Introduction to designing crossovers without measurement

Member
Joined 2023
Paid Member
Ok....this isthe crossover pcb of the Kef Reference Model One from the 90's All parts are Bennic with black caps being bipolar. Is there any use of replacing the resistors and Caps? Any advise on replacement brands and types appreciated. I know...there are loads of them and I'm trying to read and absorb as much as possible:)
crossover.PNG
 
diyAudio Moderator
Joined 2008
Paid Member
Can I assume this is from a speaker that is functioning correctly?

Your resistors are wirewound types. These are known to be fairly reliable. The most common failure mode in a speaker crossover would be overheating. Usually you can tell when this has happened because they and the board can look and smell burned. I would expect them to be OK and if you have any doubts you can measure them.

All the capacitors would be bipolar because polarised capacitors won't be suitable without DC. Modern electrolytic capacitors like this are much more predictable and reliable than they were in the past. One thing that can affect this is heat. Ideally capacitors don't get hot when you use them like resistors do, and in this case there is little to suggest they've been near anything hot. Another way they can show failure is by the package bulging or leaking. If in doubt, measure that their values are in tolerance.

Capacitors can't help but have a small, almost insignificant resistance in their leads and structure, and with electrolytics this is often a little higher. Sometimes when people change electrolytics for film type capacitors they find differences, such as the treble being too bright for example due to the slight changes in resistance. So if you need to change them, sticking with the same types can reduce this possibility and would ensure that they fit on the board.
 
Member
Joined 2023
Paid Member
Thx Allen,...well actually the speakers sound really well but then again I not a really experienced listener. So to be honest...I don't think I need to change anything besides the fact that the speakers are 30 years old so I was just thinking of replacing them with at brandnew ones. I'm gonna reverse engineer the scheme but I don't have to tools (yet) to measure the inductors nor capacitors in order to determine whether or not they are of the mark in terms of the indicated values.

reg

w.
 
Hi Allen,

I have been doing a bit of delving with my £79 multimeter which has an inductance scale.

I was surprised to find it can measure inductance of woofers on the 2mH scale passably well:

Measuring Inductance in Loudspeakers.jpg


Notice it measures this stock 0.25mh Coil exactly, even with 6.8R resistor in series! It works with 1.5mH too.
It uses a test tone around 1.5kHz in my estimation and must measure phase change.

This gave me the idea to measure some 8 ohm woofers and tweeters....

8" Sony E44 woofer 0.34mH and 6.5R DCR.
8" Monitor Audio R300-MD Elac woofer 0.34mH
6" Elac Mordaunt Short MS15 woofer 0.41 mH
8" Visaton W200S-8 woofer 1.18mH
8" Wharfedale Super Linton woofer 0.21mH and 4.1R DCR

The Visaton is a known quantity, high inductance quoted at 2.2mH in the specs:
https://www.visaton.de/en/products/drivers/woofers/w-200-s-8-ohm

I can't explain the discrepancy in value for the visaton, but the meter certainly can distinguish high inductance woofers from low inductance ones.

Thus my tuning by ear for the Sony woofer and tweeter was almost exactly on the mark for impedance correction and a 6dB rolloff commencing around 1.5kHz for a 3kHz crossover.

Sony Drivers E44 speaker.jpg


All I had to go on was this frequency response:

Sony E44 Bass.jpg

Originally the speaker just had a 3.3uF on the tweeter. It sounded rough!

I came up with this 3kHz crossover with tweeter loudness adjusted by ear:

Sony E44.jpg


With the benefit of hindsight, I realise I hit almost exact impedance correction on the Sony bass, the Visaton correction being 7.5R + 39uF!

Having an estimate of voicecoil inductance helps me decide if the bass can get away with a simple LC circuit or needs a coil and impedance correction.

Tweeter measurements were totally erratic. I think the (1.5 kHz?) test tone is hitting the tweeter Fs resonance and going wildly wrong.

For instance, my SEAS 19TAF/G which I run without ferrofluid gets 0.57mH when I would expect 0.05mH. Very odd.

Anyone with a good meter might try this with any woofers you have lying around, and then compare with the manufacturers specs.
 
Last edited:
  • Like
Reactions: 2 users
diyAudio Moderator
Joined 2008
Paid Member
Yes, inductance is inconsistent, and nonlinear, and noise during the test can affect the result, and cone movement from the test signal will alter the magnetic circuit.

However the Zobel conjugate is not terribly critical. If there ends up being a bump in the impedance because a less than ideal capacitor value was chosen then at least it should be slow and smooth. This may be audible, but also within the ability to tweak by ear due to it's low order.
 
Am I good, or am I good? :cool:

I loaded Sweepgen in Windows and compared the multimeter test-tone on the meter with Sweepgen at 1507Hz.

https://www.satsignal.eu/software/audio.html

Sweepgen 1507Hz Tone.JPG


My 1.5kHz estimate was spot on! I know my frequency scale by ear.

The splendid Mr. Troels Gravesen has done a lot of work with his bespoke JA-8008 8" woofer and the successor JA-8008-HMQ. You can find the specs at Jantzen Audio.

The newer driver is quite close to the desireable 8" SEAS W22NY001 with an L of 0.6mH, albeit with a Kapton former rather than a Aluminium one for the high Qms, and a livlier cone.

About 15-20 ohms at 10kHz corresponds to about 0.6mH inductance. It scales quite well for even higher inductance.

http://www.seas.no/index.php?option...8s-w22ny001&catid=49:excel-woofers&Itemid=359

I was pleased that his ideas were quite close to mine on a 3kHz woofer circuit on the JA-8008:

Troels Gravesen JA8008 woofer.jpg


Troels Gravesen TQWT Crossover.jpg

That is impedance correction and a 4.5kHz tank notch for the breakup on the bass, which hits about 20 ohms at 10kHz, so 0.6mH is reasonable.

I did find that smaller coils make for a livlier sound, albeit bass light. Most woofers seem to be around 0.6mH, except for the more exotic ones like the SEAS A26 which have a built-in 3mH inductance which avoids the need for an external bass coil altogether:

http://www.seas.no/index.php?option...re4x&catid=44:utv-prestige-woofers&Itemid=238

My feeling is I just need to multiply by a "Fudge Factor" of about X1.8 to bring my measurements into line with the common way of calculating Zobels.

Be interested in what other people get with known drivers if they have a good meter.
 
Hi Folks,
The question I have pertains to a satellite system with a subwoofer. I am trying to build a passive crossover for the sub and also for connecting to the two satellites. The crossovers for the satellites I made a few years ago based on the great work by Allen. I plan to use a 1st order crossover initially as subjectively the components seem to work well. The satellites have an impedance of about 30 ohms at the crossover of 150 hz. Should I try to tame the impedance as Allen suggests with a resistor or am I better to move the crossover up away from the bump? The bump appears with the drivers in the cabinet.
 
diyAudio Moderator
Joined 2008
Paid Member
There are four basic issues to consider when crossing satellite speakers to subs. One is whether to cross to the subs or to overlap with them. By crossing them to the satellites you increase the power handling of the system while increasing the bass extension, turning the speakers into three ways. When you instead overlap the subs and mains, you allow them to work together to handle room modes. It is possible to do both by adding further subs.

The second issue is whether you will use separate amps for the satellites and the sub(s). This can open up possibilities for the crossovers by doing some of the work before the amps, if you have the means to work with them.

The third is the combined response.. the natural bass rolloff of the satellites with the intended crossover response. You want something which you can match with the sub. You could either cross above the resonance as you would with a tweeter, or you can cross near the resonance and add it to the second order rolloff of a sealed box or the fourth order rolloff of a ported box.

As you've already noted, there can be impedance peak issues. The resistor may be easy to use but it may also be power hungry at lower frequencies. Capacitors and inductors don't consume power, and we can use them to advantage but it takes more work to find the right values. It can be better to work with a sealed box because the impedance peak will be simpler, and the total rolloff will not be as steep as for a ported box.

I am going to use simulated examples to show how to interact with the impedance. The simulated satellite woofer is going to start with a flat response so you can see only what the filter is doing.

This is the impedance on it's own, with a peak near 60Hz..

i1.png


This example shows the effect of the impedance on the response when using a single capacitor intended to cross 6dB down at 120Hz, next to the same thing without the impedance peak to deal with. In some situations you can compensate by changing component values but it won't help much in this case..

i2.png


Maybe the best way to approach this without measurements is second order. A second order filter can be less sensitive to impedance due to the parallel component, as well as rolling off more quickly. In the example here, the blue response is the target slope of second order at 120Hz. The red response is when using standard second order crossover values and the green response is achieved by reducing the value of the inductor which is in parallel with the driver. It is reduced to around a third of the original value as a way of compensating the impedance. This is good because higher values of inductance can be large and heavy, especially if you don't want them to have too high a resistance so the woofer doesn't overheat them..

i3.png


Finally, the string of three components on the right is the kind of circuit addition which will flatten the impedance peak. This kind of circuit can also be used for the tweeter if you can ascertain the values, which isn't too difficult to do with the help of a simulator. It is shown here with a first order filter but a second order filter could also be used.

i4.png
 
  • Like
Reactions: 1 users
Good evening, Im building a portable speakerbox with 2+20W amp with bluetooth.
Using two Scan Speak D3804 and two Focal 6 inch Polyglass 165 V2S. They are at 4 Ohm and around 93 dB. Gone try sinecap on the tweeters 6.8 Ohm in series and 0.12mH in paralell, with a 12dB filter for bass, 0.47mH in series and 8.2 Ohm in paralell. This is just for fun, just to have some music in the park.
Tweeters pointing forward and bass to the sides.

Frank
 
diyAudio Moderator
Joined 2008
Paid Member
If you have drivers which have not been documented as being used together, you might find information on them separately in a design where each has been used. Combining half of two different crossovers may or may not be OK, depending on a few things.

For starters the typical way to cross is to make the driver output look like a standard filter curve (more or less), which gives hope to this idea. This means that sometimes all you need to know is that the crossover frequency is the same and the order of rolloff (steepness) is the same, and you can attempt to use two different crossover halves. You will likely need to adjust the levels with some resistance, also to check polarity and you may need to tweak a little afterward.

Keep in mind when assessing potential candidates that the steepness is a combination of the effect of the crossover components, and the driver's natural response. This makes it a little tricky to guess by looking, but in a good crossover you can expect the woofer and tweeter to have the same slope, so you have a little more to go on when trying to guess the slopes.

Sometimes they don't follow standard curves.. sometimes intentionally and sometimes for no clear reason. You might be able to tell if the crossover has special resonant circuit parts, although sometimes these are needed to make the response smooth rather than to give it a unique shape.. Therefore unless you plan to analyse the responses you might stick to crossover examples that use more typical circuit configurations.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
Thx Allen,...well actually the speakers sound really well but then again I not a really experienced listener. So to be honest...I don't think I need to change anything besides the fact that the speakers are 30 years old so I was just thinking of replacing them with at brandnew ones. I'm gonna reverse engineer the scheme but I don't have to tools (yet) to measure the inductors nor capacitors in order to determine whether or not they are of the mark in terms of the indicated values.
A strong objective clue its about time to renew crossover bipolar electrolytics is when they are reading uF values well out of the tolerances stated for the type. They usually read too much ESR vs new equivalents too. Even a very cheap transistor & component tester with ESR capability from ebay etc. is enough as a test & comparison tool. A strong subjective clue of tired capacitors is when the speaker appears significantly duller than you remember it.

When renewing old bipolars use "rough" foil types. Such in German have the print Elko Rau for example. Smooth "plain" foil special types have much less loss and will make the speaker appear brighter than its original voicing was. Plastic film replacement caps even more. They would offer more clarity for sure but it would also take more test equipment and expertise in modifying and re-balancing the whole. But the vintage speaker even if bettered its not originally voiced anymore. Losing resale value. At least from a collector's perspective.
 
diyAudio Chief Moderator
Joined 2002
Paid Member
So we learn that in same crossover circuit positions such a pair of speakers had graded parts of almost identical values and no more than +/-5% from nominal. At least there are +/-5% capacitors available today like the Bennic P03A (J) or the Mundorf Ecap Raw. The later I suspect is FTCAP (F&T) OEM.

*The reason passive crossover e-cap bipolars are chunkier than bipolars in active line level circuits is they handle higher signal current.
Because they must pass substantial voltage swings to much lower impedance loads i.e to the drivers.
For the same reason the best MKT or MKP plastic caps for passive crossovers are of film & foil construction than simply metalized film.