I had a pair of Tang Band W5-2143 FR drivers on hand for some time and wanted to make a TL box for them but never got around to it.
Open baffle has always intrigued me but the W5-2143 did not seem well suited. Nonetheless I decided to give it a try anyhow and purchased a few IKEA door and drawer panels and used some scrap board cedar for the frame rails with the generous help of a friend with a router.
I pondered on which 12” woofer to use and with my thinking in keeping everything cheap, I selected GRS 12PF-8 woofers and a 100 Hz low-pass filter.
Once all was assembled and drivers hooked up the sound was decidedly unimpressive with the GRS woofers providing very minimal bass response. The Tang Bands were sounding quite bright and not too nice. I decided to remove the low pass filters and see what resulted. It was much, much better! But I was confused.
The GRS are rated 35Hz to 5000Hz and I thought surely that frequency overlap with the Tang Band FR would sound like a mess but it doesn’t at all, rather it is quite clean in the whole lower frequencies all the way up to the higher mids.
I have the DIY OB paired with two REL T-Zero subs and I think it sounds quite good. But I would like to learn why running the GRS pair with the Tang Band unfiltered seems to somehow work? Any ideas, or did I just get simply lucky in the pairing?
Try to be nice in your comments because I truly have little idea what I am doing and just liked the look of the Pure Audio so tried to (poorly) copy them. I think for one, I have a mismatch in the sensitivity of the 12” woofers (87dB) and the 5” FR driver (90dB) and seems I have read the woofer needs to be higher sensitivity but I failed to find an inexpensive one so went with the GRS. Despite that, as mentioned, the combo sounds quite good to my ears, realizing I could very well be biased as I might be forcing myself to approve since it was my “creation”.
Open baffle has always intrigued me but the W5-2143 did not seem well suited. Nonetheless I decided to give it a try anyhow and purchased a few IKEA door and drawer panels and used some scrap board cedar for the frame rails with the generous help of a friend with a router.
I pondered on which 12” woofer to use and with my thinking in keeping everything cheap, I selected GRS 12PF-8 woofers and a 100 Hz low-pass filter.
Once all was assembled and drivers hooked up the sound was decidedly unimpressive with the GRS woofers providing very minimal bass response. The Tang Bands were sounding quite bright and not too nice. I decided to remove the low pass filters and see what resulted. It was much, much better! But I was confused.
The GRS are rated 35Hz to 5000Hz and I thought surely that frequency overlap with the Tang Band FR would sound like a mess but it doesn’t at all, rather it is quite clean in the whole lower frequencies all the way up to the higher mids.
I have the DIY OB paired with two REL T-Zero subs and I think it sounds quite good. But I would like to learn why running the GRS pair with the Tang Band unfiltered seems to somehow work? Any ideas, or did I just get simply lucky in the pairing?
Try to be nice in your comments because I truly have little idea what I am doing and just liked the look of the Pure Audio so tried to (poorly) copy them. I think for one, I have a mismatch in the sensitivity of the 12” woofers (87dB) and the 5” FR driver (90dB) and seems I have read the woofer needs to be higher sensitivity but I failed to find an inexpensive one so went with the GRS. Despite that, as mentioned, the combo sounds quite good to my ears, realizing I could very well be biased as I might be forcing myself to approve since it was my “creation”.
Attachments
Cool to see that Pure i20 dock; are you using its built in DAC - which I found surprisingly good- or the TOSLINK out to “a real” DAC?
Re: Pure i-20.
Actually both ways, I too find the onboard DAC rather decent, but also have the Toslink output to a Cal Audio Sigma II old-school tube DAC for a bit of change in flavour as mood suits.
Actually both ways, I too find the onboard DAC rather decent, but also have the Toslink output to a Cal Audio Sigma II old-school tube DAC for a bit of change in flavour as mood suits.
The problem is that you have (roughly) reproduced the physical appearance of the PAP type speaker, but the rest seems to be left up to chance or you just randomly applied some crossover filtering that is not a good match for the task.
Here is a hint why things don't sound "right" - a speaker in an open baffle will have a response (without any filters added to it) that is similar to having the driver HIGH pass filtered, that is to say some lower frequencies are rolled off. You have to get an idea of what this highpassed response will look like in order to design a crossover for the loudspeaker that will make it sound roughly OK.
The easiest way to do that, and something that I suggest to everyone trying this kind of thing, is to download the program "TheEdge" by Tolvandata (Tolvan Data). This program lets you do some modeling of the respnose(s) of one or more drivers of the same size in an open baffle (make sure to check the box for the open baffle response!).
So, using the program, first model what the response from the two woofers will look like by simulating your baffle and placing two drivers on it where they are located in your project. Set the listening distance to e.g. 3m and position the mic on the listening axis, e.g. near the top center of the baffle. The program creates the OB "baffle loss" curve for you. This has to be added on top of the response of the drivers, which are essentially in free air so you can model that using a box modeler for a closed box repsonse and with the box volume set to a very large value that is at least 100 times larger than Vas. The sum of the infinite closed box response and the open baffle loss curve is approximately what the driver response is like in free space. Do this for the other fullranger as well. Then you get an idea of what you are up against and you can design and apply some filtering and EQ to flatten out the responses.
Good luck. Post here if/when you need help or advice.
Here is a hint why things don't sound "right" - a speaker in an open baffle will have a response (without any filters added to it) that is similar to having the driver HIGH pass filtered, that is to say some lower frequencies are rolled off. You have to get an idea of what this highpassed response will look like in order to design a crossover for the loudspeaker that will make it sound roughly OK.
The easiest way to do that, and something that I suggest to everyone trying this kind of thing, is to download the program "TheEdge" by Tolvandata (Tolvan Data). This program lets you do some modeling of the respnose(s) of one or more drivers of the same size in an open baffle (make sure to check the box for the open baffle response!).
So, using the program, first model what the response from the two woofers will look like by simulating your baffle and placing two drivers on it where they are located in your project. Set the listening distance to e.g. 3m and position the mic on the listening axis, e.g. near the top center of the baffle. The program creates the OB "baffle loss" curve for you. This has to be added on top of the response of the drivers, which are essentially in free air so you can model that using a box modeler for a closed box repsonse and with the box volume set to a very large value that is at least 100 times larger than Vas. The sum of the infinite closed box response and the open baffle loss curve is approximately what the driver response is like in free space. Do this for the other fullranger as well. Then you get an idea of what you are up against and you can design and apply some filtering and EQ to flatten out the responses.
Good luck. Post here if/when you need help or advice.
Adding to what Charlie has said...
The W5 has a slight rising response, but one that small baffle, there will be little bass and the rise will feel much steeper.
Asking the W5 to get down to 100Hz will be impossible here. I'd be surprised if it reached 300Hz, actually. So, you have a gap between 100 and 300, plus the rising response of the W5. Yep, will sound bad. Taking out the low pass will fill that gap, but there will be too much interactions with the drivers, you need a XO.
Do it right. Buy a USB mic ($100) to do measurements, that way, it won't be shots in the dark, trying to be lucky. In audio, being lucky with a design is like winning the lottery. Doesn't happen that much!
The W5 has a slight rising response, but one that small baffle, there will be little bass and the rise will feel much steeper.
Asking the W5 to get down to 100Hz will be impossible here. I'd be surprised if it reached 300Hz, actually. So, you have a gap between 100 and 300, plus the rising response of the W5. Yep, will sound bad. Taking out the low pass will fill that gap, but there will be too much interactions with the drivers, you need a XO.
Do it right. Buy a USB mic ($100) to do measurements, that way, it won't be shots in the dark, trying to be lucky. In audio, being lucky with a design is like winning the lottery. Doesn't happen that much!
I went ahead and did it for you. See attached. The speaker is modeled in the Edge as shown in the first image. I had to do it in two steps, one with the two woofers, and then a second for the middle FR driver. This listening axis is 3m away directly in front of a speaker at the height indicated by "mic". I plotted these on the same plot so you can compare the responses. It is important to note that this is ONLY the open baffle gain/loss curve and does not include the drivers' own free-air responses. Those have to be modeled separately. But we can keep those in mind and take a look at the effect of using the OB to see what might happen.
The first thing you can see is the OB gain/loss curve drops off below about 300Hz for both drivers. This depends on the baffle size only and that is why it is similar for both drivers. This is the HIGH pass effect I mentioned above.
The next thing that you can see is that the woofer gain/loss curve drops like a rock above about 700Hz. This is due to the fact that the driver is larger and with two of them you will be very far off axis from at least one of them. This causes interference, thus the dip. Can't use the woofers about about 700Hz for this reason.
Then there is the fullranger. The effect of the OB gain/loss is a significant dip in the on-axis response between 1k and 2k Hz. That's not good since you want the response to be smooth above where you cross over from the woofers. This is caused by the central mountnig of the fullranger in the baffle, and the fact that the fullranger is small compared to the baffle size, so it is like a point source. This is a major weakness of the PAP design IMHO. There is nothing you can do to completely "fix" this.
In the third plot I show the on and off axis responses for the fullrange driver. As you move off axis the response falls in level, e.g. at 300Hz on axis is the top curve and as you move off axis you move down the curves. It is easier to see how there is a hole around 1.5k Hz. Also, the depth of the dip is different for different off axis angles. This means the crossover cannot EQ the dip out the same for each angle. So you will have to live with it to some extent. It is only about 3-4dB but you will hear that. You can try to fill it in with some EQ and it might be OK.
So, with that info in mind, here is how I would approach the crossover design. Luckily you have subs. I would either cross over as high as possible from subs to woofers, boost up the woofer response below 300Hz, or cut the woofer response above your desired crossover point to the subs. This will flatten out the lower part of the response. Next you need to cross over to the fullranger, and that has to happen no higher than about 500-600 Hz. Then use some EQ to fill in the 1.5kHz dip. Finally you will need to balance the relative levels from the two woofers and the fullranger in order fort he overall tonal balance to be accurate. After that do some voicing EQ.
The first thing you can see is the OB gain/loss curve drops off below about 300Hz for both drivers. This depends on the baffle size only and that is why it is similar for both drivers. This is the HIGH pass effect I mentioned above.
The next thing that you can see is that the woofer gain/loss curve drops like a rock above about 700Hz. This is due to the fact that the driver is larger and with two of them you will be very far off axis from at least one of them. This causes interference, thus the dip. Can't use the woofers about about 700Hz for this reason.
Then there is the fullranger. The effect of the OB gain/loss is a significant dip in the on-axis response between 1k and 2k Hz. That's not good since you want the response to be smooth above where you cross over from the woofers. This is caused by the central mountnig of the fullranger in the baffle, and the fact that the fullranger is small compared to the baffle size, so it is like a point source. This is a major weakness of the PAP design IMHO. There is nothing you can do to completely "fix" this.
In the third plot I show the on and off axis responses for the fullrange driver. As you move off axis the response falls in level, e.g. at 300Hz on axis is the top curve and as you move off axis you move down the curves. It is easier to see how there is a hole around 1.5k Hz. Also, the depth of the dip is different for different off axis angles. This means the crossover cannot EQ the dip out the same for each angle. So you will have to live with it to some extent. It is only about 3-4dB but you will hear that. You can try to fill it in with some EQ and it might be OK.
So, with that info in mind, here is how I would approach the crossover design. Luckily you have subs. I would either cross over as high as possible from subs to woofers, boost up the woofer response below 300Hz, or cut the woofer response above your desired crossover point to the subs. This will flatten out the lower part of the response. Next you need to cross over to the fullranger, and that has to happen no higher than about 500-600 Hz. Then use some EQ to fill in the 1.5kHz dip. Finally you will need to balance the relative levels from the two woofers and the fullranger in order fort he overall tonal balance to be accurate. After that do some voicing EQ.
Attachments
Thanks for the input perceval. Charlie, I really appreciate you taking the time to do that. I did try but didn't fully understand what to do. Your detailed explanation does make sense to me.
Now I will have to do some research on crossover design to see how to properly implement the important frequency response information you have provided.
Again, I very much appreciate your help!
Now I will have to do some research on crossover design to see how to properly implement the important frequency response information you have provided.
Again, I very much appreciate your help!
Charlie, if you move the W5 towards an edge, not dead centered, it will probably resolve the 1.5kHz dip.
I moved my FR off center on my OB to counter that.
I moved my FR off center on my OB to counter that.
Sure, that can help, but you can chase your tail down that path.
I go the extreme route and completely eliminate the baffle! This has both advantages and disadvantages (like anything else) but these keep me on my toes and that is the way I like to build speakers these days! 🙂
I go the extreme route and completely eliminate the baffle! This has both advantages and disadvantages (like anything else) but these keep me on my toes and that is the way I like to build speakers these days! 🙂
I can recommend the software "Basta" if you want to simulate your speaker.
It practically contains "Edge", plus you can add passive and active crossovers.
It should give you a pretty good idea how to get a good sum of responses of your drivers. I´m sure you´ll get some help here if you post your progress.
I had lots of fun using the SW and getting pretty realistic results.
Basta is probably not as advanced as VituixCad but is also much quicker to get to work (IMHO). As the SW is old you will have to run it in some compatibility mode under the Windows system (at least I had to).
As a starting point you can maybe use your phone to measure the response and see if you are on a good path and your ears too.
Good luck and have lots of fun. I´m pretty sure you´ll end up with something really nice.
Nice job on the looks BTW! Now that is done you only need a couple crossover parts.
I´m almost sure the Tangband won´t play well down to 100Hz on a baffle so that may be the reason it´s much better without the filter. How did you realize the 100Hz filter?
It practically contains "Edge", plus you can add passive and active crossovers.
It should give you a pretty good idea how to get a good sum of responses of your drivers. I´m sure you´ll get some help here if you post your progress.
I had lots of fun using the SW and getting pretty realistic results.
Basta is probably not as advanced as VituixCad but is also much quicker to get to work (IMHO). As the SW is old you will have to run it in some compatibility mode under the Windows system (at least I had to).
As a starting point you can maybe use your phone to measure the response and see if you are on a good path and your ears too.
Good luck and have lots of fun. I´m pretty sure you´ll end up with something really nice.
Nice job on the looks BTW! Now that is done you only need a couple crossover parts.
I´m almost sure the Tangband won´t play well down to 100Hz on a baffle so that may be the reason it´s much better without the filter. How did you realize the 100Hz filter?
Thanks Joe, I will check out Basta and see if I can make any sense of it.
As for the 100Hz low pass filter choice, I asked a friend (who does car speaker installations-which probably means nothing regarding home stereo) and after giving him the specs of the woofers and FR driver, he recommended trying a 100Hz low pass.
As you and pereceval and Charlie before have noted, the W5 does not come anywhere close to being able to deliver 100Hz in this application. Though I showed a picture to my friend of the open baffle design I was using, obviously he had no specific experience with simulating a proper response for that design so I think he was simply working off the "specs" of the FR driver and woofer. I had even less experience to question whether that 100Hz choice made sense or not.
As for the 100Hz low pass filter choice, I asked a friend (who does car speaker installations-which probably means nothing regarding home stereo) and after giving him the specs of the woofers and FR driver, he recommended trying a 100Hz low pass.
As you and pereceval and Charlie before have noted, the W5 does not come anywhere close to being able to deliver 100Hz in this application. Though I showed a picture to my friend of the open baffle design I was using, obviously he had no specific experience with simulating a proper response for that design so I think he was simply working off the "specs" of the FR driver and woofer. I had even less experience to question whether that 100Hz choice made sense or not.
Again, thank you to all the gentlemen that have contributed to this thread. Does anyone have any experience with the ART CX-311 2-way electronic crossover? This approach seems it could be the most flexible to allow me to dial in and experiment with the crossover frequencies with the only downside being the need for additional amplification (or at least how it appears to me).
Yes, you will need as many amp channels as you have outputs.
I have no idea of the audio quality adding another piece of gear.
I do have a minDSP 2x4 HD that I used to go active with my OBs, but I only use the USB digital input, not its analogue section, which degrades the sound a bit.
I have no idea of the audio quality adding another piece of gear.
I do have a minDSP 2x4 HD that I used to go active with my OBs, but I only use the USB digital input, not its analogue section, which degrades the sound a bit.