in the past I'd just fire up MathCad and derive the symbolic equations myself except its expired again and I'd have to call in to reactivate my version 11 - and it still needs XP
but its not that uncommon a circuit
Google
with a dedicated unity buffer chip you're stuck with the unity gain equations unless you add another op amp for the 2nd order section -which does simplify/separate the design equations
just use the same 1st real pole RC values with the R gnded, I into the 1st RC node == Norton equivalent
but with free choice of LPF I'd use the multiple feedback filter with good VFA - for lower distortion
but its not that uncommon a circuit
with a dedicated unity buffer chip you're stuck with the unity gain equations unless you add another op amp for the 2nd order section -which does simplify/separate the design equations
just use the same 1st real pole RC values with the R gnded, I into the 1st RC node == Norton equivalent
but with free choice of LPF I'd use the multiple feedback filter with good VFA - for lower distortion
Last edited:
George -- to get high enough Q to do what jcx is suggesting with only Rs and Cs you'll need to use feedback. Alternatively, embrace inductors.... 🙂
To do feeback around the BUF03, I'd need another opamp not for me, I have at the moment no global feeback from the output of the PCM1704's all the way to the speakers I think I'll keep it that way, I'll just settle for the 1st order filter at TZ of -3db @125khz
Cheers George
Cheers George
The standard Sallen-Key style 3rd order filter uses a unity gain buffer so you could do it with the BUF03 you already have. But if you prefer to keep things feedback free, then inductors are the only solution I'm aware of.
if you use bigger than 4:1 unequal C in the unity gain buffer Sallen Key then Q>1 is possible with equal R in the 2nd order section considered alone
values have to be adjusted due to the interaction of the added 1st RC section for 3rd order Butterworth
less interaction can be had with further large impedance ratio between 1st section and the 2nd order part
as far as "avoiding feedback" is concerned - why do you think the BUF03 has unity gain in the 1st place? (hint: because it uses 100% voltage feedback internally)
and if you just don't like negative feedback then you may be entertained to know the Sallen Key is actually using positive feedback to boost Q
values have to be adjusted due to the interaction of the added 1st RC section for 3rd order Butterworth
less interaction can be had with further large impedance ratio between 1st section and the 2nd order part
as far as "avoiding feedback" is concerned - why do you think the BUF03 has unity gain in the 1st place? (hint: because it uses 100% voltage feedback internally)
and if you just don't like negative feedback then you may be entertained to know the Sallen Key is actually using positive feedback to boost Q
Last edited:
It sounds magnificent as it is, I think I'll just keep it at the simple 1st order at -3db @125khz on the TZ filter. With this there is just a couple of degrees of phase shift at 20khz, and the vhf noise on the output of the BUF03 is only a couple of mV at more than >100mhz, I don't think this is detectable at all. And then this way it's very simple with no global feedback anywhere in my system.
Cheers George
Cheers George
the worry about “phase shift” is also misplaced – how long do you think the digital filter is – typically low double digits of the sample rate, 100s of degrees
so “phase shift” is not so useful a quality metric, the player already has lots – if linear with frequency it is simply a pure delay
the real “time distortion” is the variation in group delay over the audio range
for a 50 kHz 3rd order Butterworth that's ~ 700 ns of delay change from 20 to 20 kHz
or about 0.2mm (0.008”) of air path difference – how's you speaker driver alignment, what's you're driver x_max?, any baffle diffraction, room reflections? what's the radius of your tweeter diaphragm?
Cary Audio uses 3rd order analog filter
so “phase shift” is not so useful a quality metric, the player already has lots – if linear with frequency it is simply a pure delay
the real “time distortion” is the variation in group delay over the audio range
for a 50 kHz 3rd order Butterworth that's ~ 700 ns of delay change from 20 to 20 kHz
or about 0.2mm (0.008”) of air path difference – how's you speaker driver alignment, what's you're driver x_max?, any baffle diffraction, room reflections? what's the radius of your tweeter diaphragm?
Cary Audio uses 3rd order analog filter
Yes they did it was based on a opa134, every time I used it between the TZ and whichever of the two madeup output buffers I was using. The sound quality suffered everytime it was inserted. Sure the output was clean down to the baseline noise of the Tek scope, but the music just wasn't there anymore.
This is why I have been so happy with the 1st order at 125khz even though there a little bit of vhf **** now seen on the Tek scope on the outputs, the music sounds real. As Zenelectro once said to me "sometimes don't let you eyes decieve you" listen to it.
Cheers George
Passive filter
Hi George, I am having great luck with using an Edcor XSM 600/600. Put a 4n7 polystyrene across the primary. Makes an excellent filter and no feedback. My Mark II 1541A dac uses this after the Pedja Rogic DDNF discrete current mirror I/V. I never liked Sallen Key filters....
Ground return
Agreed.... Use a value between 1K Ohm and 10K Ohm.... Ground side of secondary to ground side of primary. 😉
Luckily I had some old stock AD844 lying around for my modification but I want to purchase a number for further modification. Taking this horror story into account: http://www.diyaudio.com/forums/solid-state/240455-counterfeit-components.html
Could anybody in here advise me to a reliable and secure source for purchasing.
Could anybody in here advise me to a reliable and secure source for purchasing.
Last edited:
Hi George, I am having great luck with using an Edcor XSM 600/600. Put a 4n7 polystyrene across the primary. Makes an excellent filter and no feedback. My Mark II 1541A dac uses this after the Pedja Rogic DDNF discrete current mirror I/V. I never liked Sallen Key filters....
Hi, Torchwood,
You may want to use some caution in placing capacitors across transformer primaries. This creates a resonant tank circuit. By my calculation, placing a 4.7nF cap. across the 3H primary of that Edcor transformer may produce response peaking around 1.3KHz or so.
Last edited:
Hi Ken, My cousin did some testing and found that the response was nice and flat to about 20 Khz. I could ask him to look again around 1 Khz. Thanks for the information. He has the smaller Edcors on a CS4397 dac and had no peaks. Worth another look. 😉
I have now made up my mind which buffer is better the BUF03 or the AD844's. And it is the BUF03 driven from the TZ of the AD844. I found it gave just as dynamic/detailed sound as the 844 buffer but it was more musical, less highlighted around the edges which makes you want to turn it up higher and higher, it has a more relaxed quality to it, yet can still grab you by the scruff of the neck and demand your attention when needed. It's output impedance is 2ohms with 70mA current so it will drive anything even my Sennheiser HD650's. It is highly biased class A and runs very hot at +-15vdc so it needs those very rare TO-99 heatsinks.
Cheers George
Cheers George
Just have to try Mouser and cross my fingers that these new AD844 are identical. At least I understand that I should avoid offers on Ebay
RS Components UK have them and they are the real thing. I would say Mouser would be the real deal as well.
Buy Operational Amplifiers Fast op amp,AD844AN DIP8 Analog Devices AD844ANZ online from RS for next day delivery.
Cheers George
Last edited: