Ok crossover question. I took the time today to measure all of the inductors, and everyone of them is a bit off. All of the steel core inductors are high, they read around 3.12 or so. I don't imagine that a precise 3.0 is that big a deal on that inductor, but if you feel I'm wrong, please let me know, I can take a few winds off and get them down. I then measured the aircore's used in the LCR and they are wrong enough that I'm sure it matters. The 1.5mh's measure more like 1.4, one of them is 1.65 even. I imagine getting those as close as possible is a good idea, no?
In the case of the series woofer inductor, placing the mounting screws raises the inductance even more, to around 3.3 or so. How big a deal is this now? Again, I can take a few windings off. Also, what impact does adding more metal to the core do? I know that some say, don't do it, some say, you simply have to take it into account, but what about core saturation. Does it increase or decrease the saturation point? Does it even matter (I'm guessing it doesn't)? It lowers dcr by the a minuscule amount, around .05 ohms.
When it comes time to assemble the crossover, should I bother measuring every little part and making sure its close enough. I'm especially concerned about the LCR's, since small changes can cause a pretty big frequency change.
The inductance of the iron core and its resistance is so much higher than expected I could probably use my 3.2mh copper foil inductors instead, which have a dcr of around .4 (They handle more power and have no core saturation issues, otherwise I know it doesn't matter much).
In the case of the series woofer inductor, placing the mounting screws raises the inductance even more, to around 3.3 or so. How big a deal is this now? Again, I can take a few windings off. Also, what impact does adding more metal to the core do? I know that some say, don't do it, some say, you simply have to take it into account, but what about core saturation. Does it increase or decrease the saturation point? Does it even matter (I'm guessing it doesn't)? It lowers dcr by the a minuscule amount, around .05 ohms.
When it comes time to assemble the crossover, should I bother measuring every little part and making sure its close enough. I'm especially concerned about the LCR's, since small changes can cause a pretty big frequency change.
The inductance of the iron core and its resistance is so much higher than expected I could probably use my 3.2mh copper foil inductors instead, which have a dcr of around .4 (They handle more power and have no core saturation issues, otherwise I know it doesn't matter much).
I don't ever measure individual components, but I often measure the completed crossover and always the final system. I have never seen any variations sufficient to make me concerned. I have seen a wide variation on numbers reported for individual components and I usually atribute this to the testor - person and the machine (no offense intended). I really don't know the sensitivity of the crossover to the inductance values, but I suspect from my experience that it isn't that great.
Im sorry Earl, its not something I like to say, but I have to respond and say Im surpriced that you an engineer take this position
Im surpriced that any xo components leaves your place without being measured and matched first
At least as a safety, taking care of your product quality
One of those small important things that could make it different from many other
Something that people will notice and remember, giving a good reputation
Now we have rather cheap instruments for measuring caps and inductors, and I cant imagine I will ever use any component before checking the actual value
Actually I match ALL components
Ideally drivers should be matched too, but not really practical
At my place there is an audible difference even with the smallest changes
Im surpriced that any xo components leaves your place without being measured and matched first
At least as a safety, taking care of your product quality
One of those small important things that could make it different from many other
Something that people will notice and remember, giving a good reputation
Now we have rather cheap instruments for measuring caps and inductors, and I cant imagine I will ever use any component before checking the actual value
Actually I match ALL components
Ideally drivers should be matched too, but not really practical
At my place there is an audible difference even with the smallest changes
I never have my hands on the crossover parts (they are sent directly from the supplier) so I can't test them. When I make the system myself I do test the final system and the crossver if necessary. If the final result works right then it doesn't matter if the components vary.
"At my place there is an audible difference even with the smallest changes"
This you would have to prove to me!! I don't believe it.
But I will admit to this - I need to supply the transfer function for the crossovers in the manual (and I will now that I realize this issue) so that the end customer can check the final assmebly if they want. Then they can decide if its "good enough" to the design intent. I can't do that for them.
I've never seen as much as a 1 dB difference between the measured transfer function and the desired one from off the shelf un-measured components.
I'll also tell you another problem which I don't know how to control any better than I already do, and that is the Compression Drivers impedance. I see huge variations in this impedance while seeing nothing in the measured response on the waveguide. (I just did a test of six DE250s. Identical FR, widely variant impedances). This impedance has a pretty large effect on the crossover - far greater than any of the component tollerances that you are talking about. The way I handle this is to use as small a swamping resistor across the driver as I can. This minimizes the sensitivity of the design to the variation in the drivers impedance. From what I have seen this driver impedance effect swamps out anything having to do with component tollerances.
"At my place there is an audible difference even with the smallest changes"
This you would have to prove to me!! I don't believe it.
But I will admit to this - I need to supply the transfer function for the crossovers in the manual (and I will now that I realize this issue) so that the end customer can check the final assmebly if they want. Then they can decide if its "good enough" to the design intent. I can't do that for them.
I've never seen as much as a 1 dB difference between the measured transfer function and the desired one from off the shelf un-measured components.
I'll also tell you another problem which I don't know how to control any better than I already do, and that is the Compression Drivers impedance. I see huge variations in this impedance while seeing nothing in the measured response on the waveguide. (I just did a test of six DE250s. Identical FR, widely variant impedances). This impedance has a pretty large effect on the crossover - far greater than any of the component tollerances that you are talking about. The way I handle this is to use as small a swamping resistor across the driver as I can. This minimizes the sensitivity of the design to the variation in the drivers impedance. From what I have seen this driver impedance effect swamps out anything having to do with component tollerances.
Only ABX matters.
gedlee said:
Perhaps they change the formula. 15 years ago Methanol was a standard denaturation additive to all alcohol based products including shellac and bin.
http://www.e-barnett.com/MSDS/002988.pdf
The point was very simple. Use caution and proper protection then you spray paint. If you get enough of this crap in your lungs, it'll bite your behind later on.
Why take chances?
I don't make it an issue. In my NaO II and Mini kits all I offer is the plan set and the other essential part, the printed circuit board(s) for the active circuits along with the parts list for the active and passive components, and drivers. It is up to the builder to decide where to source the parts from if he doesnt'hich to use my sources which works well for non USA builders. If so desired I offer the active circuits in various stages of completion, form stuffed and tested boards to completed active units.
I used to offer completed passive crossovers but I didn't receive a single request so I don't bother offering it any more. In fact, orders have been either for plans + PCBs or plans plus completed active units. Seems no one has a problem cutting wood, building the cabinets/panels, or assembling the passive crossovers but some are a little leery of building the active unit.
By the way, if you ship internationally check out Express Mail International. It is much cheaper than UPS, Fedex, etc. Example: UPS wanted over $180 to ship a NaO II AECU (about 8 lbs) to AU. Express Mail was around $60, fully insured. Plus, no additional customs fees.
I used to offer completed passive crossovers but I didn't receive a single request so I don't bother offering it any more. In fact, orders have been either for plans + PCBs or plans plus completed active units. Seems no one has a problem cutting wood, building the cabinets/panels, or assembling the passive crossovers but some are a little leery of building the active unit.
By the way, if you ship internationally check out Express Mail International. It is much cheaper than UPS, Fedex, etc. Example: UPS wanted over $180 to ship a NaO II AECU (about 8 lbs) to AU. Express Mail was around $60, fully insured. Plus, no additional customs fees.
Man I'm sorry I even brought it up. Well let me say that, when I brought this up, it was simply to find out if Dr. Geddes felt it was a concern, and I really was only concerned about the LCR's as they are designed to attack problems at a very specific frequency. How critical the values of those parts are typically matter's most when the notch is very deep. I asked simply to see if Dr. Geddes thought it was a problem. It appears that he does not consider it an issue.
I decided to load the driver responses and schematic into Speaker Workshop to model the differences that the tolerances make, and it's pretty minimal. The filters in the tweeter seem to still get what they are supposed to just fine. The inductors are easily within +/- 5% of their rated specs. While I hear companies talk about matching parts to +/- .5% or less, you have to understand how much of that is just marketing. It's very difficult to match inductors to even 1% with even the finest equipment on the market. I happen to know that some very high end companies use some of the identical measuring equipment to me for their crossover qc and part matching. All of the caps I have measured thus far are within 1% of their rated values. When all is said and done, I will be sure that these crossovers are close enough to not make an audible difference. In my opinion certain parts, such as the tweeter cap's and woofer inductors can be as much as 10-15% out of spec and still be fine. Keep in mind that drivers are rarely even 10% within spec (break in does help improve this some). I was shocked when I first started measuring my Focal drivers to find that many of them were as much as 10hz higher than spec for the fs, and were well broken in. Turns out the same was true of my dynaudio drivers, morel, Seas, Vifa, Peerless, and Dayton. Not a single driver I had fell all that close to spec, even after break in. It was still within 10% for most of the t/s parameters though.
-I'm sure user error is a part of this too. My equipment can be very accurate, but that doesn't mean I am. I try my best, and built many measuring jigs, but I don't have the ability to insure that there is no interference, moon phase problems, or whatever else might throw things off beyond my control or knowledge.
I decided to load the driver responses and schematic into Speaker Workshop to model the differences that the tolerances make, and it's pretty minimal. The filters in the tweeter seem to still get what they are supposed to just fine. The inductors are easily within +/- 5% of their rated specs. While I hear companies talk about matching parts to +/- .5% or less, you have to understand how much of that is just marketing. It's very difficult to match inductors to even 1% with even the finest equipment on the market. I happen to know that some very high end companies use some of the identical measuring equipment to me for their crossover qc and part matching. All of the caps I have measured thus far are within 1% of their rated values. When all is said and done, I will be sure that these crossovers are close enough to not make an audible difference. In my opinion certain parts, such as the tweeter cap's and woofer inductors can be as much as 10-15% out of spec and still be fine. Keep in mind that drivers are rarely even 10% within spec (break in does help improve this some). I was shocked when I first started measuring my Focal drivers to find that many of them were as much as 10hz higher than spec for the fs, and were well broken in. Turns out the same was true of my dynaudio drivers, morel, Seas, Vifa, Peerless, and Dayton. Not a single driver I had fell all that close to spec, even after break in. It was still within 10% for most of the t/s parameters though.
-I'm sure user error is a part of this too. My equipment can be very accurate, but that doesn't mean I am. I try my best, and built many measuring jigs, but I don't have the ability to insure that there is no interference, moon phase problems, or whatever else might throw things off beyond my control or knowledge.
Driver variation will certainly swamp out crossover component variations, especially batch run to batch run. Thats why I don't worry too much about the crossover component tollerances. But I will say that it is well worth it to have me assemble the kits if just to have me measure them when they are done. This is, in the end, the only test that matters. I do a lot of these tests and I will detect a problem that is audible - and this DOES occur. In the last set of three that I made one was out of spec. I thought that it was the crossover, but it turned out that it was the fit of the compression driver to the waveguide was not seated well enough and apparantly leaked. Even though I've built a lot of these speakers this kind of thing still happens.
Building kits and DIY on systems can be fun and it can save some money, but if you cannot test the final product then there is a huge risk involved. And you certainly can't trust listening to them to determine if some small error has occured (I don't care what you think you can hear). If the two speakers sound different from one another that can be a clue, but that can also be the room. And then "what is it thats wrong?"
DIY is not as simple as it looks.
Building kits and DIY on systems can be fun and it can save some money, but if you cannot test the final product then there is a huge risk involved. And you certainly can't trust listening to them to determine if some small error has occured (I don't care what you think you can hear). If the two speakers sound different from one another that can be a clue, but that can also be the room. And then "what is it thats wrong?"
DIY is not as simple as it looks.
Vectoring off from the recent (very interesting) posts...
When I was experimenting with sealers, I also did various glues (I have a box with about 20 MDF blocks, all with varying combinations of glue, sealer, paint, handling methods etc - very dull to do).
I would say the best results I got were consistent with blocks using Polymite (was Extramite/Cascamite). This is a powdered urea/formaldehyde resin glue, and smells as nice as it sounds. I spoke to a guy who represented the UK/European MDF makers, and was told this is similar to the stuff used to make the MDF. It sets brittle and hard, so doesn't move.
I also found that West System epoxy would glue well, though again I find it very messy to use.
IRC John claimed success with normal PVA and lots of pressure, and I must admit I have had these work. I think that getting the moisture content right (as you've noted yourself) is the crucial thing.
I was considering chucking MDF parts in a small room with a dehumidifier to dry them before painting. I've not tried it, and it occurred to me you probably want the moisture content of the MDF stable for the room it's going in. Maybe too dry would be as bad as too damp.
gedlee said:
When I was experimenting with sealers, I also did various glues (I have a box with about 20 MDF blocks, all with varying combinations of glue, sealer, paint, handling methods etc - very dull to do).
I would say the best results I got were consistent with blocks using Polymite (was Extramite/Cascamite). This is a powdered urea/formaldehyde resin glue, and smells as nice as it sounds. I spoke to a guy who represented the UK/European MDF makers, and was told this is similar to the stuff used to make the MDF. It sets brittle and hard, so doesn't move.
I also found that West System epoxy would glue well, though again I find it very messy to use.
IRC John claimed success with normal PVA and lots of pressure, and I must admit I have had these work. I think that getting the moisture content right (as you've noted yourself) is the crucial thing.
I was considering chucking MDF parts in a small room with a dehumidifier to dry them before painting. I've not tried it, and it occurred to me you probably want the moisture content of the MDF stable for the room it's going in. Maybe too dry would be as bad as too damp.