recycling old woofers

[MAB]

Now those results are just like I would have expected. The lower-Q version has a droop above port resonance (much like in a closed box), and port output effectively is marginally right-shifted because the woofer's response is more heavily tilted in the frequency range in question. That's quite a small effect though and nothing like what your first comparative graph was showing. I would investigate what happens using a CB in general, going ported just seems like an unnecessary complication.

It's the same simulation result. Only difference is the second one is lower power and I didn't take the time to make nice graphs with the port velocity included. And yes, the frequency response effect is small (in both graphs), same 2dB dip and hump in both results. It's not a horrible result, I am just encouraging measurements since measuring a driver is way easier then building a cabinet. Agree sealed is much easier, especially if parameters are unknown, but glad OP is attempting to collect the parameters. Sorry we got off track here.

How well-controlled is your ambient temperature? That's probably the single largest environmental factor influencing fs. There have been cases of "burn-in" where the effect was shown to be entirely due to the driver having warmed up while playing, with fs basically returning to what it was originally after things had cooled back down.

Yeah, many posts confuse temperature changes of their drivers with break-in. It takes quite some time for a driver to cool down after operation, I've posted on the break-in and temperature phenomena that often get confused, in case the link helps. I expect warm drivers' Fs change of 5% or 10%, but not by 40% as OP is measuring.

I am still puzzled by all the noise on your measurements. Could you show what your impedance measurement cable looks like? It should be as shown in the LIMP user manual. While pickup of acoustic noise is fundamentally possible, speaker drivers are a fairly lossy affair overall and any happenings on the acoustic side should only reflect weakly on the electrical one. Have you tried increasing output volume / generator output level?

I am also puzzled.

 

[haavind]

How well-controlled is your ambient temperature

Yeah, many posts confuse temperature changes of their drivers with break-in.

I've tested the drivers cold and on different days and different times of the day. It doesn't vary that much from the new breakin curve.

, but not by 40% as OP is measuring.

Just to be clear, The graph show before/after for all 4 drivers. Each color correspond to a single driver, and the darker/lighter color indicate before/after respectively. I see I didn't explain that explicitly.
No single driver changed by 40%. They all increased Fs by 4-7%.

There's a significant mismatch (about 30% max) between the drivers though, both before and after. I don't know if it's bad quality control from the start, or different aging.

am still puzzled by all the noise on your measurements

How smooth is the curve supposed to be?
It's not that puzzling. The speaker is a microphone. Pretty sensitive it seems. If I try to block the "pinkish" PC-fan noise, then sounds from the structure gets through anyway. Someone closing a cupboard downstairs gives a 20dB burst in all the "interesting" frequencies. Rain and wind noises. People walking.
In some of the runs, something happened and the measurement got wobbly. In others, nothing happened and it turned out better.
Packing pillows around/near the driver affects the result more than the noise does by enclosing the free air.

And also the 50hz AC-mains starts to rise through the noise floor. Maybe it's the soundcard that's low quality. Maybe my cables works as an antenna. I think there could be a groundloop in my wires but I'm no expert on this and there's to much acoustic noise to test it well.

I'm using a 110Ohm R_sense resistor. I think it's on the high side, and I think I read that a lower value can give more noise resilience.

I've measured many times now and they all oscillate slightly around the final averaged curve, and the theoretical TS curve tracks my measurements dead on. The least-squares-fitting is a very robust noise filter.
If I do the max length and number of sweeps (700 seconds) i get this (computed TS curve in black):

I don't worry about it. I could maybe try and fix it, but it's not worth the hassle. The final numbers are probably correct. I got some blue-tack today to get better added mass measurement than with electrical tape and metal pieces

 

[haavind]

Could you show what your impedance measurement cable looks like?

Good luck decoding this mess

 

[haavind]

Have you tried increasing output volume / generator output level?

If I set it higher, REW complains about input clipping. If I overrule the input clipping, the result comes out wrong. Wrong values and it doesn't track well to the fitted TS curve anymore

 

[Wolf11Man]

If you use too little current, the results will be incorrect. The higher you go, the better. 110 Ohm is to much for a resistor. Try something like 20-40 Ohm. You should hear the sweep quite loud. Noise will be much lower, too. Check you cabeling, again, there may be something wrong.

 

[haavind]

It seems like the blue part of the cable is an especially good 50hz antenna

20-30-40 dB more sensitive than different audiocables i tried.
And it may be causing all sorts of noise, calibration, level, and clipping issues.

Electrical engineering is fun

 

[haavind]

It turns out my microphone input gain setting was waayy to high

Thanks for your critical questions I suppose

 

[haavind]

this is a single 256k sweep now. Before I was doing 4x512k.

 

[haavind]

I bought a pack of lower value resistors with better power handling.
I'm now using 27 ohm R_sense and 7 ohm R_ref.
I soldered my wire harness and stabilized with electrical tape.
I redid calibration, making sure to remain well inside the bounds of what my headphone_out and mic_in likes.
I did test sweeps across a range from 0dB to -40dB, and 1x64k to 4x512k sweeplength. The result was within like ~0.01 ohm regardless.

I tested the effects of warmup. This is it: Black is cold, cycling toward light gray for each run, magenta is final. It goes from 51.2 to 49.6

This is the impedance plot for all four drivers. I don't know if it's any point in comparing to my initial results anymore, to much uncertainty, but they look more like the very first results I got.

The only thing now that indicates that something might be off is this:
When I compute TS based on just a single sweep, things look fine.
When I compute full TS based on either the 6g or 12g added mass, things look fine.
But when I do the dual added mass calculations, the theoretical TS curves no longer tracks my measurements nicely.
It looks like this:

I can make it track a little bit better (but not great) by fiddling with the weight numbers, but then it complains that my Bl curve changes too much.

Anyone have any thoughts on this and is it worth trying to iron out?

 

[haavind]

properly zeroing my multimeter before doing resistance measurements on R_sense, R_ref and DC coil resistance, and measuring R_sense directly at resistor terminals instead of through the testleads fixed it.
By adjusting the added mass values a little (I don't have a very accurate scale, so I can't know the values for sure) I could track the curve pretty close to perfect, with a flat Bl curve.
It had a noticeable, but not largely important effect on the final values.

I am just encouraging measurements since measuring a driver is way easier then building a cabinet.

The second time you do it maybe it's easier