[Xymordos]

Yup, the bottom graph suddenly did that when I was testing it - the first test was okay, when I pressed record again, the graph went like that. Even though the ED is still producing sound the graph looks like this...I guess it is safe to assume the driver died.

 

[Negataros]

apart from the knowles and sonion drivers, would you guys recommend any dynamic drivers which are pretty decent?

 

[Xymordos]

So I changed the crossover a bit to make it simpler but it still looks like this: 
 
CI22955 - Full Range
ED23147 - Full Range
ED26989 - 1st order 2.2uf
ED26989 - 1st order 1uf
WBFK30019 - 1st order 1uf

Spoiler: Warning: Spoiler!

 
 
I have a feeling teflon tubing isn't a good idea.

 

[vector1]

  So I changed the crossover a bit to make it simpler but it still looks like this: 
 
CI22955 - Full Range
ED23147 - Full Range
ED26989 - 1st order 2.2uf
ED26989 - 1st order 1uf
WBFK30019 - 1st order 1uf

Spoiler: Warning: Spoiler!

 
 
I have a feeling teflon tubing isn't a good idea.

 
You're testing all 5 at once?
 
Edit: I realise that you are really testing 5 drivers at once after reading the back posts... I think you're going to have many problems with driver-driver interaction within the circuit. Remember each driver is a capacitor, resistor and inductor rolled into one and that'll impact the performance of any passive circuit you use. Some expensive CIEMs use active op-amps to buffer the signal between filtering stages to prevent any interactions and give something approximating the calculated response. It's also a common tactic to use multiple amps (bi-amp or tri-amp) to drive hi-fi/theatre theatre setups.
 
I have this feeling with so many mid strong drivers (EDs) and only one weak supertweeter (WBFK), you'll see this kind of high frequency rolled off response since the FR of WBFK is much lower than CI and EDs. 
 
I personally have quite a few problems just trying to balance the 4 driver setup of CI, DTEC and WBFK I'm using now, but I've come up with some decent sounding earphones after some A-B testing even without actually testing the frequency response.
 
How about you post your entire circuit if you can sketch it out, so we can actually see, and maybe I can run a simulation for you.

 

[vector1]

  apart from the knowles and sonion drivers, would you guys recommend any dynamic drivers which are pretty decent?

If I find any other major manufacturers, whose products are widely used and available in small batches for experimentation, I'll let you know .
 
Knowles BAs seems to be the most used simply because they're relatively easy to get. Sonion come in larger trays, which are prohibitively expensive if you only want to make a pair of earphones.

 

[tranhieu]

  So I changed the crossover a bit to make it simpler but it still looks like this: 
 
CI22955 - Full Range
ED23147 - Full Range
ED26989 - 1st order 2.2uf
ED26989 - 1st order 1uf
WBFK30019 - 1st order 1uf

Spoiler: Warning: Spoiler!

 
 
I have a feeling teflon tubing isn't a good idea.

I'm actually quite surprised you get pretty good treble extension there lol looks like the tweeters are doing well.

 

[Xymordos]

Forgot to mention that the bass part is missing because I didn't seal the mouth yet! It's just that dip at around 6k. Individually tested everything seems fine, but together it makes a dip at 6k. However, each driver is correctly wired (I believe). I understand that sealing the mouth will change the signature, but I know it won't cause that massive dip to disappear...
 
The crossover circuit is very simple: 
ED26989 - 1st order 2.2uf
ED26989 - 1st order 1uf
WBFK30019 - 1st order 1uf
 
3 Caps. That's it. Taming mids can be done by decreasing capacitance on the ED drivers? 
 
Edit: Would having each tweeter in 2nd order be better for phase?

 

[tomtomtom]

  What kind of tantalum are you using? I doubt a few volts can spoil the cap. 
 
googling around and I came across this: jimmyauw.com/2010/04/24/observing-inner-and-outer-foil-of-some-popular-capacitors/

It depends on the rated voltage of the cap, whether the cap gets damaged or not.Tantalum caps are polarized so they need to be operated with correct polarization. Using wrong polarization probably results in little to none effective capacitance.

 

[Xymordos]

So that might be why my drivers seems busted eh

 

[tranhieu]

^ I've seen most drivers getting busted for 2 reasons: you sit on them, or you short them. 
 
Quote:

  It depends on the rated voltage of the cap, whether the cap gets damaged or not.Tantalum caps are polarized so they need to be operated with correct polarization. Using wrong polarization probably results in little to none effective capacitance.

Sorry I don't get it. Why you have to mind cap's polarity when audio signal is AC to begin with (polarized caps are not to be used in AC circuits AFAIK)?
 
Yeah voltage rating is something to care about. I always grab ones with >10v rating to be sure.

 

[vector1]

Are you working from a breadboard or have you soldered the crossovers together? I'm working on a breadboard now to tune my circuits while running the individual wires into a earpiece case. If you do it like me, you should have an easier time trying to modify polarities and such.
 
I don't really understand why you have a problem with the bass? If you've sealed the tubing to the nozzle of the driver there shouldn't be any leakage, like my boot below? What is the mouth you're referring to?
 
The sound on my earphones are pretty decent now, so with minor tuning I'll be hopefully ready soon to make the final design.
 
The shells in testing. I've switched the layout to a WBFK, CI, DTEC layout from top to bottom.
 
The wires are there to hold the shells together, and for stress relief on the cables on the drivers.

 

 
Complicated circuit based on what I posted earlier

 
 
The boot of the earpiece that seals the drivers to the nozzle.

 
The shell in solidworks.

 

[Xymordos]

Heh, I clamped my driver too hard when I was trying to solder one and there's a dent on the shell >< I realized taking too long to solder them doesn't actually destroy them that easily!
 
I leave the caps hanging by wire and cover them in UV resin to prevent shorts. The shells I use are the...universal 'custom' shells. Like the Tralucent 1+2 ones, so to seal them means that I won't be able to take the drivers out to re-solder or anything like that. 
 
Also, I realized the 6k and 10k dips can be fixed by adjusting my tweeters positions in the shell. My shell is massive so there is a lot of space to move them around.

 

[vector1]

  Heh, I clamped my driver too hard when I was trying to solder one and there's a dent on the shell >< I realized taking too long to solder them doesn't actually destroy them that easily!
 
I leave the caps hanging by wire and cover them in UV resin to prevent shorts. The shells I use are the...universal 'custom' shells. Like the Tralucent 1+2 ones, so to seal them means that I won't be able to take the drivers out to re-solder or anything like that. 
 
Also, I realized the 6k and 10k dips can be fixed by adjusting my tweeters positions in the shell. My shell is massive so there is a lot of space to move them around.

 It's not that the case needs to be sealed. Only the tube needs to be sealed to the nozzle of the driver and nozzle leading to the tip to ensure there's no leakage... If the sound is leaking into the body of the shell then regardless of how you seal it, there's still going to be a loss in the bass?
 
My shells are not sealed at all, but the drivers are quite deep within the boot, which seals the channels carrying sound.
 
Be careful working with the wbfks, as excessive heat can cause the solder pads to lift of and break the wires underneath leading into the driver... Clamping helps to act like a heatsink of sorts, so the driver doesn't over heat, sticking some shrinkwrap tubing to it also functions the same way.

 

[tranhieu]

Don't use clamp... use blutak instead...
 
I also used breadboard in the past, but as I got rid of most passive components overtime the filters now are soldered directly to the wire instead. It's more secured that way and it takes only few seconds to switch the values.

 

[vector1]

  Don't use clamp... use blutak instead...
 
I also used breadboard in the past, but as I got rid of most passive components overtime the filters now are soldered directly to the wire instead. It's more secured that way and it takes only few seconds to switch the values.

Blutack is good too! I can't really work with acoustic tuning without measuring the FR which is why I'm mainly using electronic filters which are easier for me to simulate and tune.