Solving the HD650 mystery

I was initially planning to post this on our company blog, but it took too much time to get comments from Sennheiser, so I shelved the article.

Some time in January my lab guys started complaining about getting these strange Sennheiser HD650's. This headphone is the most popular among our individual calibration clients, so our lab blows through them rather intensely. The stuff below is me discussing what's found in the measurements.


The Data

First of all, let’s take a look at the data and see what we can see. The below graphs are averaged measurements taken from randomly picked HD650 headphones from Q2 2018 and “strange” looking HD650’s from Q4 of the same year when they started popping up.



It should be rather evident that the changes aren’t night and day. The new style HD650’s aren’t a different headphone. The biggest difference is in the upper mids and treble. Overall the 2 - 10 kHz region displays a sensitivity reduction by -3dB and a pronounced dip has appeared at 7 kHz. Our engineering group unanimously agreed that the difference is indeed audible - the “strange” HD650’s sound darker and the treble dip can be troublesome for precise work on say cymbals.

It also seems that the consistency for which we’ve constantly praised Sennheiser has gone down. Here are 5 measurements that were averaged in the above graph from randomly picked samples.



The consistency is spectacular right up to treble. Now let’s look at what the consistency is between the “strange” specimens.



Overall it’s similar with midrange getting more spread. Here's the kicker though - we're seeing that both variants of the HD650 can be bought right now, hence the real scatter in frequency response is rather large. We actually waited a few months, before saying anything to see if the change becomes permanent or goes away completely, too bad that it's here to stay mixed in with the old tuning variants.

So, what now?

Sample to sample variance is a reality for all headphones. Here at Sonarworks we've measured close to 6000 headphones and with many of them being of the same model we've have a unique insight into this fact of life. There's variance between different samples of the same headphone model and there is usually a difference between the tonal response of each channel, because drivers aren't ideal and differ a little from one another.

Traditionally companies have got around the natural variance of products by implementing better manufacturing and quality assurance. Of course it costs money, because by buying a pair of headphones you're also paying for the drivers that didn't make the cut. Our customers who buy individually calibrated headphones should fret much, because calibration gets rid of any tonal aberrations.

If you're just looking to get a pair of HD650's, then it's a bit harder. We took apart one of the “strange” HD650's and there's no way of visually discerning between both variants. Hence it's no use begging the vendors to guarantee that you're getting the one you want. The only way to be sure would be to measure.

Now the great reveal...

Initially I was convinced that this has something to do with Axel Grell leaving the company - you know, Sennheiser falling from grace and all that. Only today I finally solved it! As it turns out, Sennheiser has moved HD650 manufacturing from Ireland to Romania. It's likely that the discrepancies we were seeing were them adjusting to the new manufacturing process. Funnily enough, the HD600's that come to us are still made in Ireland, which may have to do with the fact that they don't sell as fast.

So yeah, the new Romanian HD650's measure almost identical to the old Irish ones, it's all good.

 

I took apart known "old" and "new" samples. Couldn't find any difference in terms of drivers. Do you know for a fact that Ireland still produces the drivers?

 

Really interesting article @Hrodulf but i do wonder @Hands, how do you know the parts still come from Ireland?

 

This is really fascinating.

I'll have to do a little homework since I'm thinking about rebuying the 650 to complement my 600. Might have to spring for an older used pair.

 

Or wait until early(?) manufacturing issues in the new plant are ironed out - @Hrodulf , please continue to monitor this on new 650s that come in and report how things change (or not) with time.

Did I see too that the (Mass)Drop 6XX is also now made in Romania?

 

I was around during the black/white screen HD650 debacle. I’d really like to avoid creating rumours that there are some better or inferior version HD650ies floating around. As far as I and my engineers are concerned, Sennheiser has nailed the manufacturing in Romania and the current stock HD650ies are identical to the regular ones before the factory swap.

And yes, at least the current batch HD6XXes are marked “made in Romania”. My thinking is that the production move allowed Sennheiser to crank out the cheaper Massdrop models.

 

So... this is just a guess, but it looks to me like the diaphragm (includes voice coil) + magnet assembly in the driver can account for this difference.

I have tried to take apart an HD650 magnet, and when reassembling, if you ain't careful enough, the worst case is... you snap the voice coil and have to find a replacement driver (the voice coil is practically impossible to re-attach/solder/connect), while the best case is... you don't align the magnet assembly right so it'll create that slight dip in the treble and bring some sensitivity down. I found that the margin was extremely small, and needless to say, that driver didn't survive for very long because I stupidly pushed the magnet in too far on the 1st try. I won't ever try to do that again.

TL;DR: I think magnet and voice coil alignment needs to be near perfect during assembly. Maybe the folks at Romania are not used to doing something like that.

 

They came back to normal mid spring. I will check once more with the lab guys, but my impression is that we’re getting units that consistently measure like those before the strangeness.

 

Is the spread still wider than with Ireland? The change in factory was mentioned in the Axel Grell thread I believe as a cost-cutting method, which somehow shows through the Massdrop discount but not the normal price which remains about the same.

 

Massdrop stuff only finds its way to us through individual calibration orders, so not that many cans from them. Don’t think we’ve measured any HD58X units so far.

And I’ll request some overview data to check what’s going on right now.

 

Don't HD58X use... different drivers? I disassembled one and it doesn't look anything like any of the HD580/600/650 drivers.

 

I'm just wondering if that also means the drivers are also manufactured in Romania, since there's no precedence of the HD58X in Ireland, unlike the HD580/600/650 line.

 

Are the original images not showing up for anyone else?

I found this thread to be pretty interesting, here are some more measurements. Both were measured at 90db and 1kz. Raw

SBAF compensation

I bought the new pair from Amazon in May of 2019 so, they probably had both variants in the warehouse at that time. Both have the made in Ireland branding on the inside and both have worn pads.

 

Hi there!
I can´t see these images either. I've seen them before, but not anymore.

So, I would like to report another case related to HD650. I have bought a HD650 in march 2017 from Thomman (German) and at the end of that year I bought a pair of HD650 drivers from Audio Sancturay (AS) and because of this thread I has try to measure those. First: the drivers bought from Thomman seems to be made in february and those from AS were from march. My gear for this measurement was a TASCAM US 366 audio interface and a panasonic WM-61A electret microphone (a circuit to feed it with 366´s phantom power) positioned in a flat surface. The heaphones were connected to a DIY amplifier based on The Wire headphone amplifier.

https://drive.google.com/file/d/1N8OR5yHro2PvLva5zj1KCyx_0l_qOcdp/view?usp=sharing
Yellowish = Thomman
Purple = AS