Hi Zero D,
Looks like I somehow managed to delete the answer to your question in Post #118, but it looks like you got it before it disappeared. I also wanted to say thank you for reminding me of the "Inductance thread"; I have added a note to my notes.
For others that may have the same question: the first T in T-TQWT stands for tapped as the driver is physically inside the inverted horn (duct, line) which is tapped like a tapped horn and radiates to the outside only through the mouth (terminus, opening). That's why in Hornresp the simulation has to be done as TH or TH1.
Regards,
Looks like I somehow managed to delete the answer to your question in Post #118, but it looks like you got it before it disappeared. I also wanted to say thank you for reminding me of the "Inductance thread"; I have added a note to my notes.
For others that may have the same question: the first T in T-TQWT stands for tapped as the driver is physically inside the inverted horn (duct, line) which is tapped like a tapped horn and radiates to the outside only through the mouth (terminus, opening). That's why in Hornresp the simulation has to be done as TH or TH1.
Regards,
I lost a reply as well.
Oliver.
I didn't mean to put down your design when I posted the equivalent ported enclosure comparison.
Your design has a few interesting attributes.
A comparison of group delay, impulse response and impedance is where the most revealing attributes will be displayed.
Oliver.
I didn't mean to put down your design when I posted the equivalent ported enclosure comparison.
Your design has a few interesting attributes.
A comparison of group delay, impulse response and impedance is where the most revealing attributes will be displayed.
I'm glad you're not uptight. Now, instead of just standing around, why don't you do something useful. You've indicated you knew how to come up with more accurate sims for those HT18s. I wonder if you could use the parameters for Oliver's T-TQWT design and do the voodoo that you do?
It might be interesting to see how your calculations compares with the results Oliver got using the SI T/S parameters.
Would you be willing to do that, JAG? I'd like to see it.
Bach On
Ok, since you asked nicely, and since Drednofred wanted to see details of the tweak, I'll do a quick run through so you can see exactly what I'm doing.
First it's necessary to find out if the driver needs the tweak in the first place. Since it's a very large coil/motor driver we can assume that it does, but since we have a measurement available we can prove it just by checking the measurement.
Here's the 2V measurement of the prototype SI HT 18 from data-bass.com.
This driver is NOT a high qts driver so it should NOT have a humped up response with a rolloff on either side of 40 hz. Sealed boxes with this qts value should have a rising response, rising up into the midbass, not a humped up peak halfway through the bass. Check the prosound drivers he measured in this same box for an idea of what a rising response from a driver not affected by this issue looks like.
So let's do a quick sim to see if a simple sim does the job or if the tweaked sim is more accurate. This sim uses the data-bass measured t/s parameters and the same box size and dimensions as the box data-bass used to measure this driver.
The blue line in the sim is a simple (non tweaked) sim, the red line is tweaked.
Comparing to the measurement above we see that the simple sim is nowhere near accurate, the tweaked sim is exponentially more accurate. Looking at a few points, the tweaked sim has the same hump at 40 hz peaking at about 90 db, it's down about 2 db at 100 hz and down about 10 db at 20 hz.
Brian Steele likes to look at impedance and that's not a bad idea at all. This tweak does not affect impedance AT ALL but let's look at impedance anyway.
Measured -
Compared to simulated (tweaked and untweaked impedance is exactly the same so you only see the red line) -
Close enough, and I think it would look closer if that notch in the measured response wasn't there, then it would peak a hz or two higher in frequency and match a bit better. Not sure what caused that notch in the measurement but if it wasn't there it would be a closer match for sure, not that it's not already pretty close.
So how is the tweak done?
This is the original generic tweak, the simplest possible way of applying the tweak. No attempt is made to get the most accurate result, this exact same tweak can be applied to all large coil/motor drivers and you will get a pretty accurate result.
First double up Re (or add Rg equal to Re). Then add power until the circled part of the displacement graph matches up.
That's it. Super easy and simple and exponentially more accurate than a simple untweaked sim every time 100 percent of the time with this type of driver.
If you don't believe it works every time with every driver of this type in any enclosure of any type, just ask, I've got examples of tweaked and untweaked sims compared to measurements of a variety of drivers (measured by data-bass) in sealed, ported, tapped horn and front loaded horn enclosures.
Now on to the original question. Here's the design in post 66 with the tweak applied.
In both graphs the blue line is the simple untweaked sim, the red line is tweaked. Top graph is stuffing included, bottom graph is stuffing removed.
I GUARANTEE my tweaked sim is more accurate than the untweaked sim.
And if this seems alarming, I did mention this issue a long time ago, and it will also affect your ported box that you built. If you took it outside and measured it you will see a humped up response a few hz above tuning and your 16 hz response is going to be rolled off in a similar way that this design is. I did mention this at the time but nobody seemed really that interested so I didn't push it since it will still beat the living hell out of what you had before. Ported boxes and all the different styles of tl, tqwt, blh, tapped horn are particularly badly affected by this issue and if not accounted for a seemingly good looking sim will measure VERY badly. Sealed boxes and front loaded horns are also affected but the effect is not nearly as bad and much easier to work with.
I could sim your ported box too and if you measure it outside the sim WILL match the measurement. But it's going to look much the same, the humped up response above tuning and a roll off below that, making your 16 hz a lot weaker than it could be.
First it's necessary to find out if the driver needs the tweak in the first place. Since it's a very large coil/motor driver we can assume that it does, but since we have a measurement available we can prove it just by checking the measurement.
Here's the 2V measurement of the prototype SI HT 18 from data-bass.com.
An externally hosted image should be here but it was not working when we last tested it.
This driver is NOT a high qts driver so it should NOT have a humped up response with a rolloff on either side of 40 hz. Sealed boxes with this qts value should have a rising response, rising up into the midbass, not a humped up peak halfway through the bass. Check the prosound drivers he measured in this same box for an idea of what a rising response from a driver not affected by this issue looks like.
So let's do a quick sim to see if a simple sim does the job or if the tweaked sim is more accurate. This sim uses the data-bass measured t/s parameters and the same box size and dimensions as the box data-bass used to measure this driver.
The blue line in the sim is a simple (non tweaked) sim, the red line is tweaked.
An externally hosted image should be here but it was not working when we last tested it.
Comparing to the measurement above we see that the simple sim is nowhere near accurate, the tweaked sim is exponentially more accurate. Looking at a few points, the tweaked sim has the same hump at 40 hz peaking at about 90 db, it's down about 2 db at 100 hz and down about 10 db at 20 hz.
Brian Steele likes to look at impedance and that's not a bad idea at all. This tweak does not affect impedance AT ALL but let's look at impedance anyway.
Measured -
An externally hosted image should be here but it was not working when we last tested it.
Compared to simulated (tweaked and untweaked impedance is exactly the same so you only see the red line) -
An externally hosted image should be here but it was not working when we last tested it.
Close enough, and I think it would look closer if that notch in the measured response wasn't there, then it would peak a hz or two higher in frequency and match a bit better. Not sure what caused that notch in the measurement but if it wasn't there it would be a closer match for sure, not that it's not already pretty close.
So how is the tweak done?
This is the original generic tweak, the simplest possible way of applying the tweak. No attempt is made to get the most accurate result, this exact same tweak can be applied to all large coil/motor drivers and you will get a pretty accurate result.
First double up Re (or add Rg equal to Re). Then add power until the circled part of the displacement graph matches up.
An externally hosted image should be here but it was not working when we last tested it.
That's it. Super easy and simple and exponentially more accurate than a simple untweaked sim every time 100 percent of the time with this type of driver.
If you don't believe it works every time with every driver of this type in any enclosure of any type, just ask, I've got examples of tweaked and untweaked sims compared to measurements of a variety of drivers (measured by data-bass) in sealed, ported, tapped horn and front loaded horn enclosures.
Now on to the original question. Here's the design in post 66 with the tweak applied.
In both graphs the blue line is the simple untweaked sim, the red line is tweaked. Top graph is stuffing included, bottom graph is stuffing removed.
An externally hosted image should be here but it was not working when we last tested it.
I GUARANTEE my tweaked sim is more accurate than the untweaked sim.
And if this seems alarming, I did mention this issue a long time ago, and it will also affect your ported box that you built. If you took it outside and measured it you will see a humped up response a few hz above tuning and your 16 hz response is going to be rolled off in a similar way that this design is. I did mention this at the time but nobody seemed really that interested so I didn't push it since it will still beat the living hell out of what you had before. Ported boxes and all the different styles of tl, tqwt, blh, tapped horn are particularly badly affected by this issue and if not accounted for a seemingly good looking sim will measure VERY badly. Sealed boxes and front loaded horns are also affected but the effect is not nearly as bad and much easier to work with.
I could sim your ported box too and if you measure it outside the sim WILL match the measurement. But it's going to look much the same, the humped up response above tuning and a roll off below that, making your 16 hz a lot weaker than it could be.
Last edited:
Some additional data to show that this method works every time with every driver of this type in every enclosure type. Some more exotic enclosure types this time with different drivers. Exactly the same tweak applied.
Mach 5 Audio UXL in Submaximus front loaded horn. Here's the sim vs the measurement. Green is measured response.
Not even vaguely close. Here's my inductance corrected prediction. Red is corrected in this sim blue is uncorrected, compare the sims to the measured response in the picture above.
I ran the numbers with Ricci's measured t/s for the UXL and ZV3 and compared with Ricci's (data-bass guy) measurements in Othorn tapped horn.
Each sim has regular results in light grey and my inductance compensated results in dark black under Ricci's measurement.
As predicted, my inductance compensation method is pretty close. The uncompensated sims are not even vaguely accurate.
Disregard actual spl levels, they mean nothing here. I did NOT increase power to offset added Re losses so the graphs would be easier to view (they wouldn't be right on top of each other). So sensitivity in the tweaked sims is not right but frequency response is correct and easily comparable to the untweaked response and the measured response.
Ricci mentioned these are indoor close mic measurements and explained that's why response droops a bit as frequency increases in the measurements.
First pic is UXL, second is ZV3.
As I mentioned, this issue causes nice looking ported, tl, blh and tapped horn sims to turn into big steaming piles of crap when built and measured. But sealed and flh are much less affected due to their impedance profile, and easier to manage. It's because of how the impedance peaks are spaced. In the flh the impedance peaks are closer together and evenly spaced so the impedance peaks keep the response propped up fairly well for a smooth(ish) response. In the ported, tl, blh and tapped horn enclosures the impedance peaks are widely spaced and not even spaced out, so the impedance peaks prop up the response in irregular spots (where the impedance peaks occur with no support at the lowest impedance peak below tuning) giving a terrible looking response with a rolled off output down at tuning.
Mach 5 Audio UXL in Submaximus front loaded horn. Here's the sim vs the measurement. Green is measured response.
Not even vaguely close. Here's my inductance corrected prediction. Red is corrected in this sim blue is uncorrected, compare the sims to the measured response in the picture above.
An externally hosted image should be here but it was not working when we last tested it.
I ran the numbers with Ricci's measured t/s for the UXL and ZV3 and compared with Ricci's (data-bass guy) measurements in Othorn tapped horn.
Each sim has regular results in light grey and my inductance compensated results in dark black under Ricci's measurement.
As predicted, my inductance compensation method is pretty close. The uncompensated sims are not even vaguely accurate.
Disregard actual spl levels, they mean nothing here. I did NOT increase power to offset added Re losses so the graphs would be easier to view (they wouldn't be right on top of each other). So sensitivity in the tweaked sims is not right but frequency response is correct and easily comparable to the untweaked response and the measured response.
Ricci mentioned these are indoor close mic measurements and explained that's why response droops a bit as frequency increases in the measurements.
First pic is UXL, second is ZV3.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
As I mentioned, this issue causes nice looking ported, tl, blh and tapped horn sims to turn into big steaming piles of crap when built and measured. But sealed and flh are much less affected due to their impedance profile, and easier to manage. It's because of how the impedance peaks are spaced. In the flh the impedance peaks are closer together and evenly spaced so the impedance peaks keep the response propped up fairly well for a smooth(ish) response. In the ported, tl, blh and tapped horn enclosures the impedance peaks are widely spaced and not even spaced out, so the impedance peaks prop up the response in irregular spots (where the impedance peaks occur with no support at the lowest impedance peak below tuning) giving a terrible looking response with a rolled off output down at tuning.
Last edited:
Very meticulous work, JAG. I'm not technically proficient to evaluate the process. But it is clear you have paid attention to the details.
It also appears that your SPL chart for Oliver's T-TQWT box is similar to his original.
You stated: "the humped up response above tuning and a roll off below that, making your 16 hz a lot weaker than it could be."
Is this a limitation with which I must live, or is there something that can be done to overcome this?
Thanks for your efforts. I do appreciate it.
Bach On
It also appears that your SPL chart for Oliver's T-TQWT box is similar to his original.
You stated: "the humped up response above tuning and a roll off below that, making your 16 hz a lot weaker than it could be."
Is this a limitation with which I must live, or is there something that can be done to overcome this?
Thanks for your efforts. I do appreciate it.
Bach On
Hi JAG and BO,
The problem I have w/ this is that while it definitely looks like needs to be addressed, and JAG has provided an easy fix it requires the building of a test box, as well as fairly accurate measurements to arrive at a curve against which to adjust the value of Rg (or add to Re), also the other method of dividing Bl by 1.414 (multiply by .707) does not give quite the same results, so what is correct?
As to fixing the response curve of the simulation, that is easy, as it requires just moving the enclosures dimensions in the simulation with the Fix/Tweak in place. Sadly, in this case it adds another ~ 100L to the box and looses a little bit of SPL.
This also brought to mind, that the actual resistance of the cable between amplifier and speaker should be entered into the simulation as Rg.
On the bright side, at least BO now knows what a T-TQWT looks like. 🙂
Regards,
P.S.: Still thinking (this could take a while).
The problem I have w/ this is that while it definitely looks like needs to be addressed, and JAG has provided an easy fix it requires the building of a test box, as well as fairly accurate measurements to arrive at a curve against which to adjust the value of Rg (or add to Re), also the other method of dividing Bl by 1.414 (multiply by .707) does not give quite the same results, so what is correct?
As to fixing the response curve of the simulation, that is easy, as it requires just moving the enclosures dimensions in the simulation with the Fix/Tweak in place. Sadly, in this case it adds another ~ 100L to the box and looses a little bit of SPL.
This also brought to mind, that the actual resistance of the cable between amplifier and speaker should be entered into the simulation as Rg.
On the bright side, at least BO now knows what a T-TQWT looks like. 🙂
Regards,
P.S.: Still thinking (this could take a while).
There's a 5 db difference at around 22 hz between the red (tweaked) and blue) untweaked lines. The rest of the graph is very similar but the humped up area is pretty different.
Technically 16 hz is about the same level in the tweaked and untweaked sim, but the big hump above 16 hz will make 16 hz seem weak compared to 22 hz.
If you wanted to do something about that you could tune lower so the big hump moves down to 16 hz and/or make the box a lot bigger, which very likely is not an option.
Or you could just leave it the way it is an eq it flat with whatever processing you have available. This approach will cut excursion a bit in the critical range above tuning so you can go a bit louder before hitting xmax (assuming you are using a hpf and the area above tuning is your excursion max).
But before you do anything to the box I'd measure the response in the listening area at several points and see how things look there. The room might flatten the sub's response out or it might make it worse. No way to know until you measure.
Nice presentation Anthony.
A little tip.
At times you can export your simulated results and display them in a REW graph together.
It gives you the same vertical and horizontal scales.
A little tip.
At times you can export your simulated results and display them in a REW graph together.
It gives you the same vertical and horizontal scales.
Some of the software I use incorporates this tweak already. But it is nice to see someone using his head and figuring it out a bit.
Definitely the t/s parameters should be measured. The measurements were from a prototype driver not a production unit, so t/s specs may be different and this is just an intellectual exercise without properly measured t/s parameters.
A test box does not need to be built though, at this point there's enough information about this driver to know that the production unit does indeed have this large coil/motor issue so the tweak is necessary to get a more accurate sim.
I haven't had time to even try the Bl reduction method that LTD02 came up with. It's going to be a big job so I haven't bothered to even start the process. I'm going to have to use his method to compare sims to measurements of the few dozen high quality measurements I have available and confirm that that tweak method is at least as good or better than mine.
Either way, it should be pretty similar. It's not going to be exactly the same but it will be similar.
The next step would be to take the more accurate of the two methods and adjust to see if there's a better fit than a simple doubling of Re (or .707x Bl).
But at this point, my simple original generic tweak is pretty accurate as it is. If you compare all the sims and measurements I posted here, they match pretty well and exponentially better than the untweaked sims.
I always use the exact same tweak (double Re and add power to compensate) and it has been effective (not perfectly accurate but exponentially more accurate than an untweaked sim) 100 percent of the time.
My measurement mic is dead at the moment and I haven't got a new one yet so I don't even have REW at this time, but I probably should get it anyway. This is a good feature.
Are you talking about the software that uses the complex inductance t/s parameters? If so that would be a better way to do it, but then you need software that will measure those parameters (since manufacturers don't provide them) and you would need a simulator that will accept them. None of the simulators I use will accept them.
What software do you use to measure the parameters and what simulator do you use that will accept them (and it needs to be able to sim horns)? Switching software would be a lot easier than refining the tweak to be more accurate.
Yes sir it is.
I'll post something in terms of simulation of an enclosure using the HT18. But pretty busy at the moment. Lots of tweeters to get shipped.
Last edited:
Do you know of any simulators that can do tl, flh, blh, tapped horn that can accept these parameters? I don't, so even if I could measure the parameters I wouldn't be able to use them unless you know of a simulator that can do it all. And if you don't have the driver and measure it yourself it's impossible to find these complex inductance t/s parameters, which is another issue.
Like I said, this would be the far better path, but if you can't get the parameters and you can't find a fully featured simulator that will accept them, my tweak is the only way forward that I can see.
Like I said, this would be the far better path, but if you can't get the parameters and you can't find a fully featured simulator that will accept them, my tweak is the only way forward that I can see.
Last edited:
There are some, but your talking serious coin.
As in good used car amounts of money.
Short but sweet you simulate, you build you measure.
Let me know if you want a measurement microphone.
If you have a smart phone you have other options as well.
As in good used car amounts of money.
Short but sweet you simulate, you build you measure.
Let me know if you want a measurement microphone.
If you have a smart phone you have other options as well.
Iterations are not an option for me anymore. My next project (if I decide to do it) is going to cost $3000, at least a couple hundred hours of work, 1200 lbs (600 lbs per box). I can't afford to make mistakes. And it will be using this type of large coil/motor driver.
Check my sims. My sims are always very accurate. My folding is very accurate. I have nearly zero worries that my sims won't match the measurement.
I do, but I want a very specific mic, this one. Cross·Spectrum - Calibrated MiniDSP UMIK-1 Microphones for Sale
Professionally calibrated by a reputable 3rd party and built in preamp for $100 US + shipping. I don't want to mess around with preamps and complicated setups, this is plug and play.
I've never had a cell phone and don't really want one. I have an Iphone 4s available if I wanted to use it but buying a decent mic from PE for it is $16 last time I checked and decent measurement software is not free either. And I haven't even looked into porting the measurements into the computer and what hassles that will bring. Not much more expensive to just buy the mic I want.
Dayton Audio UMM-6 USB Measurement Microphone
It's the same mic.
And why pay for a secondary calibration?
For your measurement requirements the Dayton is more than you need already.
As for measurement accuracy and the validity of the extra calibration I have a little news for you.
Keep your mic in your car one afternoon and your calibration is out the window.
Freeze your mic and your calibration is out the window.
Keep your money in your pocket and get the Dayton.
The Umik is the same mic.
If you want a considerably better cheaper alternative that you need to build let me know.
I can get capsules that Earthworks uses.
And I to have microphones and calibrators for comparison purposes.
It's the same mic.
And why pay for a secondary calibration?
For your measurement requirements the Dayton is more than you need already.
As for measurement accuracy and the validity of the extra calibration I have a little news for you.
Keep your mic in your car one afternoon and your calibration is out the window.
Freeze your mic and your calibration is out the window.
Keep your money in your pocket and get the Dayton.
The Umik is the same mic.
If you want a considerably better cheaper alternative that you need to build let me know.
I can get capsules that Earthworks uses.
And I to have microphones and calibrators for comparison purposes.
The Dayton cal files are garbage, it's well documented. I have several links that show the Dayton cal file is WAY off. Also the neck on the UMM mic is very fragile. Cross Spectrum won't even ship a UMM out of the US because so many necks are broken during the shipping process while it's still in it's padded box.
The price difference between the UMM from PE and the UMIK from Cross Spectrum is less than $12.
I might consider saving the $25 and getting a UMIK directly from MiniDSP but I don't want the UMM, it's too fragile.
I don't want to solder up a preamp either, or have an external preamp at all. If you have something that has a built in preamp I would strongly consider that. But how are you going to calibrate it unless you build it and test it before sending it off? If I pay you fairly for your time the UMIK is still cheaper.
It's pretty important to me that the preamp is built in. I want the preamp to be part of the calibration, I don't want an extra variable, more parts to hook up, and have to verify the accuracy of the entire setup and all it's assorted pieces. Too much work when you can get a calibrated plug and play usb mic for $100 (or $75 direct from manufacturer minus the pro 3rd party cal).
The price difference between the UMM from PE and the UMIK from Cross Spectrum is less than $12.
I might consider saving the $25 and getting a UMIK directly from MiniDSP but I don't want the UMM, it's too fragile.
I don't want to solder up a preamp either, or have an external preamp at all. If you have something that has a built in preamp I would strongly consider that. But how are you going to calibrate it unless you build it and test it before sending it off? If I pay you fairly for your time the UMIK is still cheaper.
It's pretty important to me that the preamp is built in. I want the preamp to be part of the calibration, I don't want an extra variable, more parts to hook up, and have to verify the accuracy of the entire setup and all it's assorted pieces. Too much work when you can get a calibrated plug and play usb mic for $100 (or $75 direct from manufacturer minus the pro 3rd party cal).
Hi Y'all,
Oh my, now you're opening up a whole other can of worms. But, your are correct an uncalibrated measurement system will probably not provide you w/ great measurements. Especially the low end of all these mikes is suspect, well, and the high end. LOL
What's a guy to do? 🙂
Regards,
P.S.: (It's just a rethorical observation, so - please - don't let it bother you.)
Oh my, now you're opening up a whole other can of worms. But, your are correct an uncalibrated measurement system will probably not provide you w/ great measurements. Especially the low end of all these mikes is suspect, well, and the high end. LOL
What's a guy to do? 🙂
Regards,
P.S.: (It's just a rethorical observation, so - please - don't let it bother you.)
- Status
- Not open for further replies.