-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Help with measurements
- Thread starter thatoneguy
- Start date
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #22
Thanks, got it
I made that paper towel roll encasing for the mic and got some new measurements.
The red measurement is the old measurement with just the mic. The black measurement is the new measurement with the paper towel encased mic.
Let me know what you think.
The new waterfall response
New spectrogram
Only slightly different. I guess I should focus on removing those resonances on the waterfall plot. The diaphragm is far far from being optimized as it is a tedious task to craft one.
solderdude
Grand Contributor
Only Maty optimizes things
To figure out if things are reflections you can calculate the distance between the peaks (is a bounce path so divide the lenght by 2 to see if it is a reflection. Most likely they are resonances. Also there is filter ringing from the soundcard and bandwidth limiting of the mic, mic pre-amp and ADC.
Where does the increased bass come from ?
I assume a measurement error or the membrane is not tensioned.
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #24
What is this filter ringing you speak of?
Tension doesn't have a large effect on the bass unless I over tension it.
Although the tension is a bit on the light side. Tensioned just enough so it doesn't sag, mainly just because it sounds better. Tight tension sucks the life out. I'm sure I could get better step response with higher tensions but I would need to solve the "suck of life out of the sound" problem.
The bass is just a result of the driver configuration based on a theory I had a couple years ago. Between this and an amp I've been working on for years that has no measurable distortion under any load with a theoretical 20khz distortion below -300db I'm planning on starting a little business at some point when I can find someone like-minded that wants to help.
That being said, based on my aforementioned theory which is currently working out well, the current driver is very crude and has significant room for theoretical improvement. I have months of work to do before I'll have an optimized driver, at least from a theoretical perspective.
Last edited:
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #25
It seems that the resonances are all due to the crudness of the diaphragm and the sub par manner it is mounted to the frame.
I'm using a clamping mechanism to hold it down but the clamp isn't uniformly tight.
I went through a manual sweep 1 frequency at a time and was able to eliminate every resonance just by making small adjustments to the diaphram deformities and pressing down on the clamping mechanism. Unfortunately I can't really do it all at once. I need to design a better mounting mechanism and a less crude diaphram.
Anyway this is the updated waterfall plot after tweaking it as much as I can.
The resonances are almost entirely less than 2ms other than the tips of those resonances near 1k and 2k which are both clamp related.
However keeping the resonances below 2ms is considered sufficient isn't it? At least that's what I read.
I'm using a clamping mechanism to hold it down but the clamp isn't uniformly tight.
I went through a manual sweep 1 frequency at a time and was able to eliminate every resonance just by making small adjustments to the diaphram deformities and pressing down on the clamping mechanism. Unfortunately I can't really do it all at once. I need to design a better mounting mechanism and a less crude diaphram.
Anyway this is the updated waterfall plot after tweaking it as much as I can.
The resonances are almost entirely less than 2ms other than the tips of those resonances near 1k and 2k which are both clamp related.
However keeping the resonances below 2ms is considered sufficient isn't it? At least that's what I read.
solderdude
Grand Contributor
Gibbs phenomenon would be a better description. It is seen before the pulse and thus also present after the pulse.
I use a squarewave generator + scope to view the waveform so the limit is the microphone itself.
solderdude
Grand Contributor
Do you have any AP measurements on this amp ? How do you circumnavigate the noise to show this low distortion ? by averaging over time ?
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #28
No AP measurements yet I've been trying to find someone that can do that for me for quite some time.
The numbers I have are based on many many simulations and extrapolations based on many many bench tests. Direct measurements are impossible, it will just look like a loopback. Creative practices are required. I won't elaborate here because I could probably write an entire chapter on all the tests I've done and it will completely derail the thread. To make a long story short I spent about a year trying to disprove that it was as good as the sims say it is.
The only thing I want out of it now is a reputable third party to measure it so I have official data to go to investors. It's small and cheap to make so would easily sell on the market. Same with my other inventions including these headphones I'm working on. Too much potential to walk away.
The numbers I have are based on many many simulations and extrapolations based on many many bench tests. Direct measurements are impossible, it will just look like a loopback. Creative practices are required. I won't elaborate here because I could probably write an entire chapter on all the tests I've done and it will completely derail the thread. To make a long story short I spent about a year trying to disprove that it was as good as the sims say it is.
The only thing I want out of it now is a reputable third party to measure it so I have official data to go to investors. It's small and cheap to make so would easily sell on the market. Same with my other inventions including these headphones I'm working on. Too much potential to walk away.
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #30
Yeah I know, I almost forgot about that actually. He said he won't be able to do it for me for quite some time so it's just a waiting for at the moment until he's free.
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #31
Well, that's interesting.
I accidentally destroyed my right driver so I had to make another one. Check out the results of this one.
Keep in mind NO smoothing is applied to the measurement. and 1/48 smoothing was applied to the CSD.
A naturally flat and non-jagged response.
No echos in the impulse response this time.
I took 6 other measurements that are all practically the same.
Then I attached the driver to the headphone and got a number of measurements like this
When the drivers are connected to the headband I've just been measuring the outside of the drivers because I can't fit the mic between them. The outside is functionally equivalent to the inside so the result would be the same. However, either something changed during the transition to the headband or the sound is reflecting off the other driver and my crude wood block headband. Not sure why else such a change would occur.
I accidentally destroyed my right driver so I had to make another one. Check out the results of this one.
Keep in mind NO smoothing is applied to the measurement. and 1/48 smoothing was applied to the CSD.
A naturally flat and non-jagged response.
No echos in the impulse response this time.
I took 6 other measurements that are all practically the same.
Then I attached the driver to the headphone and got a number of measurements like this
When the drivers are connected to the headband I've just been measuring the outside of the drivers because I can't fit the mic between them. The outside is functionally equivalent to the inside so the result would be the same. However, either something changed during the transition to the headband or the sound is reflecting off the other driver and my crude wood block headband. Not sure why else such a change would occur.
Last edited:
It's probably a combination of reflections from the headband, plus the headband itself resonating.
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #33
Hmmm, what if I stick the mic next to the driver at a right angle and put tape or something to cover the driver opening and mic. Would that work?Otherwise I need to disassemble it for all measurements.
Having to disasemble it would depend on what you're trying to measure. For gauging raw driver performance it might be best to leave it dissembled, but for the measuring the whole headphone you need to account for those resonances and reflections in the design of the headband.
OP
thatoneguy
Member
- Joined
- Feb 17, 2019
- Messages
- 20
- Likes
- 0
- Thread Starter
- #35
The headband is currently just a bunch of oversized blocks of wood I found, glued into a rectangular headband shape and screwed to the drivers. Definitely not something I want to "account for in a real design
I expect a final design will have no such issues so I disassembled the drivers to get a real look at what they can do.
I need to solve the low frequency resonance. Some sort of damping material perhaps, that won't skew the already near perfect response.
Any advice?
There are a number of things I can do from a driver development perspective to attempt to solve it but those tests will come later.
.Regardless of how it measures, the sound stage and sound quality are indistinquishable from reality so I don't really want to mess with them anymore, I already ruined the driver twice in 2 days. I'll wait to make a replica before I continue development and use these as my reference.
I expect a final design will have no such issues so I disassembled the drivers to get a real look at what they can do.
I need to solve the low frequency resonance. Some sort of damping material perhaps, that won't skew the already near perfect response.
Any advice?
There are a number of things I can do from a driver development perspective to attempt to solve it but those tests will come later.
.Regardless of how it measures, the sound stage and sound quality are indistinquishable from reality so I don't really want to mess with them anymore, I already ruined the driver twice in 2 days. I'll wait to make a replica before I continue development and use these as my reference.
You'll just need to experiment. There are all kinds of fabrics, felts, papers, and other things like fiberglass, rock wool, and even various kinds of first aid/medical tape you can try. You can, vary thickness, density, distance from driver, shapes and cutouts. All kinds of things.
Just go through a whole bunch of stuff. Make a change, measure it and then go on to the next. Most of the thing you try will ruin the sound and that will be obvious from the measurements. Once you've found a few with measurements that look good, actually listen to them and then make modifications from there.
You might want to look though headphone modding threads on Head-Fi or SBAF to get some inspiration on materials or techniques.
I'm not away of any published work on the methodical construction of headphones in the same way there are pleanty of books about DIY speaker design so you just need to wing it. Try lots of things, and confirm your listening impressions with measurements.
Similar threads
