.
The CCS can be built with an LM334 for a few pennies and this is a CCS that will go far below mA level into uA and pA. Using this will work. Absolutely. It however is not temp stable even when using a temp compensation circuit in high heat. This wont matter in your regular home ambient temp so still a good choice. Inside an amplifier case I would use very stable pots with low low ppm. Right now when I test outside in 100+F heat amazingly the regular trimpots are more stable than the CCS circuit.
The CCS circuit allows me even higher resistance than the trimpots of 1MOhm. Crazy. You can use the LM334 with a pot as the Rset and you are good to go. You will need two of these circuits. One for series one for shunt. You will not have balance control this way but I am sure you could figure one to put in there. Or use 4 LM334. Nice thing is LM334 ONLY puts a max of 20mA so you can not damage your LDRs.
You can use LM317 but still this is not the most stable in the mV range and that WILL effect the stability of the LDR. I like LM2757T but it operates at 52kHz which might might maybe be audible. However it has some family members that operate at MHz range and that will never be audible. 52KHz wont be audible but there are some that would probably argue that. Anyway, the LM2575 is super stable even down at .0001V. All day long I might get from 4.9604VDC to 4.9601 drift.
On the other hand the LM317Z is a TO-92 package that outputs 100mA max and needs no sinking. It also costs pennies. For under $2 you could implement LM317Z and LM334 to control the LDRs. Maybe $4 if you use 4 LM334 and an extra pot to control output voltage so now you are in complete control
Easy circuits to implement as they are all outlined in their datasheets.
Uriah
Hi
So far I have only been running tests on LDRs with LM334. Getting very lucky with using thermistors in combination with pots or thermistors with LM334. What I do is set them at 21k indoors and then put them on the porch at around 100F and see what the difference is in resistance an hour or so later. Next time I order I will try LM336. I have spent crazy amounts of money buying different parts to try with the LDRs and so far the most stable has been a trim pot with a thermistor.
Today I figured out my problem with the temp compensated circuit when I finally read over and over the math behind it. It would need a microprocessor because you can EASILY set it to a temp compensated value using the right math, not just the right circuit, but if you want to vary that resistance at all you have to do the math again OR it actually makes it more than normally susceptible to temp changes. No wonder.
But I have learned and keep learning a great deal with all the messing around and its fun. I dont know why its fun but at least with you guys I am in good company!
What I will try when my order of new wire gets in is to power and control it from indoors while the LDR itself is outside or in the oven and see how much of the problem is the temp co of the resistor and how much is the LDR.
Uriah
So far I have only been running tests on LDRs with LM334. Getting very lucky with using thermistors in combination with pots or thermistors with LM334. What I do is set them at 21k indoors and then put them on the porch at around 100F and see what the difference is in resistance an hour or so later. Next time I order I will try LM336. I have spent crazy amounts of money buying different parts to try with the LDRs and so far the most stable has been a trim pot with a thermistor.
Today I figured out my problem with the temp compensated circuit when I finally read over and over the math behind it. It would need a microprocessor because you can EASILY set it to a temp compensated value using the right math, not just the right circuit, but if you want to vary that resistance at all you have to do the math again OR it actually makes it more than normally susceptible to temp changes. No wonder.
But I have learned and keep learning a great deal with all the messing around and its fun. I dont know why its fun but at least with you guys I am in good company! What I will try when my order of new wire gets in is to power and control it from indoors while the LDR itself is outside or in the oven and see how much of the problem is the temp co of the resistor and how much is the LDR.
Uriah
georgehifi said:
OK, I'll bite.
The raw output of a switchmode power supply will often have larger-amplitude noise+ripple than a linear supply. But the noise+ripple of a switchmode is usually mostly at the relatively-high switching frequency, while the linear's noise+ripple is usually mostly at the mains frequency, which is in the audio band. The switchmode's noise is usually outside the audio band, making it much easier to filter out without affecting any audio frequencies.
But in either case, if you are searching for the ultimate, why not add an LC post-filter (making a CLC Pi filter), which should be able to lower your power supply's output noise to a few tens of microvolts? At such low currents, the inductor could be dirt cheap. Just after your PS's output filter cap, something like 10uH in series followed by 2200 uF to ground ought to do pretty well. And if the distance from the 2200uF to the point of use is more than a few inches, I'd probably also try using a small electrolytic cap in parallel with a small ceramic cap, from the load point to a ground return.
And while it might not be very critical in this circuit, in order to go for the ultimate you might want to try to make sure that all of the filter capacitors' ground returns are physically separate from any low-noise ground returns, all the way back to the power supply's main filter cap, or wherever your star ground point is located. You especially don't want any potentially-noisy ground returns sharing the same ground conductor as any audio input.
Granted, none of that should matter too much, in your attenuator circuit, since the power supply does not have to supply currents that would need to vary at up to audio frequencies; just knob-rotation frequencies.
But if anything about the power supply COULD matter in your circuit, it seems like it might be whether or not there was enough mains ripple in the output to actually modulate the LEDs' intensities, which could modulate the LDRs' resistances, which could modulate the audio.
From that point of view, it might make sense to filter the hell out of the power supply output, even using ten times as much inductance in the (C)LC post-filter. You could even add a post-regulator, after the filter. (But, of course, if there is no mains-frequency component getting into the audio output, then there's no need to worry about it.)
And I suppose that if you were making millions of them, it might make more sense to use a switchmode supply, since it could require a lower inductance value (i.e. a smaller, cheaper inductor) to filter its higher-frequency noise down to any particular desired level.
Tom
Hi,
I'm looking to purchase 25 or so Silonex NSL-32SR2 or R3 from anyone who has bought these in bulk for matching. I posted in the thread by udaily but that seems to have gone quiet. From reading this thread it seems there are a number of folks buying these in bulk, so if you have a number left over I would be interested. I don't need absolute matching, as my LDR attenuator will be driven by a digital potentiometer connected to a PC and I plan to use a lookup table on the PC (the 6 channel amplifier is dedicated to a home theater PC) to fine tune any matching issues.
I'm looking to purchase 25 or so Silonex NSL-32SR2 or R3 from anyone who has bought these in bulk for matching. I posted in the thread by udaily but that seems to have gone quiet. From reading this thread it seems there are a number of folks buying these in bulk, so if you have a number left over I would be interested. I don't need absolute matching, as my LDR attenuator will be driven by a digital potentiometer connected to a PC and I plan to use a lookup table on the PC (the 6 channel amplifier is dedicated to a home theater PC) to fine tune any matching issues.
Hi KK. The one I make, it's closest one would be the dual 10K log potentiometer with a channel balance of +&-1db.
It resembles the logarithmic because in an "average efficiency system" (speakers from 87db to 91db and average gain power amps) it has fine adjustment from a whisper at minimum to medium loud level, at which the control is at about 1 o'clock, and then the volume increases faster from 2 o'clock to max volume
Cheers George
Try calculating the power dissipated in the 100k pot with a 10Vrms signal level.
P = V*V/R =10*10/10^5 = 1mW.
If you want to reduce temperature variations in the track, try doubling the dissipation capability.
If you want to err towards zero measurable effects in the pot track, try ten times the dissipation capability.
Your 100k pot for any sensible line level input does not need to exceed 10mW.
Most will take >=250mW.
Learn to do the arithmetic. Even if it does not turn you on. It's a necessary tool.
BTW, a CD player or a preamp with significant gain and high overhead capability are about the highest voltage sources you will use.
The average signal level from these are likely to be in the range 100mVrms to 500mVrms. About <=3uW
P = V*V/R =10*10/10^5 = 1mW.
If you want to reduce temperature variations in the track, try doubling the dissipation capability.
If you want to err towards zero measurable effects in the pot track, try ten times the dissipation capability.
Your 100k pot for any sensible line level input does not need to exceed 10mW.
Most will take >=250mW.
Learn to do the arithmetic. Even if it does not turn you on. It's a necessary tool.
BTW, a CD player or a preamp with significant gain and high overhead capability are about the highest voltage sources you will use.
The average signal level from these are likely to be in the range 100mVrms to 500mVrms. About <=3uW
Last edited:
I might allready have been discussed here, but what is the minimum wattage for the 100k pot? Reason I am asking is that I have a 100K Alps motor pot with remote that could be used for a remote controlled Lightspeed.
Your pot will do just fine bequerel, a max of 40mA per half no problems.
Cheers George
Last edited: