Got a pair of speakers and I found this interesting crossover design.

Any idea what type crossover it is? I'm thinking it is simply another way to do a 6 db/octave crossover, but when I rewired it to a proper 6 db/octave crossover the sound changed some.

Well, the topology makes sense...

Low-frequency:
the inductor bypasses the tweeter so the amp is effectively connected only to the woofer

Hi-frequency:
the capacitor bypasses the woofer--amp connected to tweeter

In both cases, it's 1:1---i.e. an 8-ohm amp sees and 8-ohm load with no loss.

Transition:
I assume the crossover midpoint is where Xc = Xl (Meaning that Xc || Rwoofer = Xl || Rtweeter)*. At this point the power is split between the woofer and tweeter.


* I'm using || to designate elements in parallel

I agree that it makes sence, but I don't get the difference in sound when I connected it as a proper 6 db/octave crossover.

could you show how you are connecting it?----In a different topology, I'm guessing the part values would be different.

Like this

Hmmmm---not intuitively obvious. I think I'd need to do the math, but my brain hurts just thinking about it......

When it's wired as found, does it work properly?

I ran circuit simulations both ways (I think your inductor may be about 256 uH by the way). The circuits provide an equivalent crossover performance if the speakers act purely resistive. I think the reason they don't sound the same is that the speakers are not purely resistive which may be affecting the phase relationships of what you are hearing near the crossover frequency (approximately 5 kHz).

If I get time Ill try to find an electrical equivalent for a real speaker and see what happens.

Edited to add: I found more than I want to know about electrical models of speakers. I won't be attempting this any time soon. http://www.tymphany.com/files/resources ... edance.pdf

Curtis Eickerman

The inductor would be a different value as the capacitor was originally 12 uF, but I changed it to 4 uF because the midrange was way too strong. 4 uF tamed the midrange very nicely, but the upper highs are lacking somewhat.

They call that circuit a series crossover as
opposed to the more common parallel crossover.
The speakers are wired up correctly in the
original schematic.Looking at some crossover schematics
the cap value is larger and the inductor value
is smaller compared to the parallel crossover where the
cap is smaller and the inductor is larger.Just do a
google search for "series speaker crossover".
Regards,
Swanson

Oh ok I see.

So given I am operating it parallel now I should replace the inductor with one that provides the crossover frequency I want once I get a proper tweeter installed or can I leave the current inductor installed since the woofer sounds decent enough as is?

To do it right you will have to replace
the inductor with one of the proper
value since you are converting over to a
parallel crossover.The inductor is now
going to be the LPF for the woofer and
you will have to calculate the proper value.
Regards,
Swanson

The series circuit can be wicked as it forms a series tuned circuit which can short
the amplifier terminals through the wire resistance of the inductor if the tweeter burns out
.
Sincen you changed the circuit, and it sounded different, try reversing the polarity
of first the woofer, then listen. Then reconnect the woofer the way it was, and reverse the
tweeter. and see if your on-axis sound is more to your liking. Just do this on one speaker first.

The conventional 6dB/oct circuit you changed it to, twists the phase by 90
degrees, at the crossover -3 dB (half-power) point.

When we looked crossover hookups with Bruel & Kjaer gated SPL gear to see chart curves, the conclusion the benefits never agreed with the listener tests.

So we made the cabinet sound what people bought.

OK. Assuming they were trying to achieve flat electrical performance, the inductor might be 768 uH with an original crossover frequency of about 1660 Hz. With the 4 uF you may have inserted about a 3 dB dip centered at about 3 KHz (that would agree with what you said about turning down the mid-range frequencies). Whatever it is, good luck with the tweaking.

The problem with "flat" electrical performance is that it does not account for the relative radiation efficiencies of the speakers. This can result in the problem you encountered where it just didn't sound right (too much mid-range). So tweaking values can adjust for this to some degree by essentially building an equalizer into the crossover network.

Curtis Eickerman

With constant driver impedance the first order parallel and series crossovers are identical. If impedance changes, however, level changes in the parallel crossover while in a series crossover the level remains constant with a shift in the crossover frequency. Many consider that more benign, and is the source for some of the claims to superiority, but the picture isn't quite as rosy when one looks at back EMF.

There's a good comparison of both first and second order versions here: http://sound.westhost.com/parallel-series.htm

The transient response graph is rather stunning and the author did get 'slightly better' overall response with the series crossover but he argues it isn’t much and would be erased with driver zobels. Btw, his caveat about extravagant claims applies to the zobel as well. It certainly provides a constant impedance to the crossover but that isn't necessarily a 'constant driver response' since power into the zobel produces no sound.

I'm not sure what he's trying to get at with the alleged bi-amp limitation since there is no speaker crossover in a bi-amp configuration.

The issue of back EMF on a cone driver surely is an issue of how damping impacts
the movement, but mainly it means that inductance measurements of the voice coil alone,
and stationary are almost useless for practical crossover design because the inductance is
motional. The starting point used to be the 'impedance' defined as the resistance of the
series resistor when the driver was suspended away from everything and coil
equaled the resistor voltage. The frequency chosen was set put in the standard method
used.