Woody 1x12 - SM57 Steps IRs

[James Freeman]

I rotated only the speaker 90deg clockwise inside the cab, what was Down now is Left.
The sound sweetspots rotated with the speaker and the cab orientation and grille perforations which remained the same barely have an effect.
Anyway, I have acquired new knowledge from this experiment and I will be more aware of the non-symmetrical concentric nature of guitar speakers.

Red: Down 4.5cm from center.
Green: After rotating the speaker 90deg clockwise, now Left 4.5cm from center.

PS. I will not argue or retort on this forum, because.

 

[jay mitchell]

Before you draw more incorrect conclusions, I'll provide some essential information.

1. The wavelength of sound at 200Hz is 1.715 meters. At 100Hz it is 3.43 meters.
2. The cab will emit a coherent pressure wave in the near field at all frequencies for which the cab dimensions are smaller than a wavelength.
3. The response in the above-defined frequency range will not vary at any position within the coherent near field wave.
4. All of your cab dimensions are substantially less than a wavelength at 200Hz and below.

As a result of the above, it is evident that the apparent differences below 200Hz cannot possibly be caused by your having rotated the speaker within the cab.

When you are examining tiny differences like the ones in your graph above, you need to first verify the repeatability of your test.

1. Acquire a series IRs from the same mic position over a period of, say, ten minutes, changing nothing between measurements.
2. Overlay all those IRs together.
3. Choose one IR as the reference.
4. Normalize the others to that one.

Identical IRs will normalize to a straight line on the frequency axis at 0dB. Variations from 0dB will establish the limits on the repeatability of your measurements. It is an underyling principle of testing and measurement that differences that fall within the repeatibility of a test are not meaningful.

 

[KHAAAAAAANNNNNNN!!!!!!!!]

I say from hands-on experience that you'd be much better off acquiring IRs with the grille removed. Obviously that would require a different methodology for mic placement, but you'd then find that there are far fewer unexpected response variations from one IR to the next.

I wondered about that. Makes sense.

 

[James Freeman]

I've reshot my Vintage 30 and Greenback in four directions, Up, Down, Left and Right by rotating the speaker inside the cab and capturing 33 steps four times in the same direction resulting in 132 IRs per speaker.

@MirrorProfiles I know you are a professional studio engineer and also shoot IRs, I would love to hear your opinion on this and which direction and step you would choose as the sweet spot for both speakers.
Much obliged!

Download link (6.5MB):

https://drive.google.com/file/d/15OqvRTMmPSj9_85v4BwMQeFuS5vYOK32/view?usp=sharing

 

[MirrorProfiles]

I've reshot my Vintage 30 and Greenback in four directions, Up, Down, Left and Right by rotating the speaker inside the cab and capturing 33 steps four times in the same direction resulting in 132 IRs per speaker.

@MirrorProfiles I know you are a professional studio engineer and also shoot IRs, I would love to hear your opinion on this and which direction and step you would choose as the sweet spot for both speakers.
Much obliged!

Download link (6.5MB):

https://drive.google.com/file/d/15OqvRTMmPSj9_85v4BwMQeFuS5vYOK32/view?usp=sharing

Checking them out now, just had a full day recording guitars so my ears are a little tired but I'll give them a quick blast. 132 IR's per speaker is pretty nuts!

I'm WAY more versed with 4x12's than anything else so my frame of reference for 1x12's isn't ideal. These definitely sound great and I wouldn't guess they were a 1x12, so good job with construction and choice of dimensions. Just tested with Neural DSP Henson (2204) and their SLO OD channel which should be sort of middle of the road/typical amp sounds.

For my tastes (as in these particular speakers and mic), on first test I like the U orientation on the Greenback, and R for the V30. As expected, the difference are a bit more pronounced as you get away from the center. A lot of the time the differences are impossibly small or so negligible that a preference would be quite context dependent. R on the V30 seems nicely balanced to me as you move the mic. I like between 3 and 4 on the V30, and below 3 on the Greenback (although in real use I'd probably find uses for positions all the way to 8, particularly if blending mics).

IMO I'd combine either the U+D and L+R (so you have a diameter of the speaker captured, and make less measurements. Interpolation works annoying well for this kind of stuff and I bet some of the middle points would be indistinguishable by interpolating some either side of them. I've found with my own IR's that its really case by case whether I prefer the glue blob side of the speaker or the other side. I tend to favour the opposite side to the glue blob but sometimes it just works better.

If you can afford to budget for a dynamount for this, it will make your life way easier too. I think I did one day of making IR's before investing in one.

 

[James Freeman]

Thanks for testing.

My preferred direction for the V30 was D, and Greenback was L, as these produced a more even and gradual frequency rolloffs.
Interesting that you like a brighter sound/positions on both speakers, I find anything below 4.5 on the V30 too bright, same for the Greenback.

Interpolation works annoying well for this kind of stuff and I bet some of the middle points would be indistinguishable by interpolating some either side of them.

I initially captured in 5mm steps but found that the sounds changes significantly between 3 to 5cm where the dustcap is glued to the cone so I halved the steps to 2.5mm, I do not think interpolating in that region will yield accurate results.

 

[MirrorProfiles]

Interesting that you like a brighter sound/positions on both speakers, I find anything below 4.5 on the V30 too bright, same for the Greenback.

I definitely have a preference for brighter and more open sounding tones - I think that’s just generally what’s worked better for me in a studio context. If I’m handed darker tones to work with I generally have to boost quite a lot of top end and I just sort of gravitate to avoiding EQ if I can help it.

I didn’t test extensively with interpolating, but I did try a few different distances apart and compared them to the IR’s that I had captured in the middle. It really was way closer than I would have imagined (I think I generally have around 20 IR’s per speaker diameter, generally capturing at 2 dynamount steps per IR). Some Neural DSP plugins only use 6 horizontal positions in their cabs and IMO you can hear when it’s relying on interpolation too much.

Maybe I should test again…

 

[James Freeman]

Some Neural DSP plugins only use 6 horizontal positions in their cabs and IMO you can hear when it’s relying on interpolation too much.

I believe the new Line 6 cabs also use interpolation otherwise they could not have packed so much data per cab.
The key is to not interpolate/skip the real sweetspot region, 6 steps is not enough for the small sweetspot region let alone the entire diameter.

 

[jay mitchell]

I've downloaded the IRs and have begun examining them. As before, I've only looked at the greenback.

The first set of graphs comes from normalizing the 0.00 files to the "D" one. If the placement of the microphone and the speaker are not altered at all from one test to another - IOW, if the speaker is rotated in the enclosure without altering the position of its center and the mic is placed at exactly the same locations for each orientation of the speaker - then all of these normalized charts will lie within your test repeatibility window at all frequencies. Edit: because these IRs were all ostensibly taken on axis, they should be identical. Cylindrical asymmetry will have no effect on them. I use the same set of tests, in addition to a few others, to verify the calibration of my rotational fixture and mic placement when I'm acquiring directional data on loudspeakers.

Here they are:

Left normalized to down:


Right normalized to down:


Up normalized to down:


Note that there are significant deviations. The fact that they are there casts doubt on the accuracy of speaker placement after rotations and/or mic positioning. Note that 2.5mm is a small increment - .098 inches. If this is enough to lead to audible differences in IRs from adjacent positions, it stands to reason that very small errors in mic positioning can also cause deviations.

Possible causes for the above differences:

1. Basic repeatability of the test technique.
2. Small errors in speaker location when it is rotated. There is almost always some play in the fasteners, so you really need to make reference marks to line up with when you remove and reinstall the speaker.
3. Torque variations when you tighten fasteners. This is potentially quite significant with stamped-basket transducers.
4. Small errors in mic placement relative to the cab - including spacing off the grille.

Nothing you have done so far supports your hypothesis that cylindrical asymmetries in the speaker's cone cause audible differences at different rotational angles. If you really believe that this is the case, there is a way to establish that with a high degree of confidence using your present methodology:

1. Do the repeatibility exercise I outlined earlier in the thread. Establish reliable limits on the repeatability of your tests. For example, if the differences in the graphs above were due to repeatibility alone, your repeatibility below 10kHz is roughly +/- 2.5dB and +/- 4.5dB from 10k-20k.
2. Mark your cab so you can reliably reposition the speaker when you rotate it.
3. Establish a method to place the mic relative to the cab with 10% error (.25mm) or less.
4. Use a torque wrench to tighten the fasteners.
5. Re-acquire the "0.00" IRs at D,L,R, and U rotations.
6. Verify that the differences you see fall within your repeatability limits.
7. Re-acquire IRs at a single mic position, carefully observing all the above. Differences you see that lie outside the repeatability limits are due to some kind of cylindrical asymmetry. That is still not necessarily the cone construction, but that's one possibility.

 

[Orvillain]

This is a really interesting thread, keep it coming!