Philips SHE3580 IEM review--how can something sound so good for $10??? - Page 33

As far as I know, GoldenEars.net is the only site with measurements for the SHE3580. Those graphs were made before they started to show the raw, measured frequency response. The compensation they use is based on a diffuse field target plus some small-room X curve. If they had used only DF, the treble on the compensated graph would appear lower. GE also uses 1/3-octave averaging, which makes the graphs smoother. If we could see the raw (unaveraged) data, we would see a more uneven curve, probably with taller and narrower peaks.

If a loudspeaker with any claim to sound quality had this kind of frequency response, I wonder who would recommend it. Apparently, there are different standards for headphones. But that's another story. Here's how GE proposes to fix the response in Accudio. They believe in the missing 6 dB effect, or something similar to it.
 

The resonances that cause the two tallest peaks are affected by insertion depth. In the first graph, we can see they occur at different frequencies for the left and right channels. The heights of the peaks might be different too. Possibly, these resonances are also affected by manufacturing tolerances and variations between samples. Therefore, there may be good reasons for seeking alternatives to precision digital filtering besides the lack of (or the complexity of setting up) system-wide EQ.

I've found that the two peaks (near 5.5 kHz and 8 kHz) can be reduced with absorbent foam ear tips, like Comply T/TX series. The smaller peak near 3 kHz seems largely unaffected. The problem when you do this is that the excessive bass become even more noticeable. I suppose we could increase the air flow through the port by making a small hole through the port cover.

For less destructive bass reduction, I punctured tiny holes through silicone sleeves with a wire-size drill bit. But this makes the peaky treble really stick out, and it sounds harsh. So I tried very shallow insertion with a TX400 tip for an imperfect seal. That somewhat works, but it doesn't sit in place very well. Maybe a smaller size tip will work better.

What is really needed is a better way to get controlled leakage from an absorbent foam tip. Enter the Comply S400:
 

Description here: http://www.head-fi.org/t/643197/which-model-of-comply-tips-do-you-like-best#post_9015905
 

I found them at Target, and they're cheaper than the T/TX series. Comply says their disadvanage is that they are porous and that they don't isolate as well. But that's exactly what's needed here. To get the desired bass response, it still needs shallow insertion. But because it's porous, it can be inserted a little deeper and still have the same amount of leakage. It now feels more secure; this time, shaking my head doesn't dislodge the SHE3580.

If you look closely at the photos, you'll see that the T/TX series has a a shiny glaze over the foam, which is probably there to improve the isolation and not be as absorbent. The S series has the foam exposed, and this might make them better than the T/TX series at flattening those peaks. I haven't made a careful comparison to be able to say for sure.

Another advantage to having the bare foam exposed is that it's easier to trim the S400 with sandpaper or a pair of scissors. Just don't trim too much so that it can still absorb sound. I carved mine to make the ends oval shaped. With this customization, I don't have to pinch the S400 tips before insertion; I just stick and twist. It's really convenient.

So far, so good. From what I can hear from music and test tones, the two tall peaks in the response are almost gone, and the bass has been reduced. But what about the smaller peak near 3 kHz? A diffuse-field target curve has a peak there, but it seems to be overdone in this IEM. Moreover, the researchers who don't agree with DF equalization don't think the response at 3-4 kHz should even be as high as it is in the DF target curve. In fact, the DF-compensated graphs of many earphones and headphones show dips in the response there. Any way you look at it, the 3580's response near 3 kHz is a little too high. Humans are very sensitive to sounds at 3 kHz, so if the error is large enough, we should try to fix it. The first thing I notice if I don't reduce the level here and EQ everything else flat, are the vocals; they sometimes have a harsh glare to them.

So I played around with the right channel of the SHE3580 that I damaged. When I removed the metal mesh covering the nozzle, I could still hear the peaks near 3 kHz, 5.5 kHz, and 8 kHz. I inserted some foam into the nozzle for an acoustic damper in the tube. This seems to reduce the level at 3 kHz, relative to 1 kHz.
 

But what if we put the damper outside? I stuffed some foam into the barrel of the S400 ear tip, in front of the SHE3580's mesh grill. It took a bit of foam, but it seems to reduce the relative response at 3 kHz. The down side is that it makes the IEM a little less sensitive. I was also afraid that the damper would hurt the response above 10 kHz, which was already low to begin with. So I was somewhat surprised to find that the response there wasn't too adversely affected. I still could hear the high-frequency test tones through the foam.

When I packed enough foam into the tube, I also noticed that I got less bass. Maybe it's because air likes to take the path of least resistance when under pressure; more of it now goes into the side (and into the S400 foam) instead of through the damper and into the ear canal, and this effect is more pronounced at low frequencies. This lets me insert the S400 deeper into the ear and still get the same amount of bass.

Overall, the results are far from perfect, but I very much prefer the resulting sound over the original. It's not for everyone, but maybe you have similar preferences as me. To my ears, it now sounds a lot closer to what Accudio does to the original sound.

I can't really call this a mod since I didn't drill any holes or alter the SHE3580 itself. I just came up with a custom ear tip--one that happens to work well for me. As usual, YMMV.
 

I'll have to try this with other, similar Philips IEMs: SHE7000, SHE8100, SHE5105 Citiscape Underground. I have a feeling those will be fine, unlike JVC IEMs. Maybe some comparisons are in order?