Your remark made me doubt my design, as mentioned earlier I was pretty certain the XO was a lot higher than 2kHz. And inverting the polarity of one of the drivers does make it look like to XO point is just below 2kHz. But looking at this a little closer, the XO point really is quite a bit higher, just above 3.5kHz (see lower graph). I don't understand why the dip (when inverted) is at only 2kHz, as the phase seems to be aligned quite nicely around the XO point.
Will do!
Well spotted Stal!
If I may offer some advice for crossover layouts; an ground is not equipotential when current is flowing in it. Make sure that currents returning from the drivers (-) terminals to ground do not share the same paths on the PCB or it will introduce crosstalk between the low and high frequency outputs of the crossover.
Also for a pre-made 'breadboard' for a crossovers have a look at these three from Parts Express: https://www.parts-express.com/Crossover-PC-Board-2-Way-12-dB-260-130 , https://www.parts-express.com/Crossover-PC-Board-3-Way-12-dB-Small-4-x-7-260-132 , https://www.parts-express.com/Crossover-PC-Board-3-Way-12-dB-Large-5-x-9-260-134
.
By cutting a few sections of the solder pads, I managed to use their large three-way circuit board to implement a crossover with Zobels and LCR EQs for the drivers, and still keep the ground return current paths seperate. Below is a first order 3-way crossover with compensation for the drivers Z to achieve the desired crossover slopes, made on the Parts Express 3-way 2nd order PCB. There are about 4 cuts to the solder pads not easily visible, and the odd bridge added between pads.
If I may offer some advice for crossover layouts; an ground is not equipotential when current is flowing in it. Make sure that currents returning from the drivers (-) terminals to ground do not share the same paths on the PCB or it will introduce crosstalk between the low and high frequency outputs of the crossover.
Also for a pre-made 'breadboard' for a crossovers have a look at these three from Parts Express: https://www.parts-express.com/Crossover-PC-Board-2-Way-12-dB-260-130 , https://www.parts-express.com/Crossover-PC-Board-3-Way-12-dB-Small-4-x-7-260-132 , https://www.parts-express.com/Crossover-PC-Board-3-Way-12-dB-Large-5-x-9-260-134
.
By cutting a few sections of the solder pads, I managed to use their large three-way circuit board to implement a crossover with Zobels and LCR EQs for the drivers, and still keep the ground return current paths seperate. Below is a first order 3-way crossover with compensation for the drivers Z to achieve the desired crossover slopes, made on the Parts Express 3-way 2nd order PCB. There are about 4 cuts to the solder pads not easily visible, and the odd bridge added between pads.
Last edited:
You mean like this? I always assumed it made no difference, as all grounds will be connected at some point. But I've been wrong before.
Thanks for the advice, but I'll probably end up getting boards from Sound Imports again. Their warehouse is about a 5 minute walk from the place I work, whereas Parts Express will require international shipping (US to the Netherlands).
You can judge the crossover point by the FR of the drivers. If inverting the polarity of one driver creates a dip this is a proof of the crossover point. The 2nd graph shows the filter response, but this won't tell you where the crossover point is.
About your crossover, I think there are 2 problems. First I suspect that you simply used as-is the Dayton Audio provided frd and zma files. Unfortunately you can do that only with an infinite baffle speaker (or something that is close to it), as in a normal speaker you are omitting baffle step and diffraction effects. Second problem, you didn't provide a path difference between the drivers.
Ralf
Correct. Current can not flow without a voltage drop in accordance with Ohm's Law, namely V=IR. The R of the ground track of your crossover layout could be as high as a few tens of milli-ohms between the (-) input and (-) output terminals. In your layout most of the voltage drop caused by the return current of the bass driver is added back in to the output to the tweeter output after the high pass section of your crossover. This 'crosstalk' will limit the attenuation in the stopband of the highpass filter section in a frequency selective way that is a function of the bass driver's characteristics. But fortunately it's an optional problem! Separating the return current paths to the crossover input ground eliminates the issue without any extra cost.
I can't see your location so it is difficult to give region specific advice. The cost of overseas shipping for Parts Express is totally outrageous and there is nothing inherently wrong with the Sound Imports proto-boards, but nevertheless another DIYer in the US may find it helpful information.
Edit: Sound Imports also sell the crossover boards that I linked from Parts Express.
Last edited:
And yet they still connect at the same binding post to effectively be the same connection. In a second order arrangement, the out of bandwidth frequencies will take the path of least resistance. This means they don't travel the way you suggest under AC voltage conditions.
Yes they do, because the path of least resistance is shared by the high and low filter outputs! BTW the common return path of the speaker cable does not cause the same problem because it is before the filters.
As well as the crosstalk between the different outputs of a passive crossover in the above example, common grounds sharing currents from different functions of electrical/electronic circuits give rise to all sorts of problems in audio equipment, from inducing power supply ripple noise into audio circuits, crosstalk between channels, crosstalk between used and unused amplifier inputs and susceptibility to ground loops between interconnected pieces of equipment. Ground noise injection is simply a case of avoidable sloppy design.
There is no before and after the filters when current directionally goes both directions. The secondary return AC path of the woofer will not pass the shunt components of the tweeter circuit without going through them because the tweeter's shunt coil will be less DCR than the tweeter itself.
Diaural xovers are not required, and take twice the components for no important reason.
Diaural xovers are not required, and take twice the components for no important reason.
Of course the high pass filter for the tweeter attenuates the noise voltage induced by the common ground and hence the noise current through the tweeter. However separating the grounds of the two filter sections in the crossover from the input ground terminal onwards removes the issue entirely at no-cost or need for extra components. And it's good practice in any case. Just because I only need to do up three wheel nuts to hold on the wheels of my car does not mean I should not bother with the fourth nut!
I think there is a fundamental misunderstanding expressed in the statements above. Current always flows in all possible paths in proportion to the resistance of each possible path. It does not flow only in the path of least resistance - that is a grossly inaccurate simplification.
Yes, I oversimplified that with regards to your second statement.
However, I still don't see the separate grounds as being any better if they are tied to the same point, or post, and have nothing in the connection string otherwise from one end of the ground to the other at the other driver. I see that you are saying the woofer return does not pass through the tweeter return legs as the midpoint is the post if they are separated, but electrically, it's no different than connecting grounds right from the drivers and making it a ring or only connecting there instead.
However, I still don't see the separate grounds as being any better if they are tied to the same point, or post, and have nothing in the connection string otherwise from one end of the ground to the other at the other driver. I see that you are saying the woofer return does not pass through the tweeter return legs as the midpoint is the post if they are separated, but electrically, it's no different than connecting grounds right from the drivers and making it a ring or only connecting there instead.
The tweeter has a huge SPL hump between 800Hz and 4hKz (see below). My electrical crossover is around 3.6k, but the actual crossover point is way lower due to the response hump counteracting the crossover attenuation. I probably should've used a higher order on the highpass.
I didn't. Both the FR and ZMA files are from my own measurements.
I wouldn't expect to see these things in a crossover.
Not to imply that it is always a problem and hence always needs attending to. Besides, one could add lead resistance to Xsim and design around it.
That's a hilarious response, and BTW designing out the problem has no cost implication - doh!
