I believe PLLXO is the quintessence of crossovers. Super clean, no ill effect, just pure loveliness. But they have a huge problem. Matching with the actual amplifer and its input impedance. And especially your next amplier. Until now! Our hero enters - Adason!
Here is his contribution to humanity - a buffered PLLXO.
His own thread: https://www.diyaudio.com/community/threads/simple-active-crossover.368702/
I have a strong desire to try this out. This thing is sleak!
However, I lack any skill, or even talent, in the electrical field. I have to do this in the short brakes I have from cutting human tissue samples looking for cancer, and other unimportant tasks.
I have come up with this completing schematic which might help anyone who tries to mimic this circuit.
Some things I am trying to figure out:
¤ Bi-polar caps on input and output. Should probably be used.
¤ Dual Rail - important to understand how this works. See V+ and V- and ground in schematics.
Adasons preference is something similar to this PSU that can reduce the need for the 100uF/V35.
https://www.ebay.com/itm/164845881741?hash=item266194898d:g:ObUAAOSwxmdgj2z3&amdata=enc:AQAHAAAAwNGeP5fYh2oIx+yK4Io3k2fHDvEu5iQEkIL08ckzHZTCZERincMe2HroKGMBjI9NoePqlCIF04GR+QU43jNO/3SE6NKRPArKb8eDvh8Lhgtr3PSM1NmE3c7C2cJ99GSr5GFyHrxVm6oxJTKYz/Yuj6Xlceo0NN9gfvm1zdjGnQFIirgfSnixy3ZW1k/WGBwITaSy4U3TtIW7bX6Sjd8JnXKltnMAp0I7bkxrjvW2TcZIuZsMh6rbXCG9i4b9EM1rkA==|tkp:Bk9SR_jc5unWYA
¤ How to modify the circuit for other frequencies:
The schematic says first, choose C. Well, choose C based on what?
In Dave's pages, which Adason refers to, is says choose R.
Somewhat contradictory,
take
Let a known example of the 150Hz that Adason was working with:
If C= 0.1uF
Then,
R = 1/ (2 x Pi x "crossover frequency" x C
= 1 / (2 x Pi x "crossover frequency" x 0,1 uF
= 1 / (2 x Pi x 150Hz x 0.1uF
= 1 / (2 x 3.14 x 150 x 0.1uF) = 10k Ohm
Adason is using these values a lot in his schematic. I am not sure if every value is based on this formula.
And reversed:
C = 1/(2 x Pi x "crossover frequency" x 10 000 ohm
C = 1/(2 x Pi x 150Hz x 10 000 ohm)
= 1/(2 x 3.14 x 150 x 10 000 ohm)
= 1.06157113e-7 F
= 0.106 uF
= 0.1uF
0.1uF - same can be found everwhere in Adaons schematic.
According to Adason, if using a really clean PSU, you might not need to use the unmarked "rails ripple shaving caps", as ZM so kindly pointed out that they were called.
Anyone are welcome to help me and others to understands this circuit and solder ourselves a working example. Please correct it!
Here is his contribution to humanity - a buffered PLLXO.
His own thread: https://www.diyaudio.com/community/threads/simple-active-crossover.368702/
I have a strong desire to try this out. This thing is sleak!
However, I lack any skill, or even talent, in the electrical field. I have to do this in the short brakes I have from cutting human tissue samples looking for cancer, and other unimportant tasks.
I have come up with this completing schematic which might help anyone who tries to mimic this circuit.
Some things I am trying to figure out:
¤ Bi-polar caps on input and output. Should probably be used.
¤ Dual Rail - important to understand how this works. See V+ and V- and ground in schematics.
Adasons preference is something similar to this PSU that can reduce the need for the 100uF/V35.
https://www.ebay.com/itm/164845881741?hash=item266194898d:g:ObUAAOSwxmdgj2z3&amdata=enc:AQAHAAAAwNGeP5fYh2oIx+yK4Io3k2fHDvEu5iQEkIL08ckzHZTCZERincMe2HroKGMBjI9NoePqlCIF04GR+QU43jNO/3SE6NKRPArKb8eDvh8Lhgtr3PSM1NmE3c7C2cJ99GSr5GFyHrxVm6oxJTKYz/Yuj6Xlceo0NN9gfvm1zdjGnQFIirgfSnixy3ZW1k/WGBwITaSy4U3TtIW7bX6Sjd8JnXKltnMAp0I7bkxrjvW2TcZIuZsMh6rbXCG9i4b9EM1rkA==|tkp:Bk9SR_jc5unWYA
¤ How to modify the circuit for other frequencies:
The schematic says first, choose C. Well, choose C based on what?
In Dave's pages, which Adason refers to, is says choose R.
Somewhat contradictory,
take
Let a known example of the 150Hz that Adason was working with:
If C= 0.1uF
Then,
R = 1/ (2 x Pi x "crossover frequency" x C
= 1 / (2 x Pi x "crossover frequency" x 0,1 uF
= 1 / (2 x Pi x 150Hz x 0.1uF
= 1 / (2 x 3.14 x 150 x 0.1uF) = 10k Ohm
Adason is using these values a lot in his schematic. I am not sure if every value is based on this formula.
And reversed:
C = 1/(2 x Pi x "crossover frequency" x 10 000 ohm
C = 1/(2 x Pi x 150Hz x 10 000 ohm)
= 1/(2 x 3.14 x 150 x 10 000 ohm)
= 1.06157113e-7 F
= 0.106 uF
= 0.1uF
0.1uF - same can be found everwhere in Adaons schematic.
According to Adason, if using a really clean PSU, you might not need to use the unmarked "rails ripple shaving caps", as ZM so kindly pointed out that they were called.
Anyone are welcome to help me and others to understands this circuit and solder ourselves a working example. Please correct it!
The R and C values can be traded off, keeping their product constant, for a given crossover frequency.
Choose an exact C value that you can buy (like 0.1uF), and then calculate the closest 1% value for R.
Don't omit the local decoupling capacitors. MUST use all of the DC blocking capacitors.
Those are expensive transistors BTW.
Choose an exact C value that you can buy (like 0.1uF), and then calculate the closest 1% value for R.
Don't omit the local decoupling capacitors. MUST use all of the DC blocking capacitors.
Those are expensive transistors BTW.
Last edited:
Use the R value given, because it probably comes from a DC or loading requirement.
From your F, figure your C.
That will be a funny value. Round to nearest standard value.
Re-figure R based on your practical C.
______
I'm not sure why the inputs need blocking caps for domestic hi-fi. +/-10V won't blow it up. Yes, for an on-stage amp I would add caps AND series resistors to withstand big speaker or wall-outlet voltages, but I'd also use a stricter X-over, so that's off the table.
From your F, figure your C.
That will be a funny value. Round to nearest standard value.
Re-figure R based on your practical C.
______
I'm not sure why the inputs need blocking caps for domestic hi-fi. +/-10V won't blow it up. Yes, for an on-stage amp I would add caps AND series resistors to withstand big speaker or wall-outlet voltages, but I'd also use a stricter X-over, so that's off the table.
Thanks for the help! 150Hz coincidently is a goal for me because I want to hornload exactly that deep. Currently only down to 220Hz, but I am changing drivers and modding the horn to be longer. Going deeper would make it too large in width. 60cm width is enough for now.
Expected result from Hornresp after upgrades VS recent REW measurement using Fane S 8M 8" with no back chamber.
Regarding power supply, I found this comparison:
https://linearaudio.nl/sites/linearaudio.net/files/v4 jdw.pdf
I was about to also try Salad Ultrabib, but apparently sounds harsh compared to Jung Didden in this listening group test.
https://diyaudiostore.com/products/super-regulator
Turns out I have a super-regulator PCB as well. And a 15V-0-15V Toroidy transformer was sitting and waiting for a suitable project after I abandoned the Mezmerize preamp years ago.
Expected result from Hornresp after upgrades VS recent REW measurement using Fane S 8M 8" with no back chamber.
Regarding power supply, I found this comparison:
https://linearaudio.nl/sites/linearaudio.net/files/v4 jdw.pdf
I was about to also try Salad Ultrabib, but apparently sounds harsh compared to Jung Didden in this listening group test.
https://diyaudiostore.com/products/super-regulator
Turns out I have a super-regulator PCB as well. And a 15V-0-15V Toroidy transformer was sitting and waiting for a suitable project after I abandoned the Mezmerize preamp years ago.
Last edited:
I am using this schematic as a learning example, so I still have not identified even what exactly is the filter and what is, for example the "local decoupling caps" or "DC blocking caps." Maybe I am unlucky and some of them has the same value as the filter - 0.1uF.
One thing I did learn however, was that NP always uses a 1K resistor to "prevent parasitic oscillation with the very wide bandwidth JFETs." I saw it in B1 and in the 6-24 analog crossover. NP seems eternally afraid of parasitic oscillation and Adason seems to be not so worried.
Last edited:
There are very good reasons for preventing oscillations or RF pickup at tens of MHz.
None have to do with fear though.
A decoupling capacitor connects from a particular circuit node (often the power supply) to ground,
to provide a low inductance path.
The DC blocking capacitors prevent DC voltages from appearing at the following circuit node,
which is often the input of the following stage, or else the output terminal.
None have to do with fear though.
A decoupling capacitor connects from a particular circuit node (often the power supply) to ground,
to provide a low inductance path.
The DC blocking capacitors prevent DC voltages from appearing at the following circuit node,
which is often the input of the following stage, or else the output terminal.
Last edited:
So I should add it to the Adason XO circuit? That makes me afraid, since I don't know how to simulate it yet. Adason was motivated to reduce the amount of components for a cleaner sound.
Commercial products would be made with such a resistor at the gate of the fets.
In fact many products would not function properly at all, without the resistors.
Most simulations for home projects won't be accurate at such high frequencies, though.
In fact many products would not function properly at all, without the resistors.
Most simulations for home projects won't be accurate at such high frequencies, though.
Back to (not) understanding the filter.
The Hi-pass looks like a 1st order hi-pass.
The Lo-pass looks a bit more complicated.
Here is a classic PLLXO example of a 1st order low pass:
Here is a classic PLLXO example of a 2nd order low pass.
Here is Adasons version of a Lo-pass:
I can't even tell if this is 1st or 2nd order, but with the amount of components here, I am leaning towards 2nd order. There are two R = 10R, and two 0.1uF caps, although they are not doing anything that I can comprehend.
Maybe to understand the filter, I have to understand what the Jfets are doing. Nelson Pass writes quickly in the B1 article:
"If you put a buffer in front of a volume control, the control’s low impedance
looks like high impedance. If you put a buffer after a volume control, it
makes the output impedance much lower. You can put buffers before and
after a volume control if you want." - NP in B1 article
This might be important to the amp, but how is it important to the filter?
The Hi-pass looks like a 1st order hi-pass.
The Lo-pass looks a bit more complicated.
Here is a classic PLLXO example of a 1st order low pass:
Here is a classic PLLXO example of a 2nd order low pass.
Here is Adasons version of a Lo-pass:
I can't even tell if this is 1st or 2nd order, but with the amount of components here, I am leaning towards 2nd order. There are two R = 10R, and two 0.1uF caps, although they are not doing anything that I can comprehend.
Maybe to understand the filter, I have to understand what the Jfets are doing. Nelson Pass writes quickly in the B1 article:
"If you put a buffer in front of a volume control, the control’s low impedance
looks like high impedance. If you put a buffer after a volume control, it
makes the output impedance much lower. You can put buffers before and
after a volume control if you want." - NP in B1 article
This might be important to the amp, but how is it important to the filter?
Last edited:
It's a second order Sallen - Key low pass filter, which uses positive feedback.
Buffering the input and output for this circuit is important due to the significant impedances present at both nodes.
The circuit topology has two series RC filters (and hence is second order), except that the first RC has the bottom
of the C going to the output. Since the circuit is noninverting, that's where the positive feedback comes into play.
You can enter some values and simulate here. The op amp is much more commonly used than the fets.
http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm
of the C going to the output. Since the circuit is noninverting, that's where the positive feedback comes into play.
You can enter some values and simulate here. The op amp is much more commonly used than the fets.
http://sim.okawa-denshi.jp/en/OPseikiLowkeisan.htm
Last edited:
Aha. That is very helpful. I will look up other circuits to understand "noninverting" and "positive feedback". Thank you!
https://en.wikipedia.org/wiki/Operational_amplifier_applications#Non-inverting_amplifier
https://en.wikipedia.org/wiki/Operational_amplifier_applications#Inverting_amplifier
Positive feedback can be very bad, but small amounts, carefully controlled, can be used.
Here, the available gain of the push-pull follower is less than unity, so it can't get out of hand.
You might want to use this public domain ebook as an introduction to electronics.
https://pearl-hifi.com/06_Lit_Archive/02_PEARL_Arch/Vol_16/Sec_51/4420_The_Art_of_Electronics.pdf
https://en.wikipedia.org/wiki/Operational_amplifier_applications#Inverting_amplifier
Positive feedback can be very bad, but small amounts, carefully controlled, can be used.
Here, the available gain of the push-pull follower is less than unity, so it can't get out of hand.
You might want to use this public domain ebook as an introduction to electronics.
https://pearl-hifi.com/06_Lit_Archive/02_PEARL_Arch/Vol_16/Sec_51/4420_The_Art_of_Electronics.pdf
Will read, thanks!
Adason did write "When I was experimenting with CRC upper path before the second buffer, I noted weird behavior sometimes, when I touched something, n channel jfet started to be fully open and I got 14.9 volts on the output. When I placed 1M ohm resistor from gates to ground, this never happened. So similarly, I placed the 1M on the low path too."
Maybe this was due to parasitic oscillation since he was not as careful as NP by adding a 1K resistor before the Jfets. So maybe the 1M to ground should be 1K to Jfet instead. 🤔
Adason did write "When I was experimenting with CRC upper path before the second buffer, I noted weird behavior sometimes, when I touched something, n channel jfet started to be fully open and I got 14.9 volts on the output. When I placed 1M ohm resistor from gates to ground, this never happened. So similarly, I placed the 1M on the low path too."
Maybe this was due to parasitic oscillation since he was not as careful as NP by adding a 1K resistor before the Jfets. So maybe the 1M to ground should be 1K to Jfet instead. 🤔
Could be, with positive feedback odd things are possible. This is called "sticking to the rail".
The pcb layout will also contribute to the circuit behavior.