cotdt said:
I think you got it backwards - this should be the final board, because adjustments are difficult.
C-E filters are just a standard hi or low pass with a deep high Q notch an octave or so away. I'm working on a writeup, but you can do a single channel 2 way CE with one board.
SvErD said:
I need to upgrade my paypal account to accept credit cards. at the moment I cannot. The upgraded account will cost a couple dollars per transaction, reducing the amount to DIYaudio.
BobEllis said:
So if I wanted very steep C-E filters, I would use 2 boards for 4th order, and a notch an octave below it? How many boards do I need in order to do a dipole speaker system?
My other question is whether these boards can be used for real speakers, or only for testing purposes. Everyone tells me that active crossovers can only be used for testing.
Are you the same cotdt in the HTGuide forum?
this board will make a Cauer-Elliptic filter that approximates 8th order initial roll off and ends up at fourth order. Two boards needed for a stereo 2 way. Four boards needed for a stereo three way.
Yes, I intend to make finished/real speakers with these boards, they will be fine.
What may be confusing you is that some speaker design software has a crossover emulator - it allows you to simulate a passive crossover using a sound card and multiple amps. Probably not what you want for finished speakers, but that does not preclude using active speakers. as "real" speakers.
this board will make a Cauer-Elliptic filter that approximates 8th order initial roll off and ends up at fourth order. Two boards needed for a stereo 2 way. Four boards needed for a stereo three way.
Yes, I intend to make finished/real speakers with these boards, they will be fine.
What may be confusing you is that some speaker design software has a crossover emulator - it allows you to simulate a passive crossover using a sound card and multiple amps. Probably not what you want for finished speakers, but that does not preclude using active speakers. as "real" speakers.
Active Cauer-Elliptic filter design in 20 minutes
Some have wondered how to create an active analog Cauer-Elliptic filter. Jon Marsh over on the HTGuide board uses passive C-E filters in his designs to push tweeters down to their low limits, with reportedly excellent results. I've made some up in a simulator and thought I'd share my results.
First off, I haven’t got a clue how to run through the math to determine component values. Lancaster’s Active Filter Cookbook offers a hint, and even a topology for a second order C-E filter. He uses a state variable filter to make the notch, and the book offers no other notch. I plugged it into a circuit simulator and played with component values so I could see how it worked. The trouble with his second order was that I could not get Jon Marsh’s steep initial roll off and keep the bounce low enough to be useful. I had similar results with Jason Cuadra's topology.
Just as Butterworth Filter describes a response shape rather than a topology, Cauer-Elliptic filter describes a response that has a steep initial roll off and somewhere in the stop band the response rises to some level and ends up rolling off at the basic filter rate. For the purposes of an active XO, we want a response that meets Jon Marsh’s criteria - ~8th order LR initial roll off and a bounce that stays at least 50 db down.
Being a “rip off other people’s building blocks” artist, I figured I could get there with a fourth order filter and a notch filter. I used unity gain Sallen-Key topology. After spending the better part of a day on it, I got there with acceptable passband response.
I started with a standard LR4 filter and found that the Q of the filter sections need to be much higher than standard because the notch causes a roll off earlier than expected. You’ll see a little ripple (~1 dB) either side of the XO, but that is the price you pay for the steep initial roll off.
Interestingly, as I was adjusting values to see which one has what impact on transfer function, I stumbled onto a curve that provided built in baffle step compensation. It didn’t match the baffles I was looking to use for a RS150/27TDFC MT, and after a few iterations, I gave up and went back to flat pass band response.
I can’t offer a spreadsheet to generate C-E filters but the nutshell guidance when you sit down with a simulator is:
Start with a standard 4th order crossover topology with HP and LP sections followed by a notch filter of your choosing. (state variable, LCR, Active inductor, gyrator)
As Jon suggested, a notch about an octave away from the desired frequency works pretty well. It needs to be fairly high Q – I ended up at 6.5, with 32.5 dB of cut.
Set one section of your HP and LP filters to give an F3 of the desired crossover frequency with a Q of 1.6. Set the second section at ~15% higher (for HP) or lower (for LP) with a Q of 1.7. e.g. for a 2 KHz XO, use 2300 and 1700.
Play with filter component values to adjust the Q to reach the desired response. Increasing the F3 spread of the offset sections decreases pass band ripple, but you'll have to reduce the notch attenuation to compensate, making the bounce higher.
I haven't figured out how to sum responses or even display phase in the demo version of 5Spice I use, so this may not sum correctly. I'd appreciate any guidance.
The schematic - a mess because the program is crippled by allowable drawing size. I didn't make a serious attempt to use component values that are actually available. I used the spreadsheet poster earlier to calculate component values. Since when I originally thought of using this I was heading for a moxlite board, I used unity gain topology.
The resulting response meets the goal of 8th order initial rolloff and keeps the bounces more than 50 dB down. This will need to be tailored to your drivers' responses.
Some have wondered how to create an active analog Cauer-Elliptic filter. Jon Marsh over on the HTGuide board uses passive C-E filters in his designs to push tweeters down to their low limits, with reportedly excellent results. I've made some up in a simulator and thought I'd share my results.
First off, I haven’t got a clue how to run through the math to determine component values. Lancaster’s Active Filter Cookbook offers a hint, and even a topology for a second order C-E filter. He uses a state variable filter to make the notch, and the book offers no other notch. I plugged it into a circuit simulator and played with component values so I could see how it worked. The trouble with his second order was that I could not get Jon Marsh’s steep initial roll off and keep the bounce low enough to be useful. I had similar results with Jason Cuadra's topology.
Just as Butterworth Filter describes a response shape rather than a topology, Cauer-Elliptic filter describes a response that has a steep initial roll off and somewhere in the stop band the response rises to some level and ends up rolling off at the basic filter rate. For the purposes of an active XO, we want a response that meets Jon Marsh’s criteria - ~8th order LR initial roll off and a bounce that stays at least 50 db down.
Being a “rip off other people’s building blocks” artist, I figured I could get there with a fourth order filter and a notch filter. I used unity gain Sallen-Key topology. After spending the better part of a day on it, I got there with acceptable passband response.
I started with a standard LR4 filter and found that the Q of the filter sections need to be much higher than standard because the notch causes a roll off earlier than expected. You’ll see a little ripple (~1 dB) either side of the XO, but that is the price you pay for the steep initial roll off.
Interestingly, as I was adjusting values to see which one has what impact on transfer function, I stumbled onto a curve that provided built in baffle step compensation. It didn’t match the baffles I was looking to use for a RS150/27TDFC MT, and after a few iterations, I gave up and went back to flat pass band response.
I can’t offer a spreadsheet to generate C-E filters but the nutshell guidance when you sit down with a simulator is:
Start with a standard 4th order crossover topology with HP and LP sections followed by a notch filter of your choosing. (state variable, LCR, Active inductor, gyrator)
As Jon suggested, a notch about an octave away from the desired frequency works pretty well. It needs to be fairly high Q – I ended up at 6.5, with 32.5 dB of cut.
Set one section of your HP and LP filters to give an F3 of the desired crossover frequency with a Q of 1.6. Set the second section at ~15% higher (for HP) or lower (for LP) with a Q of 1.7. e.g. for a 2 KHz XO, use 2300 and 1700.
Play with filter component values to adjust the Q to reach the desired response. Increasing the F3 spread of the offset sections decreases pass band ripple, but you'll have to reduce the notch attenuation to compensate, making the bounce higher.
I haven't figured out how to sum responses or even display phase in the demo version of 5Spice I use, so this may not sum correctly. I'd appreciate any guidance.
The schematic - a mess because the program is crippled by allowable drawing size. I didn't make a serious attempt to use component values that are actually available. I used the spreadsheet poster earlier to calculate component values. Since when I originally thought of using this I was heading for a moxlite board, I used unity gain topology.
The resulting response meets the goal of 8th order initial rolloff and keeps the bounces more than 50 dB down. This will need to be tailored to your drivers' responses.
Nisbeth said:Bob, YGM in about 5 mins. Consider this post a request for using CC-paypal as payment
/U.
Nisbeth, you can probably deposit money in your Paypal account by doing a bank transfer (do so preferably when the $ is low).
In Germany, we can now use direct debit. Problem is, the moment I activated this, I could no longer pay by credit card. Cost-wise, it makes no difference to me, Paypal used to draw on my card in € even for payments in $ and charge about 3% for the conversion. I still pay the 3% conversion charge with direct debit.
If Bob were to upgrade his account to Premier, he'd incur a 3% spread for all incoming payments, whether they are from a positive balance, direct debit or credit card. So what this means is that Paypal gets to keep 3% twice.
Maybe it is possible for him to open two different accounts that go two to different e-mail addresses and bank accounts, otherwise, upgrading would be a loss to about everybody else.
Greetings,
Eric
CAPSLOCK (sorry couldn't resist
) Also pointed out to me the small envelope GPM rate. I figure that I can get 20 boards in that package, so for outsde the US orders 20 or less shipping is $5.25 It is still $9.50 if you want more than 20.
For those who have already paid higher shipping, I will refund the difference after the rush of orders dies down. If you want I can donate the difference to the forum. let me know if you wnat to give it to the forum, the default will be a refund.
) Also pointed out to me the small envelope GPM rate. I figure that I can get 20 boards in that package, so for outsde the US orders 20 or less shipping is $5.25 It is still $9.50 if you want more than 20.For those who have already paid higher shipping, I will refund the difference after the rush of orders dies down. If you want I can donate the difference to the forum. let me know if you wnat to give it to the forum, the default will be a refund.
Payments are starting to trickle in. I need to get the order placed to keep the discount. I will buy enough boards to cover the interest expressed on the WIKI, plus a small overage.
I'd appreciate prompt payment to reduce my risk.
If you have changed your mind, please update the WIKI to reflect it.
Thanks,
Bob
I'd appreciate prompt payment to reduce my risk.
If you have changed your mind, please update the WIKI to reflect it.
Thanks,
Bob
Bob,
I just found out about this group buy from your posting over on the HTGuide Forum. I don't know a lot about this forum or procedures, but I would really like to get in on the circuit board buy. I do not have a verified PayPal account, only one that uses my credit card, is there any other way to pay? I could probably have my account changed, but I don't know how long that takes.
I'm playing catch up here in this thread, so I haven't read everything, but I noticed you referenced a writeup by Jens regarding filter design one and two or something like that. I've tried searching and even went to Jens website, but I don't see where this information is available. If it is available, could you point me in the right direction?
Thank you,
Brian Walter
I just found out about this group buy from your posting over on the HTGuide Forum. I don't know a lot about this forum or procedures, but I would really like to get in on the circuit board buy. I do not have a verified PayPal account, only one that uses my credit card, is there any other way to pay? I could probably have my account changed, but I don't know how long that takes.
I'm playing catch up here in this thread, so I haven't read everything, but I noticed you referenced a writeup by Jens regarding filter design one and two or something like that. I've tried searching and even went to Jens website, but I don't see where this information is available. If it is available, could you point me in the right direction?
Thank you,
Brian Walter
Brian,
send me an email and I'll give you the address to send a money order.
Jimbo,
Thanks for the help.
The spreadsheet posted earlier uses the formulas in Jens' manuals (an excellent read for practical filter design). The MOXLite tab does the calculations for for the unity gain Sallen Key topology. You can do that with these boards, but they are also capable of using the equal component value topology, which is a bit more flexible. I will update the spreadsheet to include equal value calculations and post it once I get the order placed.
send me an email and I'll give you the address to send a money order.
Jimbo,
Thanks for the help.
The spreadsheet posted earlier uses the formulas in Jens' manuals (an excellent read for practical filter design). The MOXLite tab does the calculations for for the unity gain Sallen Key topology. You can do that with these boards, but they are also capable of using the equal component value topology, which is a bit more flexible. I will update the spreadsheet to include equal value calculations and post it once I get the order placed.
Active Filter Cookbook...
For anyone interested I just purchased the Active Filter Cookbook by Don Lancaster through his own distribution. It is the 1995 printing for $19.95, not the volume '7' for 30+ dollars, I searched a bit a discovered from other reviews that there is no real new information in the 'seventh' AND it is autographed by the guru himself...
http://cgi.ebay.com/AUTOGRAPHED-Don...019742QQcategoryZ2228QQtcZphotoQQcmdZViewItem
For anyone interested I just purchased the Active Filter Cookbook by Don Lancaster through his own distribution. It is the 1995 printing for $19.95, not the volume '7' for 30+ dollars, I searched a bit a discovered from other reviews that there is no real new information in the 'seventh' AND it is autographed by the guru himself...
http://cgi.ebay.com/AUTOGRAPHED-Don...019742QQcategoryZ2228QQtcZphotoQQcmdZViewItem
Paypal eats the cake and keeps it
Well, this is the first payment outside the Euro zone that I initiated after getting my bank account confirmed for direct debit. And guess what: neither can I use my credit card nor direct debit, they insist on my transferring the money first, which means Bob will have to wait two business days for the transfer to clear. They still charge me 3% above interbank rate for the €/$ conversion.
Just for comparison:
- my credit card company charges me 0% for payments made in €
- my credit card company charges me 1.5% for conversion fees (e.g. €/$)
- my bank charges me 0% for outgoing transfers, direct debits or incoming payments
- my bank charges me about 0.1% for currency conversion on outgoing electronic money transactions
I'm glad Bob still maintains a non-premier account, otherwise Paypal would get to keep 6% at zero risk and near zero cost. Now they get to keep 3%.
Better donate the surplus to diyaudio, although they never kept up to their promise of a custom title for my last donation
Well, this is the first payment outside the Euro zone that I initiated after getting my bank account confirmed for direct debit. And guess what: neither can I use my credit card nor direct debit, they insist on my transferring the money first, which means Bob will have to wait two business days for the transfer to clear. They still charge me 3% above interbank rate for the €/$ conversion.
Just for comparison:
- my credit card company charges me 0% for payments made in €
- my credit card company charges me 1.5% for conversion fees (e.g. €/$)
- my bank charges me 0% for outgoing transfers, direct debits or incoming payments
- my bank charges me about 0.1% for currency conversion on outgoing electronic money transactions
I'm glad Bob still maintains a non-premier account, otherwise Paypal would get to keep 6% at zero risk and near zero cost. Now they get to keep 3%.
Better donate the surplus to diyaudio, although they never kept up to their promise of a custom title for my last donation
Nisbeth said:
Unfortunately I can't use my bank account:
I don't want to be the reason why everyone else has to pay 3% extra, but still....
Bob, can you open another account for cc-payments?
/U.
Nisbeth, if this is is a problem, transfer the money to my bank account and I will transfer it to Bob via Paypal. According to EU legislation, transfers between EU countries must be charged at the same rate as national transfers. As we do everything by direct transfer or debit, this means free as far as the German part is concerned. My experience has been that transfers to A, UK, NL are also free. Not sure about countries like F who are more like USA (lots of personal checks or credit card transactions).
Last time I used my credit card in Scandinavia, DKK had a 1.5% conversion charge, SKK did not, have no idea why.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.