And this was for my Mk I version with a series resitor and just one shunt LDR per channel.
The MK II sounds better than the MK I, and has a much more constant input and output impedance around 7K-10K And it does not have the distortion peak at -6db like the MK I has.
Cheers George
The MK II sounds better than the MK I, and has a much more constant input and output impedance around 7K-10K And it does not have the distortion peak at -6db like the MK I has.
Cheers George
Attachments
georgehifi said:
That's an interesting graph.
You can see the point where the series and shunt LDR's are the same
value, at this point their non linearities should cancel provided each
LDR has very similar characteristics.
I'd like to see an FFT of just the single version and how far down
H3 is from H2. If the LDR is used in fully balanced shunt mode then
H2 will be largely cancelled and we will be left with mainly H3. If H3 is a
long way down - well that's a good thing
I'm sure Nelson is already thinking of this!
Terry
I will try to make some FFTs of the Philips LDR07 cell tonight, just out of curiosity and maybe it's of interest here also. I like discrete cells more than LED controlled ones because I can light them with an incandescent lamp, which gives me a resistance ratio of ~1e5 (~30Ohms...~3Megohms) with the LDR07
If one runs tracking LDRs with DC-offset applied in balanced mode (includind the DC), it's maybe only even order harmonics and those cancel... just an idea...
Regards, Klaus
If one runs tracking LDRs with DC-offset applied in balanced mode (includind the DC), it's maybe only even order harmonics and those cancel... just an idea...
Regards, Klaus
Silonex graphs are confusing
I am really glad that I am not the only one who didn't understand that graph. I can't find the relation between dBr and x-axis. Normally, or as far as I understood, the dB-value has to rise the higher the current (or voltage) is. But not as shown.
Regards
Thomas
I am really glad that I am not the only one who didn't understand that graph. I can't find the relation between dBr and x-axis. Normally, or as far as I understood, the dB-value has to rise the higher the current (or voltage) is. But not as shown.
Regards
Thomas
LDR FFTs
I did some qualitative SA with the LDR07 cell in a series connection (see attachment):
1) high level drive (top),
2) low level drive (center) and
3) low level drive with DC offset (bottom).
ad 1) The amount of H2 is apparent but completely counter-intuitive to me. H2 is ~90dB down, H3 is ~60dB down.
ad 2) Reducing voltage drive (-18dB) reduces harmonics (as one would expect) but also changes even/odd distribution. Now H2 is ~-95dB, and H3 is ~90dB.
ad 3) Same AC voltage across the cell but with significant DC offset, way more than AC magnitude, only slightly reduces H3 but adds a lot of H2 again, -60dB (plus a H7 component).
Some (very preliminary) conclusions:
- cancelling of even harmonics with a balanced setup might be only effective if there is really an excellent tracking of the LDR characteristics.
- using lowest possible drive levels seems to be beneficial, from noise considerations this implies low LDR working resistances. Maybe a current mode setup around an OpAmp might give better results than a voltage divider working at line levels.
- with channel tracking using DC to measure the actual resistance it will not be easy to get low THD.
Personally I'd rather go for good VCAs like THAT2180...
Regards, Klaus
(arrghh, what is that 1000x1000px image size restriction nonsense, when the pic is a measily 43kB?)
FFTs
I did some qualitative SA with the LDR07 cell in a series connection (see attachment):
1) high level drive (top),
2) low level drive (center) and
3) low level drive with DC offset (bottom).
ad 1) The amount of H2 is apparent but completely counter-intuitive to me. H2 is ~90dB down, H3 is ~60dB down.
ad 2) Reducing voltage drive (-18dB) reduces harmonics (as one would expect) but also changes even/odd distribution. Now H2 is ~-95dB, and H3 is ~90dB.
ad 3) Same AC voltage across the cell but with significant DC offset, way more than AC magnitude, only slightly reduces H3 but adds a lot of H2 again, -60dB (plus a H7 component).
Some (very preliminary) conclusions:
- cancelling of even harmonics with a balanced setup might be only effective if there is really an excellent tracking of the LDR characteristics.
- using lowest possible drive levels seems to be beneficial, from noise considerations this implies low LDR working resistances. Maybe a current mode setup around an OpAmp might give better results than a voltage divider working at line levels.
- with channel tracking using DC to measure the actual resistance it will not be easy to get low THD.
Personally I'd rather go for good VCAs like THAT2180...
Regards, Klaus
(arrghh, what is that 1000x1000px image size restriction nonsense, when the pic is a measily 43kB?)
FFTs
No Mr.Pass, it was typical CD level, 0dB==2Vrms.
The lower two curves had -18dB on the send side and +18dB make-up gain at the reciever.
With the signal being -15dBr, for the upper trace the actual voltage across the LDR must have been on the order of 1.7Vrms
For the center/bottom traces I back-engineer to ~0.2Vrms.
Only a quick test, not a real measurement of any sort. I might repeat this with more precision / documentation when I run into some of the PE or Silonex parts.
Or do you consider using these discrete cells? (AFAIK, the LDR07 is obsolete now).
Regards, Klaus
The lower two curves had -18dB on the send side and +18dB make-up gain at the reciever.
With the signal being -15dBr, for the upper trace the actual voltage across the LDR must have been on the order of 1.7Vrms
For the center/bottom traces I back-engineer to ~0.2Vrms.
Only a quick test, not a real measurement of any sort. I might repeat this with more precision / documentation when I run into some of the PE or Silonex parts.
Or do you consider using these discrete cells? (AFAIK, the LDR07 is obsolete now).
Regards, Klaus
Trace 1: soundcard out --> LDR --> soundcard in (15k Z-in)
Trace 2: soundcard out --> LDR||220k --> 0.22uF --> soundcard in
Trace 3: soundcard out --> LDR||(8V+220k) --> 0.22uF --> soundcard in
Soundcard is a semi-professional 8xIn/Out, 24bit/96kHz model
FFTs (64k) were made over a 10sec time, 95%-overlap, and averaged to get a smoother noise floor
Trace 2: soundcard out --> LDR||220k --> 0.22uF --> soundcard in
Trace 3: soundcard out --> LDR||(8V+220k) --> 0.22uF --> soundcard in
Soundcard is a semi-professional 8xIn/Out, 24bit/96kHz model
FFTs (64k) were made over a 10sec time, 95%-overlap, and averaged to get a smoother noise floor
http://www1.silonex.com/audiohm/levelcontrol.html
This webpage from Silonex explains what the graphs that were posted here mean, under what conditions they were measured, etc.
The X-axis is the control voltage of a constant current source circuit used to drive the LEDs, i.e. no meaning for the audio performance. You might wish to colorate the left Y axis (attenuation) to the right Y axis (THD).
From there I think you can draw your own conclusions what attenuation band you would use most often and then which configuration offers the least distortion.
Patrick
This webpage from Silonex explains what the graphs that were posted here mean, under what conditions they were measured, etc.
The X-axis is the control voltage of a constant current source circuit used to drive the LEDs, i.e. no meaning for the audio performance. You might wish to colorate the left Y axis (attenuation) to the right Y axis (THD).
From there I think you can draw your own conclusions what attenuation band you would use most often and then which configuration offers the least distortion.
Patrick
Member
Joined 2002
Too less resistance
Hi
today I got my 12 NSL-32SR2S: 6x2C, 4x2D, 2x2F.
I started measuring with my 5V PSU and my 10 Euro multimeter. I measured only resistance! My main measuring was after 2 minutes. First I tried 20 minutes but that was too boring. There is a little R shift in time.
I soldered a 100k log pot and 100 Ohm R between V+ and LED+.
I tested with 0 Ohm (~25 Ohm LDR), 18.6 kOhm (~426-528Ohm LDR after 2 minutes) and 92.5 kOhm (~1.9 - 2.64 kOhm LDR).
But with 95 kOhm (max of the pot) I got 3.3 kOhm. How will I simulate a 10k pot with the Lightspeed mkII circuit if the max. LDR resistance is only 3.3 kOhm???
Regards
Thomas
Hi
today I got my 12 NSL-32SR2S: 6x2C, 4x2D, 2x2F.
I started measuring with my 5V PSU and my 10 Euro multimeter. I measured only resistance! My main measuring was after 2 minutes. First I tried 20 minutes but that was too boring. There is a little R shift in time.
I soldered a 100k log pot and 100 Ohm R between V+ and LED+.
I tested with 0 Ohm (~25 Ohm LDR), 18.6 kOhm (~426-528Ohm LDR after 2 minutes) and 92.5 kOhm (~1.9 - 2.64 kOhm LDR).
But with 95 kOhm (max of the pot) I got 3.3 kOhm. How will I simulate a 10k pot with the Lightspeed mkII circuit if the max. LDR resistance is only 3.3 kOhm???
Regards
Thomas
Nelson, easier to read PDF of the same.
http://www1.silonex.com/audiohm/pdf/levelcontrol.pdf
Cheers George
http://www1.silonex.com/audiohm/pdf/levelcontrol.pdf
Cheers George
Re: Re: Too less resistance
Sorry, but I got other results.
If you assume the Lightspeed configuration as a voltage divider with input voltage of around 2V. Then you have R1 with 3.3K Rmax (with 100k pot for LED) and R2 with 25 Ohm Rmin, you got an attenuation of about -42dB as highest attenuation. With a 10k pot you got -80dB and with a 7k pot -77dB.
The main problem is that the lowest resistance is about 25 Ohms. To receive the same effect like a 10k pot (-80dB), the value of R1 has to increase to about 250k. But then your impedance for most amps is very high!
Regards
Thomas
georgehifi said:
Sorry, but I got other results.
If you assume the Lightspeed configuration as a voltage divider with input voltage of around 2V. Then you have R1 with 3.3K Rmax (with 100k pot for LED) and R2 with 25 Ohm Rmin, you got an attenuation of about -42dB as highest attenuation. With a 10k pot you got -80dB and with a 7k pot -77dB.
The main problem is that the lowest resistance is about 25 Ohms. To receive the same effect like a 10k pot (-80dB), the value of R1 has to increase to about 250k. But then your impedance for most amps is very high!
Regards
Thomas
With matched NSL-32SR2S controled by a 100k log pot for the 5vdc this is what I measured, between 7-10k depending on matching, this was backed up by pannelhead and others who have built it as per my circuit diagram, if you are getting something else, maybe your matching isn't tight enough?
Cheers George
Cheers George