electronic crossover cap resistor ratio

[ostie01]

Hi, I want to build a 4way 24 dB linkwits-riley electronic crossover and would like to know what is the best ratio between capacitor and resistor.Should I go with high value capacitor like 200 or 300 nF and smaller value for resistor or should I use smaller capacitor value like 1 to 50 nF and higher resistor value.Also I have a good quantity of polystyrene capacitor value between 1nF and 100nF and polypropylene value between .002uF and 1uF.I will use the NE5532 dual OP amp.Should I put some low value cap between the plus and minus voltage pins.I someone have some good advices to share,I will appreciate. Thanks

[AndrewT]

Hi,
look up the ESP site.
He has various articles and designs for crossovers.
Download his L-R design software. It will let you select Rs and Cs at will to suit each frequency.

There are many other design software packages that will help you choose your frequency components.

Aim to keep your Rs at about 10k +-2k, provided you can obtain accurate capacitors in the range you need. Aim for +-2% for caps or better (5% is too wide unless you can match from a batch) and remember to keep the values consistent between lopass & hipass as well as between channels.

[consort_ee_um]

I would not be so tight on the resistor value as Andrew. I think 5k6 to 56K would be OK for the resistor value. If you go too high in resistance I found you could get hum pickup and too low a value the op amps will have trouble driving. One problem usually is that high value precise tolerance capacitors can be difficult and expensive to find. So I normally pick a cap value and check if the resistor value is in the range above. However you already have some capacitors.
There is a nice on-line calculator here:
http://www.beis.de/Elektronik/Electronics.html

[MBK]

... also to consider, stay clear of very low capacitor values. Around 1 nF or so (which implies high-ish R's), parasitic capacitances become significant factors. Best stay above 10 nF.

[AndrewT]

Hi,
I'm a disbeliever.
Cap range should be 1nF to 100nF. This should cover the audio range. A 10pF parasitic is 1%. Careless layout could achieve this and careful layout could achieve 4 to 10 times better.

Sub-bass hipass at 10Hz to 20Hz will need bigger, similarly treble lopass above 20kHz may need smaller but try to avoid it by using smaller R.

But if you stay wideband at the extremes, 20Hz & 20kHz never need filtering actively.

[MBK]

Speaking from experience, not from belief...

I have a MOX active X-O from this forum. There are several ranges of frequencies. The lowest uses a 1 nF cap. The next lowest uses a 10 nF cap. The range with the 1 nF cap gave problems to all users: ca 1 dB gain offset compared to the other frequency ranges. The problem was traced (sic!) to parasitic capacitance w/regard to the ground plane.

The 10 nF range was fine.

Of course if the layout had been perfect etc etc. But more often than not, it isn't

[consort_ee_um]

As you say something wrong with your implementation. I have used 1nF for more years than I care to remember with no problems.
Incidentally why are 1% polystyrene caps which I used for this application, so expensive now?

[richie00boy]

I have also had no problems whatsoever with 1nF and other low values, but yes in general it's good for DIYers to try to keep above 10nF for the reasons mentioned.

[AndrewT]

Hi Mbk,
MOX was intended as a development tool. Maybe all those selectable caps, resistors, connectors and associated traces had a bit of excess capacitance. But this should only affect the Q and F3 of the highest crossover frequency.

I stand by my original recommendations 1nF to 100nF and 8K to 12K (range 50Hz to 10kHz) with further provisos. When using unity gain filters (as Moxlite) you need to double some of the resistor and cap values to maintain the correct Q. If you need to go outside that frequency range then a wider range of values have to be used.

[MBK]

I just meant this 10 nF thing as a rule of thumb, allowing for less-than-perfect implementations etc. Without a doubt, and recognized belatedly, the original MOX had a flaw in its implementation, not too serious ( it didn't affect Q, strangely, but it affected the gain of all frequencies using the 1 nF cap, by almost exactly 1 dB). Equally, proper layout vastly expands the range of useable values.

Still - I now keep my caps above 10 nF when I can. Once burned, ... you know!

YMMV...