Also, please follow Craig's instructions, as I am not familiar with WinPCD. PCD, i can usually get away without blending - you just need to ignore stuff under 300 Hz.
So I do need to take both tweeter and woofer near field measurements to input in the blender or just the woofer? Should these measurements be done the same way as the Farfield both at tweeter height and 1/4" from tweeter.
On the blender sheet toward the bottom are the adjust spl buttons to extract minimum phase what is that spl adjustment for? Do I need to match it to what I adjusted in the blender for the woofer.
The woofer just needs the near field. You can trim the tail with the tweeter in the Blender by adjusting the starting point of the tail.
When you measure near field the mic is 1/4 inch from the device under test.
The SPL adjust is helpful to put the FR curve in the middle of the screen for PCD. If you adjust the SPL you need to do it for every FR curve by the exact same amount. The SPL adjust can be used to also set a response to the factory SPL.
I don't agree Ani. Based on my experience the only way to generate accurate phase data is to have the FR extended as far as possible above and below for each driver and then extract their minimum phase.
Ok thank you again gentlemen. My plan is to grab the nearfield measurements for the woofers when I get off of work tonight. I will fiddle with the excEl sheet more and hopefully everything will make more sense when I get the correct files.
If I make headway I'll post some pics. It's kind of a hassle because I lost my office disc so I have to Rylan my old pc to use excel but I use my new one for omnimic because it's significantly more quiet.
I believe the reason the new blender tutorial doesn't mention blending in the near field measurement is that you can create a viable working FRD just using the box simulation on the bottom, and blend that into a gated FF measurement using blender.
Craig points out the importance of using blender (or another tool) to get the roll-offs tacked on for proper minimum phase extraction. If you are going the minimum phase route, that is a critical step.
= Howard Stark: "This is the key to the future. I'm limited by the technology of my time, but one day you'll figure this out."
I don't agree Ani. Based on my experience the only way to generate accurate phase data is to have the FR extended as far as possible above and below for each driver and then extract their minimum phase.
Thanks Craig. I wasn't looking at the phase only at the FR below gating.
Also, I don't completely follow regarding the tail... what are we looking for while adjusting the tail?
One advantage to using a simulated near field for woofers is you do not have to screw around with merging the port response to get a reasonably accurate response curve. With rare exceptions, most drivers are virtually ruler flat below 500 hz, so blending a simulated response curve with a measured far field response (provided your gating extends below 500 Hz) should not present any serious problems.
Ok, I think, I got something backwards. Mike, sorry to derail you...
Are we talking about blending the near field measurements with the mic very close to the dust cap to avoid room reflection with the far field response, taken with the mic say 1m away, which includes room interactions at lower frequency and then blending to get a good response of the woofer... As specified in Jeff B's accurate measurement to 10hz?
Or are we talking about blending the far field response with the woofer box simulation response which simulates the woofer inbox resonpse?
When and where do we use what?
And just to confirm, for tweeters we need only the far field?
And just to confirm, for tweeters we need only the far field?
For the tweeters, far field only yes. Though you still want to run them through blender, using a 12db roll-off attached below say 500-1000hz (tweeter dependent) to make sure the minimum phase comes out right.
= Howard Stark: "This is the key to the future. I'm limited by the technology of my time, but one day you'll figure this out."
Ok i ran the NF measurements for both woofers. It ended up being 3/8's away. I turned the volume down because it was way too high so I just level matched it at 90SPL like i did for the other measurements. This is what I came up with when I tried to splice it. It's not calculating the minimum phase on my pc. I am not sure why or what I am missing.
I loaded my FF woofer measurements on the DIFF_FRD_DATA tab I loaded the NF woofer on the LF_FRD_DATA tab I loaded the FF Tweeter measurement on HF_FRD_DATA tab
Something that concerns me is that I had to lower the "SPL to be added to the LF data set" by -85db so I don't know if I am missing anything.
Well, first things first - the blender is for blending near field and far field response of a single driver. For example, you take a gated far field measurement of your woofer and import that as the HF. Then you take a near field or do a simulation and import that as the LF. Then you adjust your levels, select your blending region, etc.
The "HF" and "LF" designations are not at all intuitive when using this software. I am guessing you can use it to import a gated, full system response and splice it with a simulated or near field response of the systems bass driver, though. When I get home from work, I will see what I can do for creating a quick how-to on using this.
Let's say I go with this woofer blending even though it does look kinda funky. What are the next steps. I just save these frd's as seperate "woofer A and B blended" do i leave the tweeters as is and input directly to pcd?
Messed around last night with the current model. I know I don't have everything I need in the model perfect but for a first prototype crossover its not bad. I don't have all the exact parts on hand so im off on a few values but it's fairly representative of the model. There is work to do in the 2k around for the woofer/tweeter xover point and a reduction in the top end. I'll try a larger resistor on the tweeter.
I lowered the time from peak to just under 2 seconds but the low end still looks weird. I don't know if I did the low end blending correctly so maybe that is causing it too.
Two milliseconds is a pretty short gate - you must really be in a reflective environment. I know you mentioned you were, but that is really unusual. Most rooms you should be able to grab at least a 3.5 millisecond clean measurement. If you have extra comforters, pillows, whatever around it can pay to drape/cover stuff. Also, whatever your mic distance you really want to try and keep any other object in the room that far away from it - preferable twice as far.
Another thing you can do to get a feel for how your speakers are really performing is to take a measurement at 1cm from the woofer cone, then maybe 6cm, 12cm, 24cm, then 36cm, etc up to your standard distance. What you will glean from this process is where the baffle step loss kicks in and also where your room starts dominating - you will also see some other interesting effects if you have your tweeter hooked up at the same time. Kind of fun, actually.
It takes some time to get a handle on measurements, enjoy the ride and take lots of notes. It will eventually become second nature
^ I got to the part were I need to make a frame and grill and set the project aside. A week ago I bought some screen door track with the rubber roll in for the screen mesh. My plan is to miter the ends and use that to hold in the speaker grill cloth. Hopefully the works out.
Some good progress has been made on the passive xover front. I emailed my measurements to Craig S. And he was nice enough to blend them correctly for me. So thanks Craig! I got them into pcd and worked for a few hours. I got a pretty good sounded build right now. Not perfect but a nice elevated bottom end that I wanted. I ended up at 2k hz xover with 2nd order on woofer and 3rd on tweeter. 8 parts total 3 being resistors.
I tried a few xover points from 2000 to 3000 but 2k was the easiest for me to get a decent response. I need to measure the current mock up xover with decently close values to see how it matches the model but from what I hear I suspect it will be very close to watch the model predicts.
I think im on to something good here now. The bass is strong as I prefer and compared to the previous versions I've simmer donna'ed everything past 10k it had a pretty healthy rise after that before. I'm using a 5ohm and 10 ohm resistor on the tweeter ill monitor it to see if they get crazy hot or not. Aside from the crazy low end spikes it's plus or minus 3db.
Comments
On the blender sheet toward the bottom are the adjust spl buttons to extract minimum phase what is that spl adjustment for? Do I need to match it to what I adjusted in the blender for the woofer.
The woofer just needs the near field. You can trim the tail with the tweeter in the Blender by adjusting the starting point of the tail.
When you measure near field the mic is 1/4 inch from the device under test.
The SPL adjust is helpful to put the FR curve in the middle of the screen for PCD. If you adjust the SPL you need to do it for every FR curve by the exact same amount. The SPL adjust can be used to also set a response to the factory SPL.
If I make headway I'll post some pics. It's kind of a hassle because I lost my office disc so I have to Rylan my old pc to use excel but I use my new one for omnimic because it's significantly more quiet.
Craig points out the importance of using blender (or another tool) to get the roll-offs tacked on for proper minimum phase extraction. If you are going the minimum phase route, that is a critical step.
Also, I don't completely follow regarding the tail... what are we looking for while adjusting the tail?
Are we talking about blending the near field measurements with the mic very close to the dust cap to avoid room reflection with the far field response, taken with the mic say 1m away, which includes room interactions at lower frequency and then blending to get a good response of the woofer... As specified in Jeff B's accurate measurement to 10hz?
Or are we talking about blending the far field response with the woofer box simulation response which simulates the woofer inbox resonpse?
When and where do we use what?
And just to confirm, for tweeters we need only the far field?
I loaded my FF woofer measurements on the DIFF_FRD_DATA tab
I loaded the NF woofer on the LF_FRD_DATA tab
I loaded the FF Tweeter measurement on HF_FRD_DATA tab
Something that concerns me is that I had to lower the "SPL to be added to the LF data set" by -85db so I don't know if I am missing anything.
The "HF" and "LF" designations are not at all intuitive when using this software. I am guessing you can use it to import a gated, full system response and splice it with a simulated or near field response of the systems bass driver, though. When I get home from work, I will see what I can do for creating a quick how-to on using this.
Is this approximately what I should be striving for on the blending.
Another thing you can do to get a feel for how your speakers are really performing is to take a measurement at 1cm from the woofer cone, then maybe 6cm, 12cm, 24cm, then 36cm, etc up to your standard distance. What you will glean from this process is where the baffle step loss kicks in and also where your room starts dominating - you will also see some other interesting effects if you have your tweeter hooked up at the same time. Kind of fun, actually.
It takes some time to get a handle on measurements, enjoy the ride and take lots of notes. It will eventually become second nature
I tried a few xover points from 2000 to 3000 but 2k was the easiest for me to get a decent response. I need to measure the current mock up xover with decently close values to see how it matches the model but from what I hear I suspect it will be very close to watch the model predicts.