In ear monitor crossover

[tranhieu]

I'm building my own custom iem atm, everything goes according to plans but I'm stuck with the crossover part as there is not much info available but some frequency response graphs. I tried emailing the manufacturer asking for impedance graphs yet they don't have anything like that in database.

Anyways, my plan is to build 2 designs, one use 2 receivers per ear (1 tweeter + 1 woofer) and one use 3 or 4 receivers per ear (1 dual for mid + bass, 1 for high, or perhaps another 1 for low if I want to go for a quad). Below are the datasheets:

- 2 receivers:
ED29689: http://www.knowles.com/search/prods_...-29689-000.pdf
BK26877 (the datasheet attached is for 26824 but they are not that much different in terms of response) the impeadance of 28677 is 28ohm.
using the online calculator with ed29689 being the tweeter and the bk26877 being the woofer and cutoff at 1.5khz, I bought some 2.9mh inductors and 15uf caps.

Is that setup ok? I'm a bit worried about the lack of impedance graphs which makes it impossible for me to work out the resistance value I should add to dampen the peak (if there's any).

- 3 or 4 receivers:
WBFK30095 (this one is a dual receiver) :http://www.knowles.com/search/prods_...-23910-000.pdf
DTEC30008: http://www.knowles.com/search/prods_...-30008-000.pdf

The wbfk will take care of the high, while the dtec covers the rest of the frequency. However I'm thinking about adding another receiver for the low

2015: http://www.sonion.com/Products/Trans...0/2015_v3.ashx

Due to the lack of measuring equipments I cannot go any further than a 3 way design, even a good 2 way would be more than enough. However I don't know how to design one given these data sheets. Or perhaps I will just keep it as simple as possible without adding an additional driver?

Thanks for reading

[sreten]

Hi,

No idea about what you are doing but the impedance data gives an approximation.

AFAICT drivers for hearing aids and hearing loss are not intended for hifi.

rgds, sreten.

[weltersys]

Those transducers look like they need a boatload of EQ, I'd suggest DSP, passive EQ and crossover would be a complete crapshoot.

There are in ear monitor companies that build two transducer units, any reason in particular you want to DIY such a thing yourself?

[tranhieu]

Originally Posted by sreten

Hi,

No idea about what you are doing but the impedance data gives an approximation.

AFAICT drivers for hearing aids and hearing loss are not intended for hifi.

rgds, sreten.

I got the same impression at first, but susprisingly loads of custom iem companies out there manage to implement them for their hiend iems. I'm not saying they can rival speakers by any means but their sound quality is pretty good to be considered.

Originally Posted by weltersys

Those transducers look like they need a boatload of EQ, I'd suggest DSP, passive EQ and crossover would be a complete crapshoot.

There are in ear monitor companies that build two transducer units, any reason in particular you want to DIY such a thing yourself?

oh... thanks for the suggestion, I will have a look at the EQ.
well, the thing is if I diy one myself the cost is only 1/4 or even 1/5 of a commercial product, and I am free to tweak the sound signature.
I know custom iems are not very well known here but imo they are worth diy-ing considering their quality that is packed in such small housings.

[speaker dave]

Have at it.

I worked with Westone for a exploratory IEM project. They use knowles capsules and are very well regarded by musicians. I have their 1, 2 and 3 way units (the 2 and 3 ways are pretty good). Their networks are very simplistic, as in blocking capacitors. You can do much better.

One issue you will have with network design is the unknown driving impedance, unless you only want a fixed system.

(If anybody has some money to waste and real marketing savy, I'd like to build up some designs and a brand behind this concept.)

David S.

[tranhieu]

Originally Posted by speaker dave

One issue you will have with network design is the unknown driving impedance, unless you only want a fixed system.

By driving impedance are you referring to the output impedance of the source? If I'm not mistaken the output impedance is always close to 0?

Yeah I reshelled my UM3X and to my surprise the network inside is so simple that it cannot get any simpler. The sound is ok, not that good imo but Westone uses the same transducer configuration for the ES3X which is a monitor for recording (as it's advertised), the only difference between the 2 is the xo.

[gedlee]

Originally Posted by tranhieu

I'm building my own custom iem atm, everything goes according to plans but I'm stuck with the crossover part as there is not much info available but some frequency response graphs. I tried emailing the manufacturer asking for impedance graphs yet they don't have anything like that in database.

I have a ton of data having worked at Knowles and recently doing an IEM project. Much of it is proprietary, but if you E-mail me on the side then we can discuss what it is that you want and maybe I can direct you around a few pitfalls.

The impredance gets quite complex and no two models will even be similar so you need specific data, but its not that hard to get. I also have a lot of in-ear measurements of deveices that I know are correct (many you see are not).

I also have computer models of the devices, ear couplings, crossovers and in-ear simulations using real ear impedance loads. Of course that is proprietary, but I might be able to run some examples for you.

[speaker dave]

Originally Posted by tranhieu

By driving impedance are you referring to the output impedance of the source? If I'm not mistaken the output impedance is always close to 0?

Yes, there isn't really any standard for source impedance of headphone circuits. Some have near zero source R, others may be 40 to 100 ohms. This makes for a great deal of variation of driver response and especially for the response of any passive network you might want to add.

David S.

[XuMo2]

Interesting thread, I would like to follow. I have been checking in internet for same.
My thoughts :
I think the crossover used in this IEM are just marketing stuff. As weltersys suggested they dont make real sense.
There are restrictions on use of inductors. So the crossover needs to be done using RC filter. I think most of big names use factory tuned transducers (custom). And some basic crossover is done by caps and register values.
Would be interesting to know what you people think. Gedlee & Dave seem to be experts here from audio industry.

[XuMo2]

tranhieu , Did you make any progress ?
"2.9mh inductors and 15uf caps" . Inductors should not be used with BA, the manufacturers recommend avoiding them. 15uF seems too big value.