BHOOPALI (TM design)
For Bhoopali, the TM design, this is what I decided to go with as the crossover circuit. It's fourth order electrical.
The values in the circuit need to be rounded off to sensible values. The numbers here are more accurate than we need. For instance, the resistance value of 2.040 Ohm can be just 2 Ohm, and 7.582 Ohm can be 7.5 Ohm or thereabouts. Capacitor value of 6.510 uF can easily be 6.3 uF + 0.1 uF + 0.1uF in parallel, and so on.
This is the SPL curve: fairly flat:
There is a very gentle rise in the SPL curve with frequency. It is possible that the speaker will sound slightly on the detailed side. This rise will of course disappear as soon as you position them to listen slightly off-axis, maybe with a bit of toe-in. This gentle rising characteristic is a result of the tweeter's SPL curve -- there is nothing much one can do to fix it in the crossover. If I don't like this rising curve, I'll avoid using this tweeter in future projects. The midbass driver was a peach -- it was super simple to mate it with any crossover I could have wanted, because it had a wide, uniform, and smooth SPL characteristic. Typical of inexpensive poly-cone midbass drivers.
In this SPL curve, ignore anything below about 500Hz, since I took SPL measurements using a gated noise method which does not give me accurate SPL measurements at low frequencies. If I wanted to measure down to the 20s, I'd have to take nearfield measurements of the low frequency region and splice it with the gated high-frequency portion. Not needed for crossover design, since my crossover area is comfortably in the gated-measurement region.
And this is the impedance graph the software is predicting:
Ignore everything in the graph below, say, 100Hz, because the impedance in that region will be determined by the box volume, the impact of the bass reflex port, etc, all of which have been kept pending when the measurements were taken. These measurements were taken just to let me design the crossover, which means anything below about 500Hz is unimportant for me here.
Therefore, if we just look at the range above, say, 100Hz, then the impedance does not drop below about 4.5 to 5 Ohm at any point. It's a fairly easy load for any modern amplifier. The final impedance curve will only be obtained after the crossover is built and the bass reflex port fitted.
BHAIRAV (MTM design)
For Bhairav, the MTM standmount design, this is the crossover circuit I arrived at. A fourth-order electrical topology for the tweeter, third-order electrical for the midbass drivers:
I started with fourth-order electrical for the midbass too, but after optimisation to align with a fourth-order acoustic slope, the software told me that I could drop the second capacitor, which would have been the fourth component of the 4th order filter. So I'm getting 4th order acoustic slopes with a 3rd order electrical midbass filter.
You may be confused seeing only one midbass driver symbol in this circuit. You know that an MTM has two midbass drivers. Well, in my software, I do not work with two drivers because I take measurements of both drivers together, paralleled into one. So, both for SPL and impedance, there is only one measurement. The software would not be able to process this if I had entered two midbass drivers in the circuit -- that would have needed separate SPL and impedance measurements for each driver, something which is very difficult to do with an MTM box. You can't take an SPL measurement of just one driver if two of them are fitted in one chamber -- Driver #2 would become a passive radiator when you're pumping signal into Driver #1, and
vice versa, making the SPL measurements meaningless. Therefore, the only way to take midbass measurements in an MTM setup is to hook them up as one, and then take one reading.
Once again, the values are down to three decimal places here, and will need to be rounded off in reality to get practical values. An error of 2-3% or so this way or that doesn't have any impact at all on the sound.
The crossover gives me this SPL curve, shown again with the total summed curve and the slopes for the individual drivers:
The summed SPL curve is quite pleasantly flat, which is partly just a testament to the drivers. The slight notch at about 4KHz to 4.5KHz both in the tweeter and the midbass driver curves is probably an artefact of reflections: I've always seen some small unevennesses introduced by the MTM driver layout on the front baffle. The emitted sound waves bounce or reflect from adjacent drivers a bit. That dip can be ignored. And dips in the SPL curve are very hard to notice by hearing anway -- it's peaks which colour the sound in more obvious ways.
One thing to note is the four or five squiggles of the midbass drivers from 5KHz to 10KHz. This is due to cone breakup. I hadn't expected this in this driver -- my previous experience with Kevlar cone drivers has been smoother SPL. However, I suspect there won't be significant audible impact of this. I believe metal cones have much more audible cone breakup modes. In fact, it's partly because of the cone breakup modes that I wanted to set a lowish crossover frequency and use a higher-order slope -- a 2nd order slope may risk the cone breakup region becoming audible.
The crossover frequency here is about 2.5KHz like the Bhoopali's case. In MTM, the crossover frequency is supposed to be set as low as feasible, but in this design's case, the tweeter would not be able to handle a very low frequency well. An MTM design therefore places demands on the tweeter to deliver clean sound to quite low frequencies, and this one, being a more budget-constrained design, uses a good tweeter which does not go very low.
The impedance curve of this crossover, as per the modelling software, is as follows:
This impedance curve indicates that we can expect a 3-Ohm or slightly higher load on the amplifier for most parts. This is not easy for a weak amp, but most modern solid-state amps, even Gainclones and chip amps, will be able to deliver the few Watts of power we are expecting. If you remember, we are designing relatively low-budget standmounts for small Bombay bedrooms. This means that we don't expect to play very loud.
This lowish impedance is a problem with standard MTM designs. People like to connect their mid-bass drivers in parallel in MTM configurations for some reason. This means that they fit a pair of 8-Ohm drivers, and these give you a 4-Ohm nominal load after paralleling. This often means that the curve dips into the 3-Ohm region in reality, as is shown here.
When I had built my first Asawari (floorstander MTM), I had chosen 4-Ohm midbass drivers and connected them in series. That may be a better approach if you expect to drive them with an amp which will be uneasy with a 4-Ohm load.
Hindsight
Looking at the measured data of the drivers, the following thoughts arise:
- For Bhoopali, I'd choose a different tweeter, just because that very slight rise in the SPL curve is not "nice". In fact, the tweeter I have used for Bhairav (the Tympany XT25SC90 ring radiator with small flange) may be perfect for a Bhoopali v2.0. It has lovely flat frequency response, and Bhoopali does not require a low crossover frequency. The XT25SC90 tweeter could have done the job.
- For Bhairav, I'll use a different tweeter if I do it again. I'll want a tweeter which can be crossed over lower, so that I can suppress the cone breakup region of the midbass drivers (about 5KHz to 10KHz, as I have explained above) more strongly. I may also try a steeper slope. That will push up the crossover price by adding some more coils and capacitors, but may give a cleaner sounding speaker.
Now to order the crossover parts, build the crossover, fit the bass reflex port, and sit back and listen. That will take some time -- may be a month or two. Parts take time to arrive.
In the meantime, will welcome questions and opinions.
