Similar magnitude problem happens with eg. accelerometer measurements of box panels. How to convert that to spl at 1 meter?
This kind of measurements are relative, and I like to watch distortion from nearfield or 1m measurement of a speaker. Peaks of 3rd harmonic are often mechanical resonances from eg. loose screws or wires. Then just use REW's signal generator to play sine at peak F and listen and palpate!
This kind of measurements are relative, and I like to watch distortion from nearfield or 1m measurement of a speaker. Peaks of 3rd harmonic are often mechanical resonances from eg. loose screws or wires. Then just use REW's signal generator to play sine at peak F and listen and palpate!
Here a compilation of simulation and measurements of a testbox for the Purifi PTT8.0X08-NAB02
Standing modes-Box calculation:
TIB:
Note: It looks like the 3 close modals around 490Hz created sidebands, but it turned out there were also panel resonances in that range. So difficult to separate.
(I could not find a TIB after reinforcing the testbox panels.)
Impedance measurements, with the usual large Rsense, 27 Ohm.
Empty box not reinforced (no damping stuff ):
Note: The wiggle between5-10Hz is the very flexible springmounting of box.
The main cabinet modes are clearly present, as are some other resonances.
Now with 600grams of wool as damping:
Pretty effective in mitigating the main cabinet modes, also something else seems to become (more) visible. This is not just the spider resonance.
In fact in first image it is also observable although somewhat masked by cabinet modes.
Experimenting with fibration measurements of panels etc, I decided to reinforce the panels with another layer of MDF, and , and some strenghtening of seams and baffle. Also with the metal thread inserts for mounting driver much better pressed into the wood.
Since then I used another PTT8.0X08 (other serial number), and as one can see what remains are some driver resonances.
I also switched to an Rsense of 0.135 Ohms. Note I use a power resistor of 6.8 Ohms for calibrating the LIMS application, this resulted in an Rsense value of 0.135 Ohms as apparently “seen” by LIMS. I have no tool to measure accurately such low resistance values.
000 gr wool:
With 700grm of wool:
Or superimposed, with same y-scale as above:
Standing modes-Box calculation:
TIB:
Note: It looks like the 3 close modals around 490Hz created sidebands, but it turned out there were also panel resonances in that range. So difficult to separate.
(I could not find a TIB after reinforcing the testbox panels.)
Impedance measurements, with the usual large Rsense, 27 Ohm.
Empty box not reinforced (no damping stuff ):
Note: The wiggle between5-10Hz is the very flexible springmounting of box.
The main cabinet modes are clearly present, as are some other resonances.
Now with 600grams of wool as damping:
Pretty effective in mitigating the main cabinet modes, also something else seems to become (more) visible. This is not just the spider resonance.
In fact in first image it is also observable although somewhat masked by cabinet modes.
Experimenting with fibration measurements of panels etc, I decided to reinforce the panels with another layer of MDF, and , and some strenghtening of seams and baffle. Also with the metal thread inserts for mounting driver much better pressed into the wood.
Since then I used another PTT8.0X08 (other serial number), and as one can see what remains are some driver resonances.
I also switched to an Rsense of 0.135 Ohms. Note I use a power resistor of 6.8 Ohms for calibrating the LIMS application, this resulted in an Rsense value of 0.135 Ohms as apparently “seen” by LIMS. I have no tool to measure accurately such low resistance values.
000 gr wool:
With 700grm of wool:
Or superimposed, with same y-scale as above:
https://www.diyaudio.com/community/threads/drive-current-distortion-measurement.402566/
#81 there
"volume: 100l
fill: 4,375 kg sonorock (idea: Esa Meriläinen)
resistor: 20R"
Best regards
Bernd
@JanRSmit Great, great work
Thank you for measuring
!
in your picture: "Now with 600grams of wool as damping:" ?
Maybe: ??
Best regards
Bernd
I was not able to measure without the sonorock, as it is part of the loudspeaker now (glued shut).
Thank you for measuring
!you refer to the wiggles around 300-500Hz?
in your picture: "Now with 600grams of wool as damping:" ?
Maybe: ??
Best regards
Bernd
Last edited:
Regarding USB microphones: why do people use them?
I can tell you what my thought process was when I first started. I had no experience with sound recording equipment or procedures, and no experience with microphones. I looked at the setup for a two channel rig, and saw that it required something called an audio interface, which I had never heard of. The mics required cables with XLR fittings, which were not available from the local computer store. The online manuals for the audio interface used pro-recording terminology which I was unfamiliar with. There seemed to be multiple gain adjustments in the signal path, and multiple ways to get clipping. It seemed like there was a lot of potential for a newbe to make errors.
For my first entry back into speaker design after 25 years, I wanted something plug-and-play. The Pocket CLIO looked ideal, but I was reluctant to pay the price. So I bought the OmniMic.
Later when I moved up to an XLR mic and an audio interface, I was grateful I had learned the basics of measuring a driver with a basic USB mic. The OmniMic provided a baseline to compare my measurements to, and no surprise, I was making several errors in setting up the audio interface. I also had a defective Dayton EMM6 microphone. I was able to resolve all of this because I had a solid USB mic to compare to.
I have not used the OmniMic in several years, but I am grateful I started with it.
j.
I can tell you what my thought process was when I first started. I had no experience with sound recording equipment or procedures, and no experience with microphones. I looked at the setup for a two channel rig, and saw that it required something called an audio interface, which I had never heard of. The mics required cables with XLR fittings, which were not available from the local computer store. The online manuals for the audio interface used pro-recording terminology which I was unfamiliar with. There seemed to be multiple gain adjustments in the signal path, and multiple ways to get clipping. It seemed like there was a lot of potential for a newbe to make errors.
For my first entry back into speaker design after 25 years, I wanted something plug-and-play. The Pocket CLIO looked ideal, but I was reluctant to pay the price. So I bought the OmniMic.
Later when I moved up to an XLR mic and an audio interface, I was grateful I had learned the basics of measuring a driver with a basic USB mic. The OmniMic provided a baseline to compare my measurements to, and no surprise, I was making several errors in setting up the audio interface. I also had a defective Dayton EMM6 microphone. I was able to resolve all of this because I had a solid USB mic to compare to.
I have not used the OmniMic in several years, but I am grateful I started with it.
j.
The wiggles or small variation in impedance curve in thecrange 300 to 500Hz is most likely related to panel resonances. In the later measurements they do not show or very minimal.@JanRSmit Great, great work
I was not able to measure without the sonorock, as it is part of the loudspeaker now (glued shut).
Thank you for measuring
you refer to the wiggles around 300-500Hz?
in your picture: "Now with 600grams of wool as damping:" ?
Maybe: ??
Best regards
Bernd
Thanks for sharing this story!
For real, it fascinates me what different background people have and start something.
I was already surrounded with diy people who used measuring systems.
So knowledge and experience was already widely shared.
Most people in these days used either speakerworkshop or boxsim with just single channel measurements.
With WinXP this only gave minor artifacts but nothing to worry about.
Although I think a lot has to do with personality as well.
I am extremely curious in nature (even when I was a small kid).
So I'll just somehow figure it out.
Of course it doesn't require an interface, the standard computer sound in/out will do. This is the way most did it in the days before. Mics that needed phantom power were sometimes served with a small DIY battery box (a few 9V) with a few non-critical resistors and capacitors.
Or phantom power injectors were being used if not (mostly) just being build ourselves.
Hey all, sincerely not trying to create controversy...just trying to understand motivations....
For me, I use manufacturers' T/S for initial box design. One sub's or low mid bass chambers only.
Adjusting params in hornResp doesn't seem to make the T/S matter much, compared to bigger picture volume, port area/length etc.
I've never ever thought to apply T/S to mid driver sections or above, in the initial design stage.
So my curiosity... what is the scrutiny for very fine specs about?
Simulations? Or do box designs change significantly with better T/S specs?
By significantly, I mean more than what taking transfer functions of the actual build will show.
Or is the scrutiny for predicting crossovers and EQs', implemented passively by LCR?
The scrutiny on predictive measurements doesn't make much sense to an active DSP guy like me, where realized acoustic measurements and the ability to act on them, is all i focus on.
it's hard to imagine how the big picture, physical box design goals, are very much effected by predictive nuances.
And once built, measurement reality takes over.
So wondering what the predictive measurement motivations are all about...
If it's cause you simply enjoy them, and the subject, that I totally get
For me, I use manufacturers' T/S for initial box design. One sub's or low mid bass chambers only.
Adjusting params in hornResp doesn't seem to make the T/S matter much, compared to bigger picture volume, port area/length etc.
I've never ever thought to apply T/S to mid driver sections or above, in the initial design stage.
So my curiosity... what is the scrutiny for very fine specs about?
Simulations? Or do box designs change significantly with better T/S specs?
By significantly, I mean more than what taking transfer functions of the actual build will show.
Or is the scrutiny for predicting crossovers and EQs', implemented passively by LCR?
The scrutiny on predictive measurements doesn't make much sense to an active DSP guy like me, where realized acoustic measurements and the ability to act on them, is all i focus on.
it's hard to imagine how the big picture, physical box design goals, are very much effected by predictive nuances.
And once built, measurement reality takes over.
So wondering what the predictive measurement motivations are all about...
If it's cause you simply enjoy them, and the subject, that I totally get
For me like you first step to reduce possible selection for bass and midrange drivers for sealed box .
Then when having the unit(s) test for real behaviour around Fs, and for mechanical issues. And i listen to the bare units at low signal volume to observe for issues.
A behaviour i learned back then when we did not have the measurement capabilities we have now.
Once in cabinet and stable in behaviour(damping etc is ok) i do not care about the TS parameters anymore.
Must say though that i measured my current system, and to my surprise after 18 years they still measure the same.
Last note: The impedance measurements are important for passive crossovers.
Then when having the unit(s) test for real behaviour around Fs, and for mechanical issues. And i listen to the bare units at low signal volume to observe for issues.
A behaviour i learned back then when we did not have the measurement capabilities we have now.
Once in cabinet and stable in behaviour(damping etc is ok) i do not care about the TS parameters anymore.
Must say though that i measured my current system, and to my surprise after 18 years they still measure the same.
Last note: The impedance measurements are important for passive crossovers.
- -
I tend to agree... Of course with the kind of speakers I (mostly) build, sealed box with LT EQ, exact bass modelling is not so important. But even a fully passive non-EQ sealed box system has a lot of forgiveness in it... What structures engineers called "fudge factor", which means a design which is not sensitive to tolerance variations or tolerance build up.
If we model a typical vented box system, with a typical bass reflex style woofer with a Qts in the range of 0.3 to 0.35, we can see that it actually takes a fair bit of variation in Vb, Fb, Fs, Vas, and Qts to drive the simulated response off the ideal curve by more than 1.5 dB. So even these sytems are somewhat forgiving.
There is a human tendency, when faced with a complex problem, to over-focus on those aspects within our control as a way of soothing our mind about the other aspects which are beyond our control. I have caught myself falling into this trap before... Could this be part of what we are doing? "So much complexity and unknowns about the speaker design process, but By Golly at least I can design the vent tuning frequency to 4 significant figures... 43.27 Hz is the RIGHT answer"
j.
OT, but I just noticed this
@JanRSmit
Btw, start a conversation with me if would like to chat about reducing/mitigating the resonance just above 400Hz.
Thanks for the replies, JanRSmit and hifijim,
Jim, I think that's a rather profound observation.
To help mitigate that tendency, I keep trying to ask myself "what matters most.....what aspect(s) offer the greatest marginal improvement potential".
Helps keep me big picture focussed...admittedly at the risk of missing the improvements finer scrutiny of aspects deemed less likely to have a larger impact.
Jim, I think that's a rather profound observation.
To help mitigate that tendency, I keep trying to ask myself "what matters most.....what aspect(s) offer the greatest marginal improvement potential".
Helps keep me big picture focussed...admittedly at the risk of missing the improvements finer scrutiny of aspects deemed less likely to have a larger impact.
That is absolutely the most important thing!
I call it the list with priorities as well as constraints.
Obviously we want all things to be as good as possible, but some things just have an higher significance then others.
I would also highly recommend making a list like that for yourself.
Because you will be surprised how many of these points are maybe actually quite subjective.
Things like, looks, budget, size, something that fits the space/room etc.
Although quite understandable, it also requires experience to develop a reliable sense of what matters most. That is the "tension" in this statement .
So in line with @hifijim objective of this thread, what are the aspects that matter most?
For me at this moment is the on- and off-axis fr-resp quality of the midrange, and to achieve this within the design limits i have (aka "battery limits" , see my post https://www.diyaudio.com/community/...unfinished-after-15-years.391827/post-7637506)
In other words how to model the baffle for the best contribution of the diffraction to the fr-resp quality.
To that end i made a choice to use AKABAK for simulation, to 3D print the baffle shape and to measure the fr-response. Time consuming for sure to get going , but i do not have a workshop space anymore to make physical baffles and test. However a sim has to correlate with the real thing, and luckily with AKABAK it iturned out to be close enough.
Why the focus on fr-respone of midrange? Wel in my experience the sound and the soundstage are very depenent on the response quality in the midrange. (The depth of the soundstage by the way mainly by the bass quality, thus my choice for sealed enclosures) All this assuming the recording is not screwed up wrt to soundstage ;-)
Once shape is selected, the physical construction and driver mounting, because cabinet issues (resonances, rattles) are killing for the soundstage (audibly locating the enclosure destroys the soundstage perception) and of course the sound itself.
Fair point. I guess i take the experience i've gained from a multitude of builds for granted sometimes.
My big picture design list starts like yours, with freq response on-and-off axis.
Next on the list is how low do I want the speaker to go.
Then, uncompressed, unclipped SPL for whatever average SPL the design goal is. This is for the entire spectrum, including lowest desired bass extension. Must also meet uncompressed, and unclipped for headroom above average SPL goal.
Fourth is flat phase across the spectrum. Whether or not phase is audible doesn't really matter to me....flat phase makes achieving prior goals, particularly the first one, much much easier ime.
And like you again, i key off the mid-range..or rather i should say I key off vocal clarity, intelligibility, and natural soundingness. I see vocals as predominantly mid-range.
Now through all that, other than the sub portion of a speaker, I never use sims of any kind.
It's too easy for me with a pretty decent garage workshop, to just slap a speaker on a temporary baffle, and grab a transfer function of it.
I can totally see without that ability, I'd be doing a lot more simulation work.
But for me, to answer the threads central question....the driver measurements i need for initial design are generally all contained in manufacturers published specs..
It's not too hard to pick out suitable candidates
Then, after a prototype is constructed, the measurements I need to refine the design are simply a bunch of transfer functions, and maybe an impulse response or two.