New for this installment is the breaking in of a few new hand tools, a decent solder station and a nifty component tester from Parts Express: The DATS V2 speaker test dongle and app.
I'd like to talk first a little about crossovers. One "dirty little secret" in loudspeaker design is a common practice of inverting the phase of a tweeter to provide decent summation in the crossover frequency of a typical two way design. Sometimes the reactances of the speakers have signal content that works out substantially out of phase at the matched polarity. Usually it's easier to move the red dot than to explain to the user why his tweeter is out of phase. Good summation in the crossover frequency arguably fixes an output problem, but it's not without its own issues. What may work on axis, may not work so well off axis, and at all frequencies. Many very good loudspeakers use this technique.
I'm not a polarity bigot as a rule, but I have never been comfortable with the idea that phase is not always known and dealt with a priori; that is, figure out the right thing to do and do that. As opposed to flipping things around until the response is less broken. Call me a purist, or at least superstitious. I've respected phase since my early encounters with schizophrenic tweeter polarity..
Attempts to bring order to this evil include some admirable solutions. A constant voltage crossover has the virtue of forcing a unity summation at the crossover frequency, and phase is forced to play nice too.
One principle of this type of the crossover is that it is possible through adjusting values to effect a high amplitude slope near the crossover point. This topology is sometimes called a "Quasi 2nd order" for that reason. It's been described by Small (1), Ashley& Kaminsky (2) , and others.
Another concept for dealing with phase at the crossover points is the B&O Phase Link topology. I came across this first in a 1977 paper by Erik Bækgaard (3). A recent re-cap by Geoff Martin is posted in his interesting blog (4) Not to take anything away from the 1970's, but it's clear B&O continues to do some very interesting loudspeaker work today.
Phase link describes a method for inserting an additional driver
The M70 Crossover has elements of both a series crossover and a phase link topology. The next figure shows both types of elements, after factoring out the zobel and pad resistor elements.
The Tweeter and midrange crossovers are a little more complex, but note that there is a constant voltage section set up by C8 and L4
So, this should be an interesting piece to study further.
To help in breaking down the measurements for further analysis, I relaid the crossover to more closely emulate the schematic. (I would flip it top to bottom to line up the same element positions, but we'll let that be my nod to the OCD gods. Not superstitious about that at all).
Astute newsletter editors will notice that I took steps to eliminate the mutual inductance problem cited in part II:
The original layout had a characteristic of about 0.1% of mutual inductance. Certainly smaller than the working tolerances of the project, but no sentiment for carrying over the issue.
I have been taking a number of measurements using my DATS impedance tester. It's a remarkable little device. A USB speaker parameter tester and LRC wrapped in one app. I'll have more to report on the usefulness of that device next time.
It has given me a few new problems to worry about. Dammit. I believe in order for the narrow range of the filler driver to do what it must, it is desirable, possibly crucial, to have drivers that are matched pretty well in the filler range. Unfortunately, since one of my filler drivers was defunct, I had to get a replacement driver from another production lot.
Here is the difference in resonance between the two drivers:
The transfer impedance of the crossover network is as follows:
Clearly it is time to stop cutting bait, and put this thing together. Other projects are wanting their share of the lair.
More on the deconstruction and reconstruction of the Bang & Olufsen M70 speaker next time.
(1) Richard H. Small, Constant Voltage Crossover Network Design, JAES Volume 19 Issue 1 pp. 12-19; January 1971
(2) Robert J. Ashley, and Allan Kaminsky, Active and Passive Filters as Loudspeaker Crossover Network, JAES Volume 19 Issue 6 pp. 494-502; June 1971
(3) Erik Bækgaard, A Novel Approach to Linear Phase Loudspeakers Using Passive Crossover Networks, JAES Volume 25 Issue 5 pp. 284-294; May 1977
(4) Geoffrey Martin, Blog Post: B&O Tech: Uni-Phase Loudspeakers
http://www.tonmeister.ca/wordpress/2015/10/29/bo-tech-uni-phase-loudspeakers/
