That’s the fun part off DIY audio!! Push the limits.
For the plainness; the pictures were taken during the test. The circuit is like the one I posted.
In the version I built in in my pre-amp, the reference (IC2, IC4) is a LM336-2.5 with 100uF and 100nF in parallel, the transistor (T1, T2) is a BC517 darlington and R2 and R4 are 130 Ohm. I use the regulated filament supply of the tube amplifier to power the circuit.
Peter
I just put a 100uF cap across the two diodes use for CCS Vref and it did indeed bring another level of refinement to the sound 
my experience has taught me not to put a small bypass cap across a larger one unless the small cap has a snubber resistor in series - I have seen in spice* & heard with my ears what confusion it creates so once I get a new supply of caps I will try 1uF & 0.2 ohms across all my 100uF caps.
mike
*estimated lead inductances need to be included
my experience has taught me not to put a small bypass cap across a larger one unless the small cap has a snubber resistor in series - I have seen in spice* & heard with my ears what confusion it creates so once I get a new supply of caps I will try 1uF & 0.2 ohms across all my 100uF caps.
mike
*estimated lead inductances need to be included
Mike,
Take a LM285-1.2 reference (at the cost of only 1 Euro each) to replace the two diodes and you will gain about the same amount in performance as you did with the capacitors! (And leave the capacitor in place.)
It’s funny, but more and more I become convinced that driving the LED’s is as critical as normal amplifier circuitry. So treat the LED’s with the same care as the rest of your circuitry. And it’s explicable, because every little change in the amount of light the LED emits is directly translated in a change of resistance of the LDR. And that’s why we will end up with something really sophisticated.
Peter
Take a LM285-1.2 reference (at the cost of only 1 Euro each) to replace the two diodes and you will gain about the same amount in performance as you did with the capacitors! (And leave the capacitor in place.)
It’s funny, but more and more I become convinced that driving the LED’s is as critical as normal amplifier circuitry. So treat the LED’s with the same care as the rest of your circuitry. And it’s explicable, because every little change in the amount of light the LED emits is directly translated in a change of resistance of the LDR. And that’s why we will end up with something really sophisticated.
Peter
Mike,
I don’t know what will give best results sonically. With a TLV431 the current source becomes a feedback design because the voltage at the ref input is held constant. I have measured the output of this CCS in 1 kOhm with a 5 ½ digit DVM and it is rock steady. Even the last digit. So 1mA in 1 kOhm gives a reading of 1.00000 V. When you take a LM285BYZ-1.2 the last two digits will show some fluctuations because Vbe changes are not compensated. But what is best sonically, I don’t know. I am using now a LM336-2.5 and darlington and that outperforms my test setup which was based on a TL431 with a BC550C. But signal wiring is much shorter now so I am not sure where the benefit comes from.
Peter
I don’t know what will give best results sonically. With a TLV431 the current source becomes a feedback design because the voltage at the ref input is held constant. I have measured the output of this CCS in 1 kOhm with a 5 ½ digit DVM and it is rock steady. Even the last digit. So 1mA in 1 kOhm gives a reading of 1.00000 V. When you take a LM285BYZ-1.2 the last two digits will show some fluctuations because Vbe changes are not compensated. But what is best sonically, I don’t know. I am using now a LM336-2.5 and darlington and that outperforms my test setup which was based on a TL431 with a BC550C. But signal wiring is much shorter now so I am not sure where the benefit comes from.
Peter
Without having tested the circuits, I would personally go for TL431 rather than 336 or 285. It makes much better current control.
If you really want a open loop design for whatever reasons, then just use 2x or 3x red LED driven by a J511 as your voltage reference. Not rock stable, but then very low noise, and no global feedback (if that is the sex appeal).
Patrick
If you really want a open loop design for whatever reasons, then just use 2x or 3x red LED driven by a J511 as your voltage reference. Not rock stable, but then very low noise, and no global feedback (if that is the sex appeal).
Patrick
Arthur,
There is no such thing as a latest design. But there are several voltage reference options to suit the attenuation range and input – output impedance you need. The rest of the circuit remains the same. See post #1001 (and later) for details.
Patrick,
We don’t know yet what gives the best results sonically. It doesn’t make sense to make extended theories because the only way to find out is to try things and let you ears judge.
Badge,
There is no error in the circuit diagram as far as I know.
Peter
There is no such thing as a latest design. But there are several voltage reference options to suit the attenuation range and input – output impedance you need. The rest of the circuit remains the same. See post #1001 (and later) for details.
Patrick,
We don’t know yet what gives the best results sonically. It doesn’t make sense to make extended theories because the only way to find out is to try things and let you ears judge.
Badge,
There is no error in the circuit diagram as far as I know.
Peter
Increase Zin of amp
Excuse me for butting in with a slightly off-topic question -
The Zin of the F3 (ZV9, etc) power amp is only 10K and I'm using a buffer after the L'speed to make it function (remarkably well) but, as George has said, this is not ideal.
Can this input Z be increased to 100K without an input buffer stage?
I have also found that TL431's benefit from simple noise filters in CCS situations.
Thank you George for your design and years of development.
... James
Excuse me for butting in with a slightly off-topic question -
The Zin of the F3 (ZV9, etc) power amp is only 10K and I'm using a buffer after the L'speed to make it function (remarkably well) but, as George has said, this is not ideal.
Can this input Z be increased to 100K without an input buffer stage?
I have also found that TL431's benefit from simple noise filters in CCS situations.
Thank you George for your design and years of development.
... James
Hi James, Nelsons amps are not the norm input wise, his earlier Threshold Stasis could be increased easily, via the input resistor (also fet input), I've done a couple of them without having dc offset problems, but the F3 is a different beast.
I know he reads this Lightspeed thread as I've had comunications with him, he is very interested in the Lightspeed progress, so maybe he will reply to this question better than I, if not in a couple of days I will ask him to contribute.
Cheers George
I know he reads this Lightspeed thread as I've had comunications with him, he is very interested in the Lightspeed progress, so maybe he will reply to this question better than I, if not in a couple of days I will ask him to contribute.
Cheers George
Attachments
increased Zin for F3
Thank you, Nelson -
And for all your designs, publications and "shared time" over the years - will look forward to this one, too.
Currently using a fairly standard k170, j74 cct with shunt supplies that works fairly well but "sucks out" some of the dynamics (unlike the NS10 cct) - could be better, no doubt.
L'speed has a different sound to the X2.5, especially with the NOS dac.
Both F3 amp and preamp are working perfectly and do like silver IC's.
Regards ... James
Thank you, Nelson -
And for all your designs, publications and "shared time" over the years - will look forward to this one, too.
Currently using a fairly standard k170, j74 cct with shunt supplies that works fairly well but "sucks out" some of the dynamics (unlike the NS10 cct) - could be better, no doubt.
L'speed has a different sound to the X2.5, especially with the NOS dac.
Both F3 amp and preamp are working perfectly and do like silver IC's.
Regards ... James
pietjers said:
Peter,
I want to build a version of your CCS to test. My setup has mono controls, about 125K log pots, parallel tracks from a stereo 250K Alpha.
Looking at your circuit, the main change is a single LDR per circuit. I think the value of R2 and R4 needs to be doubled. Would the voltage reference also need to be increased.
I should have some TL-431 2.5 volte references, but also some LM329DZ, Low noise, but 6.95 volt refferences. Would a reference this high be usable?
My power supply is a raw 17 volts. Now how tracking regs dropping to 12 volts and then down to 5. This can be changed to any supply voltage from 14 volts down.
I would ask this offline, but your email is not enabled.
Thanks in advance
George
Panelhead George,
The 250k pots and TL431 are just fine! The other references are too high.
Let me explain how to calculate the range of the CCS. It’s VERY simple.
First you need to calculate the voltage at the emitter of T1, T2 (Ve). For 3-pin references such as TL341 and TLV431 this is very simple because these IC’s maintain Vref at their ref pin, so Ve=Vref. For two pin references you will have to subtract the Vbe of the transistor from the voltage of the reference. Because the base current is very small, 0.5V per p-n junction is a good value to calculate with. So subtract 0.5V for a normal transistor and 1V for a darlington.
Now divide the Ve by the resistance between the emitter and the anode of the reference, and you have the current of the CCS. For Imax (the maximum current) of the CCS I take 10 mA. First calculate R2 and then calculate the range of the CCS.
Formulas for TL431 / TLV431:
R2 = Vref / Imax
Imin = Vref / (P1 + R2)
Formulas for two pin references (such as LM336-2.5, LM285BYZ-1.2, etc.)
R2 = (Vref – Vbe) / Imax
Imin = (Vref – Vbe) / (P1 + R2)
So with a TL431, a 250k pot and a Imax of 10 mA:
R2 = 2.495 / 0.01 = 249.5 Ohm
Imin = 2.495 / (250k + 249.5) = 10 uA
This will give a LDR resistance range of approximately 50 Ohm to 20 kOhm.
Peter
The 250k pots and TL431 are just fine! The other references are too high.
Let me explain how to calculate the range of the CCS. It’s VERY simple.
First you need to calculate the voltage at the emitter of T1, T2 (Ve). For 3-pin references such as TL341 and TLV431 this is very simple because these IC’s maintain Vref at their ref pin, so Ve=Vref. For two pin references you will have to subtract the Vbe of the transistor from the voltage of the reference. Because the base current is very small, 0.5V per p-n junction is a good value to calculate with. So subtract 0.5V for a normal transistor and 1V for a darlington.
Now divide the Ve by the resistance between the emitter and the anode of the reference, and you have the current of the CCS. For Imax (the maximum current) of the CCS I take 10 mA. First calculate R2 and then calculate the range of the CCS.
Formulas for TL431 / TLV431:
R2 = Vref / Imax
Imin = Vref / (P1 + R2)
Formulas for two pin references (such as LM336-2.5, LM285BYZ-1.2, etc.)
R2 = (Vref – Vbe) / Imax
Imin = (Vref – Vbe) / (P1 + R2)
So with a TL431, a 250k pot and a Imax of 10 mA:
R2 = 2.495 / 0.01 = 249.5 Ohm
Imin = 2.495 / (250k + 249.5) = 10 uA
This will give a LDR resistance range of approximately 50 Ohm to 20 kOhm.
Peter
New Lightspeed owner.
Living in Australia,I decided to take a punt on George's little creation.
I've never met George,but it turns out I surfed on George's turf for a few years('72-'76).
From the time I sent George the money,to the day I received the Light speed,was almost lightspeed in itself.I think George bumped me up the list a bit.
My system is comprised of a collection of some great sounding vintage components.Japanese,Solid State.Awesome sounding.
My phono stage is brand new(Simaudio Moon LP5.3) and my speakers are diy activated Altec Valencias.
The Lightspeed was inserted into the system as a substitute for an EchoWars modified Kenwood 700C preamp.
First-up I had to turn-up the input level for my active crossover,since unlike the 700C,the LS doesn't provide any gain.
The initial listening impressions were very favourable for the LS.In my system the EW 700C wipes the floor with the other new(Transcendent Sound Grounded Grid)and Vintage preamps(High quality Japanese Solid State)that I have.
The LS initially sounded on a par with the EW 700C.However further use of the unit has elevated it above the 700C.
They both sound cut from the same cloth.Wonderful dynamics,richly harmonic,great openness and spaciousness,well defined imagery.great focus,natural tonality,great rhythm pace and timing and a general sense of being 'unprocessed'.
Both extremely musical,and musically involving.
It's like the LS takes all the individual qualities of the EW 700C,and extends each parameter a further step or two.
Very nice indeed.
I may incur the wrath of George(insert the sensation of the ground shaking here),and admit that I have inserted a Burson Audio Buffer stage after the Lightspeed to no apparent ill effect.
I'll try taking the Burson out of the loop again,to test it's contribution.
So far it's a big thumbs-up from me for the Lightspeed.As long as you have high output sources(no vintage tuners,cassette decks),and your impedances aren't contrary, it's a very,very nice sounding unit.
Thanks George.
Living in Australia,I decided to take a punt on George's little creation.
I've never met George,but it turns out I surfed on George's turf for a few years('72-'76).
From the time I sent George the money,to the day I received the Light speed,was almost lightspeed in itself.I think George bumped me up the list a bit.
My system is comprised of a collection of some great sounding vintage components.Japanese,Solid State.Awesome sounding.
My phono stage is brand new(Simaudio Moon LP5.3) and my speakers are diy activated Altec Valencias.
The Lightspeed was inserted into the system as a substitute for an EchoWars modified Kenwood 700C preamp.
First-up I had to turn-up the input level for my active crossover,since unlike the 700C,the LS doesn't provide any gain.
The initial listening impressions were very favourable for the LS.In my system the EW 700C wipes the floor with the other new(Transcendent Sound Grounded Grid)and Vintage preamps(High quality Japanese Solid State)that I have.
The LS initially sounded on a par with the EW 700C.However further use of the unit has elevated it above the 700C.
They both sound cut from the same cloth.Wonderful dynamics,richly harmonic,great openness and spaciousness,well defined imagery.great focus,natural tonality,great rhythm pace and timing and a general sense of being 'unprocessed'.
Both extremely musical,and musically involving.
It's like the LS takes all the individual qualities of the EW 700C,and extends each parameter a further step or two.
Very nice indeed.
I may incur the wrath of George(insert the sensation of the ground shaking here),and admit that I have inserted a Burson Audio Buffer stage after the Lightspeed to no apparent ill effect.
I'll try taking the Burson out of the loop again,to test it's contribution.
So far it's a big thumbs-up from me for the Lightspeed.As long as you have high output sources(no vintage tuners,cassette decks),and your impedances aren't contrary, it's a very,very nice sounding unit.
Thanks George.
majkisk said:
As yet no, one day maybe.
But I can see a problem already with the 4 x matched NSL-32SR2S, I could send them out matched in a kit, then a heavy handed someone and soldering iron too hot will unmatch them again, yes you can change the resistance curve with too much heat from a very hot soldering iron.
The built version is 500aud which is 296euros, a kit would be around 100aud cheaper, it's the matching of the NSL-32SR2S that is the most labour intensive part of the Lightspeed Attenuator, and there's no way that part can be left to the diy'er who buys the kit.
Cheers George