There's more than one way to DIY

ScottS

One of the projects I have had in the queue for a while is an 8" two way based on the SB Acoustics SB20PFC30-8 and a tweeter in the SEOS8 dome tweeter waveguide. Thanks to JR, I have some tweeters in house that look like reasonable candidates to try with the SEOS8 (the Seas 27TFFC). Last weekend, a friend let me know about a pair of speakers at a thrift store and wondered if he should pick them up and how much to offer. I couldn't resist so I said $20 and they took it. They are some old Mitsubishi's that probably aren't good, but the cabinets are pretty heavy. The model number is M-FS8.

This particular pair had some interesting customization.

Apparently these needed a midrange, so the port was replaced with another driver. Pretty impressive, whoever did that must have redesigned the crossover.

Nope! Now I will have to measure one of these for sure before I repurpose the cabinets.

The bottom few inches is just a riser, so the overall box for modeling purposes is 32" x 10" x 12.5" out of 3/4" particle board for an overall internal volume of around 47 liters. This happens to be a favorable vented alignment for the SB's.

I was originally thinking about doing this project as a smaller sealed design, but maybe not...

Silver1omo

I like that, I'm working in a polk refresh/recycle.

6thplanet

Do those Polks have the "PowerPort" out the bottom?

Silver1omo

@6thplanet said:
Do those Polks have the "PowerPort" out the bottom?

Yes, they were 3 inches, had to reduce to 2 to work better for the tuning. Not sure how to measure the near port thou.

dynamo

Hmm. For fun you should measure and play them (without the added mid..). Providing the crossover isn’t as awful as possible they may not be bad. Those drivers appear to be shielded versions of the drivers in the Phase Tech 7T, which wasn’t awful. I’d be willing to bet they are a mitsu rebrand. Please post pics of the crossover when you tear them apart.

ScottS

I'll do some measurements before I take this project further. The first step will be T/S parameters on the woofer and then I will have to rebuild at least a makeshift port.

ScottS

I'm getting back into this project. The first thing I did was take a measurement at 1 meter on axis with the goofy midrange in place.

There's clearly some interference and excess energy around 4-6 kHz, but it isn't nearly as bad as I imagined. Impedance also wasn't horrible.

The fuzziness around 100 Hz was due to the midrange driver only being held in by 2 screws and being a little loose as a result. That fact also likely contributed to the 50-ish Hz peak being fairly low and doubled.

Next, I took the woofer out to get some data on it. These are the T/S parameters I measured.

Pretty high Qts and VAS is sufficiently high that the bass reflex alignment won't be great. I plugged this into WinISD. I set Qa to 20 (instead of the default 100) because the Mitsubishi cabinet has a fair amount of lining/stuffing, which is understandable. I then used the estimated cabinet volume based on cabinet size and tried sealed (yellow line) and vented (white line) alignments.

The vented alignment uses a 3" diameter 8" long port, which gives about a 34 Hz tuning and F3 around 44 Hz. In this case, I would be tempted to just run the woofer sealed as f3 is 51-52 Hz and there is a broad 1 dB peak centered at 100 Hz for just a little extra warmth. With that, I will probably just cover the midrange hole and take another measurement.

ScottS

I put a plywood scrap over the hole where the port/midrange was and remeasured.

Now we don't have all that crazy interference at higher frequencies. The treble is elevated by a dB or two, but in the as designed speaker that is probably somewhat offset by the bass reflex tuning - which is likely similar to the WinISD white line in the previous post. The only other issue is some reduced response around 200 Hz. In order to get the high frequency measurement as smooth as it is, I had to use a very short gate time of around 3 ms. Praxis uses a flex windowing algorithm that tries to estimate the lower frequency response from data beyond the gate time, but it's definitely not perfect, so I wondered if that dip would still show up with a longer gate time. I reran the measurement with a 7 ms gate. It's a lot messier, but the dip is still there.

The impedance plot looks cleaner with the plywood where the midrange was.

The sealed box peak is around 55 Hz and there is a small blip around 178 Hz.

All-in-all, this appears to have been a pretty well designed speaker.

ScottS

After Dynamo's comments and the encouraging measurement results, I actually buttoned both speakers back up without the midrange in order to give them a listen alongside some of my other rotating pool of speakers. They are actually decent. The vinyl wrapped particle board cabinets are in rough shape. I would have to fill in a few divots and either paint them or redo the vinyl to return them to the wild - but that is actually an option. It would have been easier if they sounded terrible...

ScottS

Here is a pic of the current state...

The rogue midrange was spliced into the wiring going from the crossover to the tweeter, so that explains why the interference in the measurement with the midrange attached was all at higher frequencies. I think they sound quite a bit better than a pair of Infinity Reference 20's (also an 8" two-way, but a standmount) that a neighbor gave me.

ScottS

Did more listening and comparisons yesterday. Despite the fact that the Mitsubishi's are reasonably well balanced, listener fatigue does eventually set in (predictably for an 8" woofer, 1" dome speaker without BBC dip). I could take on an exercise to document and try to fix off axis behavior - but I think I will stay with the original plan instead (SB20PFC woofer, Seas 27TFFC in a SEOS8 waveguide). Cabinet will get a new front and sides will be rewrapped.