Two questions
Great board, but I was wondering about a couple of things:
Great board, but I was wondering about a couple of things:
- What high-accuracy caps do you suggest I use for the freq-setting components? I'm wondering about which caps (and from which reseller) will fit into that space.
- I tried the spreadsheet, and it gives me LP and HP only for q of 0.5 or greater. What if I need lower Q? Optimised active filters sometimes have funny Q requirements
High-accuracy caps
Some other options I found were
Which of these will fit?
So, am a bit confused about what caps to look for.
Some other options I found were
- Panasonic PPS film caps, from their ECHS series, on Digikey. 2% tolerance. Digikey part number for the 100nF cap is PS1H104G-ND. Priced $1.95 in single pieces. Unfortunately you'll need 14.5x6mm board area for each cap, and 10mm lead pitch.
- Panasonic P series polypropylene caps, Digikey part number P3104-ND for the 100nF, 2% tolerance, $0.86 in single pieces. But 17.5x7mm board real estate needed.
- Panasonic ECQ-P(Z) series polyprop film caps, Digikey part number P3925-ND for the 100nF value, 2% tolerance, $0.82 in single pieces, but 17x7.5mm board area needed, 7.5mm lead pitch.
Which of these will fit?
So, am a bit confused about what caps to look for.
Hi,
cascade a pair of Butterworth (Q=1/sqrt(2)) and the output is Q=0.5 (Bessel) just right for 4pole Linkwitz Reilly.
Cascade a pair of Q=0.5412 (gain =1.1523) and Q=1.3065 (gain =2.2346) gives Q=0.7071 (Butterworth and overall gain of 2.575, +8.2db)
The resulting Qt=Q1*Q2....*Qn
That's the beauty of this topology, ECV S&K, gain and Q can be varied without changing F, they are independant.
Unity gain filtering involves a lot of maths and unmatched component values that require absolute accuracy rather than just matching but at a fixed gain cf +8.2db of the ECV Butterworth.
cascade a pair of Butterworth (Q=1/sqrt(2)) and the output is Q=0.5 (Bessel) just right for 4pole Linkwitz Reilly.
Cascade a pair of Q=0.5412 (gain =1.1523) and Q=1.3065 (gain =2.2346) gives Q=0.7071 (Butterworth and overall gain of 2.575, +8.2db)
The resulting Qt=Q1*Q2....*Qn
That's the beauty of this topology, ECV S&K, gain and Q can be varied without changing F, they are independant.
Unity gain filtering involves a lot of maths and unmatched component values that require absolute accuracy rather than just matching but at a fixed gain cf +8.2db of the ECV Butterworth.
And any idea what caps to use with this board? The manual specifies 7.5x5.5mm cap footprint, with 5mm lead pitch. I can't seem to find any reasonably tight tolerance cap of the size we need in that footprint.
Maybe I should just switch to 5% tolerances. After all, for passive XO we use 10% tolerance caps anyway.
Maybe I should just switch to 5% tolerances. After all, for passive XO we use 10% tolerance caps anyway.
Hi,
buy 100s of 5% and select MATCHING values for each frequency. Try to better 1% for caps and better than 0.3% for resistors.
Polyester are rarely available in 1%.
1% polystyrene used to be readily available, but more difficult now.
1% or 2% polypropylene are available at more cost and in the larger values in bigger packages.
1% Mica are available in smaller values but have the wrong pin pitch.
No easy answer.
BTW,
single pole fiters can get away with 5%, 10% is rubbish,
as the filter steepness increases the required accuracy increases to match.
for 2pole you may get away with 2% and for 4pole 1% but I would go for closer matching, absolute accuracy is not required for frequency, but it is needed for gain (just as well DMMs are more acurate on DC voltage, pity about the resistance accuracy)
buy 100s of 5% and select MATCHING values for each frequency. Try to better 1% for caps and better than 0.3% for resistors.
Polyester are rarely available in 1%.
1% polystyrene used to be readily available, but more difficult now.
1% or 2% polypropylene are available at more cost and in the larger values in bigger packages.
1% Mica are available in smaller values but have the wrong pin pitch.
No easy answer.
BTW,
single pole fiters can get away with 5%, 10% is rubbish,
as the filter steepness increases the required accuracy increases to match.
for 2pole you may get away with 2% and for 4pole 1% but I would go for closer matching, absolute accuracy is not required for frequency, but it is needed for gain (just as well DMMs are more acurate on DC voltage, pity about the resistance accuracy)
Found a cap -- almost
This is BC Components metallised polyprop filter caps, 416 series. 2% tolerance. The 100nF has footprint of 6x 7.2mm, lead pitch 5mm. Digikey part number BC2054-ND. Sold for $0.63 each.
This one is a shade shorter in length and half a millimetre wider in fatness, than what the board seems to allow. Maybe it'll do, maybe it won't.
What do the rest of you use?
This is BC Components metallised polyprop filter caps, 416 series. 2% tolerance. The 100nF has footprint of 6x 7.2mm, lead pitch 5mm. Digikey part number BC2054-ND. Sold for $0.63 each.
This one is a shade shorter in length and half a millimetre wider in fatness, than what the board seems to allow. Maybe it'll do, maybe it won't.
What do the rest of you use?
I guess Bob and Jens had some cap in mind when they did their designing? What were they planning to use? And I don't remember this factor being discussed in the threads either. Is it only me who's struggling to find good caps? Some people have posted photos of populated XO boards, I wonder what they used.AndrewT said:
If you can live with only 10 nf, Wima FKP2 1% will fit.
I have also used Panasonic ECQ series 2% polyprops - they are a bit big at 100 nf but if you can live with the lead length of them standing above (or below) the board they sound nice.
Otherwise I pull out the big bag of 5-10% caps and my meter and pick matching caps.
The cap footprint was a tradeoff between available board space and the functions available on the board. One of the design constraints was to make the board fit the free version of Eagle's 80 x 100 mm footprint. I would have preferred two all pass sections, but I cannot cope with SMDs...
I have also used Panasonic ECQ series 2% polyprops - they are a bit big at 100 nf but if you can live with the lead length of them standing above (or below) the board they sound nice.
Otherwise I pull out the big bag of 5-10% caps and my meter and pick matching caps.
The cap footprint was a tradeoff between available board space and the functions available on the board. One of the design constraints was to make the board fit the free version of Eagle's 80 x 100 mm footprint. I would have preferred two all pass sections, but I cannot cope with SMDs...
Who sells these?BobEllis said:
A bit big??? They are huge by the standards of this board. As per my Digikey data sheet, they're 17mm long, as against this board's limit of 7 or 7.5mm.
You mean let them float so high that they are above other components? Okay... that's a thought. Options increase.
Yes, that'd do too, I guess. I can just use ordinary polyester caps I guess.
Any pointers about which types of caps "sound" better?
Oh absolutely. I'd agree on that SMD front totally. Thank God you haven't gone SMD, though Jens seems quite comfortable with the idea.
And about Q less than 0.5, do I just cascade stages as AndrewT was suggesting?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.