XO3: The 1/2/3-way Active Crossover/filter PCB. Design suggestions welcome! - diyAudio
Go Back   Home > Forums > Source & Line > Analog Line Level
Home Forums Rules Articles diyAudio Store Gallery Wiki Blogs Register Donations FAQ Calendar Search Today's Posts Mark Forums Read

Analog Line Level Preamplifiers , Passive Pre-amps, Crossovers, etc.

Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Thread Tools Search this Thread
Old 29th September 2010, 09:43 AM   #1
diyAudio Member
s3tup's Avatar
Join Date: Dec 2004
Location: Israel
Default XO3: The 1/2/3-way Active Crossover/filter PCB. Design suggestions welcome!

Hi all!

I'm on finishing stages of crossover PCB development, and need yours input on my design.

The crossover board consists of:
- Input LPF for HF noise elimination (passive stage)
- Gain/input buffer with limited bandwidth feedback
- Bafflestep
- Single global notch + buffer after the notch
- LP for woofer
- HP for midrange
- Variable phase for midrange
- LP for midrange
- HP for tweeter
- Variable phase for tweeter
- Each output stage for woofer,mid,tweeter consists of single notch + output buffer.

The filters chaining is as described at LinkwitzLab.com - the tweeter signal travels the following path:
- Input - bafflestep - notch - hp mid - phase mid - hp tweeter - notch tweeter - output buffer, so there is no phase errors between woofers, midranges and tweeters.

Currently i'm doing the following tasks:
- Upgrading the PSU - from simple chunked-up capacitors to LM317/337 regulators on board (they shouldn't produce much heat)
- Adding optional balanced input at input stage (simple single opamp scheme)
- Adding ability to bypass/reroute all the sections (say, you want to use it for full range speakers, but with 3 notches and no phase shifter? no problem!, just chain-up the output notches and bypass the phase shifter/LP section).

- All the HP/LP filters are 2/4-th order (selectable), Sallen-Key topology.
- All the passive elements footprints allow placing both SMT and thru-hole components
- All the filtering capacitors could be up to ~40mm length, 5mm wide
- All the filter resistors are built from 4 paralleled/chained SMTs, or 2 paralleled thru-holes - for getting exact values from limited E12/E24 values range. (the values finder will be supplied as a windows program or web service - you type the required value, select E12/E24, and get the closest value from chained/paralleled resistors).

As a side note:
Each opamp bypassing circuit requires:
- 1206 NP0 ceramic caps (value range is around 1n-3n, or anything you have around, that is not too small for the application)
- 5mm diameter electrolytic cap (anything you have, rated at 16v+, 1uf+).
- Opamp sockets are required - you have to use the gold=plated ones with round pins - they stand out of PCB for a 1-2mm - this space is required for placing PSU bypassing SMT caps.

And... everything is stuck on 4x4in PCB! You'll need 2 boards for stereo.

Design highlights:
- Two-sided ground plane pours
- "Starry" power delivery to ICs
- Both SMTs and thru-hole components are possible in most cases, in various sizes (good for huge film caps)

Pricing guestimation? 10$-20$ for single board.
Ordering? You'll be able to buy the boards from seeedstudio.com, after i test/fix the first PCB batch for design errors.

Attached are:
- Top/bottom board views without ground pours (to show the traces on opposite layers).
- Schematics in PDF (please disregard the component values, they are for layout purposes only).

Limitations... - board size, should be 4x4in - hence the tight layout.
Attached Images
File Type: jpg TS.jpg (699.7 KB, 1403 views)
File Type: jpg BS.jpg (526.4 KB, 1244 views)
Attached Files
File Type: pdf xo3.pdf (89.4 KB, 486 views)

Last edited by s3tup; 29th September 2010 at 09:57 AM.
  Reply With Quote
Old 29th September 2010, 10:02 AM   #2
AndrewT is online now AndrewT  Scotland
diyAudio Member
Join Date: Jul 2004
Location: Scottish Borders
read up MOX and threads that it spawned.
regards Andrew T.
Sent from my desktop computer using a keyboard
  Reply With Quote
Old 1st October 2010, 04:27 PM   #3
diyAudio Member
s3tup's Avatar
Join Date: Dec 2004
Location: Israel
Parametric EQ both subtracting and adding is a nice feature, will be added.
Rerouting the default signal path will be added too, yet i doubt it will be useful, as there are already enough features on each signal thread, and the topology is ok (as Linkwitz described on his site).

I'm playing right now with software, that going to simulate the crossover behavior (plotting graphs based on RC values in the circuit, and generating RC values based on sliders position). Going to be coded in VB.NET, and will require .NET framework, 3.0-3.5.
Say, you want to place a notch at some frequency. You move the F, then choose Q and depth, click "simulate" and get a picture how it will look like on frequency response.
A further step - is to load frequency response of each speaker in their enclosure, and tweak it visually by sliders, and then to get values for the implementation circuit.
Summing the response is a bit tricky, as there are phase/directivity issues of driver placement/size, so the flat frequency response won't tell the whole picture.
Maybe i'll do the "ideal" summed response, assuming every speaker get perfectly aligned phase, and then to get that response, user will need to tweak phase shifters to align phases in the real world.

The last, and furthermost option would be creation of software EQ/channel splitter, that will be able to play simulated filers using computer multichannel sound card. That's voodoo for me at this step of my evolution
I'm okay on hardware DSPs, FIR/IIRs, yet i have no idea how to implement it on regular CPUs with regular C/C++. Probably some audio buffer processing.
Then, designing a speaker will be an easy task - throw in some amplifiers, build enclosures, play with filers in software, build them in hardware - and forget about filters existence as your darkest nightmare.
  Reply With Quote
Old 2nd October 2010, 06:34 PM   #4
diyAudio Member
s3tup's Avatar
Join Date: Dec 2004
Location: Israel
Ok, as promised: The software tool for crossover simulation.
Just a screenshot...

What was done:
- Calculated transfer functions for HP, LP, Bafflestep, Notch for the circuits i have. Each trandfer functions receives a R,C,f parameters as input and returns a magnitude for particular combination of resistors, caps and frequency, in complex way.
- Made some kind of "platform" for transfer function analysis - graphing, frequency sweep, layered display of graphs.

To do:
- Calculate a phase transfer function.
- Add some controls! Changing Q, F, depth e.t.c for filters > calculation of actual R,C values > Simulation.
- Chaining functions. Probably by complex multiplication, and then plotting as it was single function.
- Frequency response loading. That will be a hard thing to do. I'd like to split it into 2 steps:
- - Graphical graph loading and analysis - datasheet graphs could be used, as well as any other.
- - Data import from major measurement programs (LEAP, ARTA, LspCAD etc).
- Phase, phase phase. I have to think what to do with it... Importing the phase graphs of measured speakers should be nice, and alteration of the graph based on phase filter response - even better.
Attached Images
File Type: gif Response.gif (33.5 KB, 1072 views)
  Reply With Quote
Old 6th October 2010, 01:05 PM   #5
diyAudio Member
s3tup's Avatar
Join Date: Dec 2004
Location: Israel
I wonder whether it's sufficient to put just a single first order all-pass circuit on tweeter output? Won't i almost always need at least two all-pass stages?

In the other hand, the woofer-midrange physical offset shouldn't affect the frequency response, as the phase difference between woofer-midrange acoustical centers are really small...

So it's better to use both all-pass networks just on tweeter's filter...?
  Reply With Quote
Old 1st November 2010, 03:33 AM   #6
Francec is offline Francec  Australia
Join Date: Jul 2006
Location: Port Melbourne, Australia
How is progress? Things seem to have slowed down.

  Reply With Quote
Old 3rd November 2010, 12:48 AM   #7
diyAudio Member
Join Date: Mar 2007
Location: California
Default UAF42 board for 2nd order filter, equalizer, active crossover

I have finally arrived at a scheme using the UAF42 IC from Burr Brown for line level crossovers in active speakers. Each UAF42 IC has simultaneous outputs for second order LP, BP, and HP functions. What is very nice is that, using three multi-turn trim pots, I can independently adjust Q and crossover frequency of each section. One circuit can be used for any 2nd order filter function within the audio band, making it more or less "universal".

I recently realized a way to used the bandpass output to make a single band, variable Q, variable frequency, second order equalizer section. The Q and center frequency are varied using the UAF42s controls, and another circuit takes the BP output and turns it in to a boost or cut.

In addition to these functions, I have a separate board that will handle input (balanced or single ended), and baffle step with a variable center frequency and attenuation.

These are pretty much all the building blocks that you need to make active crossovers for speakers. The second order sections can be cascaded to make higher order filters and the equalizer can be used to notch out resonances, fill in dips in the frequency response, remove the hump in the baffle step response, etc.

I don't have custom PCBs for these, so I am using a general IC prototyping board that has some traces on it. Several of these boards are needed depending on how many filter stages and equalization bands are used. Since the prototyping boards are all the same size, I can easily mount them together inside a speaker somewhere as a stack.

If any one is interested in collaborating on a run of custom PCBs for this kind of thing, please contact me. I don't have any experience in PCB design, routing, etc. for low level signals so I am a little out of my league. That's why for now I am only using the pre-fab boards. But I think a lot of people would be interested in this kind of PCB because amplification that you can put in a speaker is cheap and plentiful. The one thing that is missing to allow people to easily build their own active speakers are the line-level crossover boards. Since I have figured out a way to make a universal crossover with the UAF42, which I can get on Ebay for under $2 US each, this would be a great way to get more people interested in to speaker building. If you are a company selling chip amps, wouldn't this be a great match to your product line???

  Reply With Quote
Old 3rd November 2010, 01:14 AM   #8
diyAudio Member
abraxalito's Avatar
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Blog Entries: 145
Send a message via Yahoo to abraxalito
Not at all wishing to rain on your parade, but a few things about the UAF42 on the surface don't bode too well for getting good sound. Firstly the THD is 0.1% and only specified at 1kHz and unity gain. There's the likelihood of more severe HF distortion than this and it'll also get worse at higher gains. The GBW product of the opamps looks barely adequate for audio at 4MHz. The quiescent current for 4 JFET input amps is a fairly measly 6mA - that's just 1.5mA per opamp. Finally, as the list price for these is over $7 (1k), how do you know the eBay parts aren't fakes?
The heart ... first dictates the conclusion, then commands the head to provide the reasoning that will defend it. Anthony de Mello
  Reply With Quote
Old 3rd November 2010, 03:15 AM   #9
diyAudio Member
Join Date: Mar 2007
Location: California
Default that's not what I measured

Thanks for mentioning the specs from the datasheet. Honestly, I don't believe that that the bandwidth is a problem. You are not using the chip as a gain stage, it's a filter and the gain will be unity or less, so I think that there is plenty of GBP.

On the other hand, the THD spec would be worrisome, except that I have built and measured a circuit using this chip. The actual circuit construction was pretty ugly, I didn't do any bypassing, I took the measurements without any special attention to shielding, etc. and I still got way less distortion than 0.1%, in fact it is less than 0.01% at all frequencies and is generally below 0.005%. OK, I just noticed that I only measured 2nd and 3rd order distortion, and the measurements only go up to 500Hz (this was a low frequency HP filter), but you get the idea. See image below...

Here is a graph of the distortion of the circuit from my project using the UAF42:

Click the image to open in full size.


Last edited by CharlieLaub; 3rd November 2010 at 03:17 AM.
  Reply With Quote
Old 3rd November 2010, 03:37 AM   #10
diyAudio Member
abraxalito's Avatar
Join Date: Sep 2007
Location: Hangzhou - Marco Polo's 'most beautiful city'. 700yrs is a long time though...
Blog Entries: 145
Send a message via Yahoo to abraxalito
Thanks for sharing the plot. It looks like you've a very low fundamental frequency there - with the harmonics below 100Hz, you might be looking at 10 - 30Hz for the fundamental. That's not inconsistent with the datasheet figure at 1kHz, here's why.

The opamps in your chip look to have very similar performance characteristics to the TL074 (with the exception of THD - on the TL074 its low, around 0.003% at 1kHz.) So its reasonable to assume the OL gain curve is going to look similar to the TL074. This part shows an OL -3dB bandwidth around 50Hz. There's a factor of 20 between this and 1kHz, so the distortion will be the same factor higher there. That's for me the explanation of the difference between 0.005% and 0.1%.
The heart ... first dictates the conclusion, then commands the head to provide the reasoning that will defend it. Anthony de Mello
  Reply With Quote


Hide this!Advertise here!
Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off

Similar Threads
Thread Thread Starter Forum Replies Last Post
Active Digital Loudspeaker Crossover - Part Suggestions solicitation speakerguy79 Digital Line Level 0 2nd December 2009 04:26 PM
active crossover state variable filter Altecboy Multi-Way 0 23rd June 2009 01:31 PM
<AUS> B3S active crossover design - Value woofer suggestions? TheSeekerr Multi-Way 0 25th August 2008 12:46 PM
Generic active filter / crossover... board... thingie... JohnR Parts 5 14th October 2002 10:13 PM
Opamps for active crossover/filter harvardian Solid State 10 3rd January 2002 02:01 AM

New To Site? Need Help?

All times are GMT. The time now is 07:41 AM.

vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2017 DragonByte Technologies Ltd.
Copyright 1999-2017 diyAudio

Content Relevant URLs by vBSEO 3.3.2