Hi Djim,
Nice. 🙂 I have a few newer toys than the old Tek too, but I just love that old thing. Hope things have been going well for you.
Regards,
Nice. 🙂 I have a few newer toys than the old Tek too, but I just love that old thing. Hope things have been going well for you.
Regards,
Hi Djim,
By the way, what's your take on just a guy's tweak/fix for the long coil overhang high power drivers (simulation v. measurement) w/ a high Le?
Regards,
By the way, what's your take on just a guy's tweak/fix for the long coil overhang high power drivers (simulation v. measurement) w/ a high Le?
Regards,
I understand that many will want to buy a calibrated mike they can trust. And my Dayton EMM-6 with the Dayton cal file probably isn't very reliable across the full audio spectrum. And if I was going to be doing a lot of these kinds of measurements, I might invest in something better. And looking at the linked threads, it appears the problems with the EMM-6 mikes is in the upper frequencies. I doubt our organ will be putting out many sounds above 18K Hz.
BO
BO
Last edited:
I've never had any problems getting decent measurements out of any of my Dayton mics, I've got the imm6 for phones, the emm6 and omnimicv2.
Hi Bach On,
Post #164: "... it appears the problems with the EMM-6 mikes is in the upper frequencies..."
Let's hope DrDyna is correct, and that is the case, but, let me repeat the link from Post #160:
http://www.diyaudio.com/forums/subwoofers/160879-build-your-own-2x12-th-kraken-212-th-22.html
The two response charts posted there by Djim show clearly that the low end is badly affected in some of these microphones/microphone capsules. If you are interested in response below 30Hz you have to make sure that your measurement system does not have the kind of roll-off as shown in these charts, which are typical.
The whole microphone/calibration thing is why I'm so upset about not being able to just take the T/S parameters into a simulation, and get reliable results, and why JAG's Tweak is so important.
And it is a measurement system problem as all the low cut (high pass) filters from the signal source through the preamplifier/amplifier/speaker/microphone/preamplifier/A/D are cumulative, and have to be evaluated at some point or another. Oh, and then there is the environment...
Regards,
Post #164: "... it appears the problems with the EMM-6 mikes is in the upper frequencies..."
Let's hope DrDyna is correct, and that is the case, but, let me repeat the link from Post #160:
http://www.diyaudio.com/forums/subwoofers/160879-build-your-own-2x12-th-kraken-212-th-22.html
The two response charts posted there by Djim show clearly that the low end is badly affected in some of these microphones/microphone capsules. If you are interested in response below 30Hz you have to make sure that your measurement system does not have the kind of roll-off as shown in these charts, which are typical.
The whole microphone/calibration thing is why I'm so upset about not being able to just take the T/S parameters into a simulation, and get reliable results, and why JAG's Tweak is so important.
And it is a measurement system problem as all the low cut (high pass) filters from the signal source through the preamplifier/amplifier/speaker/microphone/preamplifier/A/D are cumulative, and have to be evaluated at some point or another. Oh, and then there is the environment...
Regards,
I guess the blunt truth is that all the measurements in the World are really only to satisfy a collective curiosity. We have what we have. For good or for ill, that is the system we will use in the initial setup. 😱
BO
BO
I've got it. You are re-stating Plato's parable of the Cave, not just educating those of us urging you to measure to figure out just what "it" is you "have".
Well, that's swell, but I think I'll stick with Aristotle when it comes to acoustics.
Ben
Last edited:
Hi TB46,
Le is a product of the VC moving in a magnetic field driven by an AC current based on impedance. Only a loss-factor in the form of resistance can be simmed with Rg or Re. If you use Re it will show the influence on impedance. With the Stereo Integrity that seems to be the case.
The SI HT18D2 with an Le of 3.86 mH is probably build up with many turns in the VC to create 'height' for excursion; Xmax = (Hvc-Hg)/2. More turns means more losses (lossy-inductance). In order to stabilize impedance/inductance during travel, they used shorting ring(s).
Therefore your own 'old' method using a certain value for each mH (perhaps forgotten?), is still defensible from a theoretical point of view : -)
The differences in sensitivity between sim (after loss-correction) and measurements from Data-bass.com can be found by comparing both impedance plots. The 'average' value for Z is lower in reality. Zmin and the slopes of the curves are different too. This is a good example of the limitation of a sim. For more details I can suggest Leach's paper: "Loudspeaker Voice-Coil Inductance Losses".
Cheers,
Djim
Le is a product of the VC moving in a magnetic field driven by an AC current based on impedance. Only a loss-factor in the form of resistance can be simmed with Rg or Re. If you use Re it will show the influence on impedance. With the Stereo Integrity that seems to be the case.
The SI HT18D2 with an Le of 3.86 mH is probably build up with many turns in the VC to create 'height' for excursion; Xmax = (Hvc-Hg)/2. More turns means more losses (lossy-inductance). In order to stabilize impedance/inductance during travel, they used shorting ring(s).
Therefore your own 'old' method using a certain value for each mH (perhaps forgotten?), is still defensible from a theoretical point of view : -)
The differences in sensitivity between sim (after loss-correction) and measurements from Data-bass.com can be found by comparing both impedance plots. The 'average' value for Z is lower in reality. Zmin and the slopes of the curves are different too. This is a good example of the limitation of a sim. For more details I can suggest Leach's paper: "Loudspeaker Voice-Coil Inductance Losses".
Cheers,
Djim
Last edited:
Post #170
Hi Djim,
I get the basic premise, and have not forgotten about adjusting Le.
Where I have a problem is that this says the T/S parameters are of little value for designing the low end of subwoofer loudspeakers w/ long voice coils and high Le/Re ratio drivers; because these drivers are so non-linear in their behaviour that a fudge factor(s) must be used to arrive at useful simulations.
In the past I have mainly used the test box method for measuring/calculating T/S parameters. In this (as in the added mass method) method the measured values are frequencies, resistances and voltages, that define the shape of the resonance peak in free air, and the shape of the resonance peak in a test environment (in a sealed box, or with an added mass). From these values the remaining T/S parameters are calculated.
I feel that parameters acquired like this should reflect the behaviour of the driver, particularly around the respective resonance frequencies.
The driver in question here is the SI HT18. The fs of this driver has been listed as either 15Hz (data-bass.com) or 17Hz (manufacturer). We are looking to define the behaviour at the very low response range, w/ particular attention to 16Hz.
If the parameters derived from the free air response and the sealed test box behaviour of this driver cannot be used to design the bottom end of a subwoofer for this driver in a simulation, then the whole measurement/simulation thing would be of only questionable value to me. I don't want to go back to designing 1/4 Wave enclosures w/ cut and try methods in wood (Let me restate that: I will not go back.).
And, once I start adjusting the T/S values to fit my assumption of what they should be I feel like I'm on a slippery slope that leads me back to building multiple test boxes, etc...
On the one hand I get where just a guy comes from, and I appreciate his efforts, but I'm not quite there yet to just double Re (or something similar), and be done with it; on the other hand this means building a suitable test box, measuring the response, and adjusting T/S parameters until they fit the test box frequency response. Doable, but not very satisfying.
See why I have a problem w/ this?
Regards,
Hi Djim,
I get the basic premise, and have not forgotten about adjusting Le.
Where I have a problem is that this says the T/S parameters are of little value for designing the low end of subwoofer loudspeakers w/ long voice coils and high Le/Re ratio drivers; because these drivers are so non-linear in their behaviour that a fudge factor(s) must be used to arrive at useful simulations.
In the past I have mainly used the test box method for measuring/calculating T/S parameters. In this (as in the added mass method) method the measured values are frequencies, resistances and voltages, that define the shape of the resonance peak in free air, and the shape of the resonance peak in a test environment (in a sealed box, or with an added mass). From these values the remaining T/S parameters are calculated.
I feel that parameters acquired like this should reflect the behaviour of the driver, particularly around the respective resonance frequencies.
The driver in question here is the SI HT18. The fs of this driver has been listed as either 15Hz (data-bass.com) or 17Hz (manufacturer). We are looking to define the behaviour at the very low response range, w/ particular attention to 16Hz.
If the parameters derived from the free air response and the sealed test box behaviour of this driver cannot be used to design the bottom end of a subwoofer for this driver in a simulation, then the whole measurement/simulation thing would be of only questionable value to me. I don't want to go back to designing 1/4 Wave enclosures w/ cut and try methods in wood (Let me restate that: I will not go back.).
And, once I start adjusting the T/S values to fit my assumption of what they should be I feel like I'm on a slippery slope that leads me back to building multiple test boxes, etc...
On the one hand I get where just a guy comes from, and I appreciate his efforts, but I'm not quite there yet to just double Re (or something similar), and be done with it; on the other hand this means building a suitable test box, measuring the response, and adjusting T/S parameters until they fit the test box frequency response. Doable, but not very satisfying.
See why I have a problem w/ this?
Regards,
This is why Leach and others came up with complex inductance parameters, the simple t/s parameters are simply not adequate to define the behavior of some drivers. The complex parameters include a factor for lossy inductance and an extra component for resistance, very similar to what my tweak does, and the complex parameters are also derived right from the impedance curve. If you check REW it can measure t/s including complex inductance parameters. But there's no simulator that will accept these extra parameters.
So you see this isn't just fudging or poking around in the dark. My generic tweak doubles Re, and you can get a more accurate result by taking the measurement and finding the best amount of Re to match, but then you need a great deal of trust in the measurement. Simply doubling Re works fairly well for ALL drivers of this type and it's super easy so it's what I use when explaining the method and when I don't need the most accurate possible result.
And as mentioned adding Re and adjusting power makes some of the other graphs inaccurate so the Bl tweak may prove to be better, at least in that aspect.
Even so, the simple generic tweak is exponentially more accurate than a simple sim for this type of driver.
There's lots more I would like to say but my keyboard had a big drink of coffee and I have to use onscreen keyboard and type with the mouse and it takes forever. If you have any specific questions I can answer but I can't go on at length and go into great detail now.
BTW,adjusting Le alone doesn't work, not even remotely close, Re or Bl NEEDS to be adjusted.
Last edited:
Hi just a guy,
Post #172: "...There's lots more I would like to say but my keyboard had a big drink of coffee and I have to use onscreen keyboard and type with the mouse and it takes forever..."
That's OK, don't torture yourself w/ the onscreen keyboad/mouse typing. I got the concept, and your Tweak seems to be an improvement over just adjusting Le as far as the curve fit is concerned (e.g.: to the data-bass.com sealed box SPL curve).
Regards,
Post #172: "...There's lots more I would like to say but my keyboard had a big drink of coffee and I have to use onscreen keyboard and type with the mouse and it takes forever..."
That's OK, don't torture yourself w/ the onscreen keyboad/mouse typing. I got the concept, and your Tweak seems to be an improvement over just adjusting Le as far as the curve fit is concerned (e.g.: to the data-bass.com sealed box SPL curve).
Regards,
Just a bit more...
If we just had fs vas and qts you can get in the ballpark but not as accurate as using all the simple parameters. Adding the complex inductance t/s gets you more accurate in the same way.
But the only simulator that will accept them is Unibox, and it can't sim complex enclosures AND it won't accept the complex parameters REW generates - there's more than just Leach's model for these parameters, and manufacturers never give these specs.
So you are really stuck if you want an accurate sim, and this is why it's good to have such a simple tweak that works really well.
If we just had fs vas and qts you can get in the ballpark but not as accurate as using all the simple parameters. Adding the complex inductance t/s gets you more accurate in the same way.
But the only simulator that will accept them is Unibox, and it can't sim complex enclosures AND it won't accept the complex parameters REW generates - there's more than just Leach's model for these parameters, and manufacturers never give these specs.
So you are really stuck if you want an accurate sim, and this is why it's good to have such a simple tweak that works really well.
The problem with fudging with parameters in the model to make the sim'd FR match the measured FR is that by doing so your fudge is essentially including all the inaccuracies involved in that specific FR measurement and it may not work for another FR measurement made under different conditions.
That's why I think the impedance curve is considerably more important component of matching a sim to actual results. If the measured impedance curve is a match for the one produced by the fudged sim, then you may be onto something. If not, well...
I have to agree with you Brian.
Not negating Anthony's observations.
But it's not a sure thing. It's a curve fit. Not a calculated method.
Not negating Anthony's observations.
But it's not a sure thing. It's a curve fit. Not a calculated method.
As I mentioned previously the tweak doesn't affect impedance curve at all if you add the extra impedance via Rg (instead of adding extra Re). Reducing Bl does have a small effect but it just reduces the magnitude of the peaks a bit without shifting their location.
I addressed this specifically earlier because I know you guys (especially Brian) would question the effect on impedance, and because I also appreciate the importance of the impedance curve. I even showed the impedance curve sim vs measurement in my example, and it was a pretty good match. Not perfect match, but being less than perfectly accurate has nothing to do with the tweak as the tweak has no effect whatsoever on impedance.
As to Brian's 1st point, accurate measurements are essential and that's why I only use data-bass measurements (except the one flh measurement). He is very good and reliable. Probably not perfect like everything in this world but very very good and it's the biggest collection of reliable measurements of this type of drivers available.
Last edited:
Anyway, I posted 4 examples, a sealed box, a flh, and 2 tapped horns compared to Ricci and aarose's measurements. Can't argue with results, and correlation is quite accurate despite using the generic tweak (double Re) with no attempt made to adjust Re more accurately for best curve fit. Even the very simple generic version of the tweak works very well.
The flh example is least accurate but I don't think it was simmed or folded with high degree of accuracy AND it wasn't measured by Ricci but it's the only flh with this type of driver that I could find with measurements. And the 2 tapped horn measurements were indoor close mic with effects explained (it's hard to find reliable measurements of this type of drivers in exotic enclosures so I used everything I could find), but in all 4 cases correlation between sims and measurements is very good, exponentially better than an untweaked sim.
The flh example is least accurate but I don't think it was simmed or folded with high degree of accuracy AND it wasn't measured by Ricci but it's the only flh with this type of driver that I could find with measurements. And the 2 tapped horn measurements were indoor close mic with effects explained (it's hard to find reliable measurements of this type of drivers in exotic enclosures so I used everything I could find), but in all 4 cases correlation between sims and measurements is very good, exponentially better than an untweaked sim.
Last edited:
You are running against a few problems.
Our specs for the HT18 for example are taken from Josh's measurements. His measurement is on a prototype driver. And the measurements are three years old.
That's quite a few driver generations back.
It would be awesome to have some reliable measurements of a current driver.
And I just happen to know someone that has one. He is in Ottawa. I'll ask if I can measure it.
I'll post the data up on this thread if I can do the measurements.
Our specs for the HT18 for example are taken from Josh's measurements. His measurement is on a prototype driver. And the measurements are three years old.
That's quite a few driver generations back.
It would be awesome to have some reliable measurements of a current driver.
And I just happen to know someone that has one. He is in Ottawa. I'll ask if I can measure it.
I'll post the data up on this thread if I can do the measurements.
I already mentioned all this and you are correct.
As I said, without accurate t/s this is just an intellectual exercise, but at least in the case of my example it's fine, as I was just showing how the method works using the measurements of the prototype driver and using the prototype t/s as measured to correlate sim to measurements and show the tweak works.
But as I said, those are old specs and I wouldn't use them to sim a new driver, and I also don't trust published specs on the SI website.
- Status
- Not open for further replies.