Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
71.67.188.239
In Reply to: RE: Are my ears really that sensitive? posted by Davy on September 13, 2014 at 10:40:31
The option of raising the mid LP inductor from .4 to .7 mH is interesting, but may end up cutting you off too low- that is nearly an octave's worth - if you have something like 0.1-0.2 mH to add on to the 0.4 that might not be too drastic, but if you like what the simulation shows, go ahead - you already have the parts. I would not worry about the 18ga wire for now, as you say, it is good enough for experimentation.
As for reproducing the MG20 XO in the IIIa, I think that you can start with the top filters and change the 18/12 to 18/6 later on. It is just a wire snip and possibly a cap substitution. In my setup doing anything other than 1st order on the midrange is so disuptive to SQ that I just don't do it at all. The initial experimentation with the free midrange was incredibly instructive as to how good things can sound. I only added the filters because the amp and drivers needed protection from loud bass and in order to do away with the head in a vise problem due to low dispersion of the midrange beyond 6-8 khz.
Follow Ups:
Ok I tried several things on the MGIIIas last night. I think I am slowly dialing things in.With 60uF worth of external HP caps (first order filter) in the 18 6 (6 12) config I flipped the tweeter polarity as you suggested, and this did make the sound more to my liking, but it was a bit too sucked out in the presence region and lower treble.
So instead (still running 18 6 6 12 I put the polarity back (to + - + for the drivers) and increased the mid LP inductor from 0.4 to 0.72 mH. Obviously this spaces the mid/tweeter crossover more and brings down the XO point. This was better than flipping the tweeter polarity, but again things were too sucked out in the presence region. I think you are right, with a less drastic change, - maybe to 0.5 or 0.6 mH things might work well. Another additional thing which might help is to bring the tweeter XO down slightly by increasing the cap (and possibly reducing the tweeter inductor).
Anyway, the next move (with 0.72 mH mid inductor still installed) was to reconnect the parallel midrange 10 uF cap, going to 18 6 12 12 (MG20/20.1 type spec). This change the sound a lot and it did not sound as good as 18 6 6 12 in my opinion as the sound was much thinner and anaemic and the presence region was becoming more prominent. The midrange output had gone up significantly I think. Quite a bit of work would probably be needed to get 18 6 12 12 sounding better, - maybe increasing the Mid LP inductor more to around 1mH or higher (as in the 20.1). I did try decreasing the tweeter cap to 10 uF and that improved things slightly, but not enough. I think the midrange output is higher in 18 6 12 12 than 18 6 6 12.
So I think I'll go back to 18 6 6 12 and try to get that to work. I'll start unwinding the 0.72 mH inductors and go down slowly to around 0.5 to 0.6 mH. Might also try lowering the tweeter, and maybe try increasing the external HP from 60 uF to 70 uF or higher.
This is the electrical plot I am getting from the spreadsheet with 18 6 6 12, 70uF external HP, and 0.55 mH mid LP and the tweeter adjusted to 15uF and 0.4 mH parallel inductor (reduced from 0.7 mH). I can use the previously used 0.4 mH midrange inductor on the tweeter instead. The ribbon is still safe as its not going as low as it does in the 3.3, 3.5 or 3.6
The plot looks good, but one issue is that I have noticed is that when I change the tweeter crossover values on the spreadsheet, the midrange curve also changes. I guess this must be wrong. I am trying to get to grips with LSPCAD demo version instead (which is limited to 2 drivers, but still usable).
Edits: 09/16/14
Been running some sims and things don't look too great increasing the mid LP inductor to 0.7 mH whilst running the stock external 18 12 setup with (with 3.5 mH inductor and 60uF caps on the external HP)
18 6 external with 60 uF HP caps looks better.
I think the problem I had with thinness increasing the external HP caps to from 35uF to 60uF with the first order external HP yesterday is that it increases the mid output meaning bass is less prominent. With a resistor on the mid it might work much better. The FR is quite flat with that particular config. When AndyR altered his MGIIa crossover he also had to use resistors so perhaps this is something to bear in mind. Increasing the caps higher to 75 or 90 looks even better but I'm not sure the mid would be happy running that low in first order.
If I used a resistor on the mid what wattage should I use? Several 10 watts in parallel or a much higher rating?
My other idea of adding an 2nd mid LP inductor to make it 3rd order won't work unless the present inductor is increased substantially in value, to 8mH or higher, whilst also increasing the parallel cap significantly to around 50uF. This is basically the 3.3 or 3.5 internal crossover setup. That's quite a bit of work.
It is possible that the 60uF swap was making the lower mids (250-500 hz) too prominent and creating a dip from destructive interference lower down. If it is just a matter of mid prominence then you might get rid of the problem with a smallish resistor.
The resistor can be made of a few parallel 5 or 10 watters. You probably don't need any massive resistors like 50 watt rated ones that often have heat sinks.
I would not go to 3rd order in the mid or treble. It starts sounding rough.
Replicating the XO you built for the 3.3 or 3.5 is not necessarily a bad idea since you liked it all things considered. Your lower levels of playback are friendlier to the 1st order XO than my need to shatter glass with operatic sopranos.
Thanks Satie. I think even with the new mid LP inductor value I'd still need to lower the midrange output level with a resistor. I see Andy used 30w resistors so that seems sensible.Like I said I was not happy with the 18 6 6 6 crossover I put in the 3.3rs after not hearing it for a long time. Also way too much peakiness in the prescence region. It would need adjusting, maybe back to stock MGIII 18 6 6 12 (which actually worked well in them before, unlike in the iiia where the mid is too sucked out), or I might actually go back to the stock 18 12 18 12 crossover in them when I get the ribbon repaired. You are right things do start to sound a bit rough with third order on the mid LP but I would not mind trying it again anyway as it was designed for the speaker. Does not help that the cap Magnepan use in parallel on the mid LP is an electrolytic...
Wow you do listen loud. Front row of concert stuff! I remember that from before! Be careful with your hearing though!
Another option ( though it's harder to do as you have change all the values) are Bobwire's two 18 12 12 12 crossovers which are very flat. Much flatter than stock MG IIIa . I would probably want to introduce a slight dip in the prescence region if possible but I might play around with them in simulation to see. AndyR ran them in Lspcad for me a few years back and thought they looked extremely good. They might need the relative driver levels adjusting too.
Edits: 09/14/14
You guys conversation makes my head hurt. :) This kind of tedious, iterative approach to tweaking is MUCH easier handled by a DSP-crossover and multiple amplifiers. You can experiment endlessly and ultimately come to a solution (if you're careful) that can be recreated with passive components.
And always remember, just looking at simulations of the electrical curves are limited in value when determining suckouts, peaks, phase responses, etc, etc. You MUST factor in the acoustic responses of the drivers and be skilled with an acoustic measuring acumen or it can be pointless.
Cheers,
Dave.
Thanks for the input. I do kind of personally like plodding along, adjusting stuff slowly, even if its no the best way to do it. Guess it depends on what floats your boat. I am also however quite interested in your suggestions. Which DSP crossover(s) do you recommend?, seem to be quite a few out there.
From my (reasonably limited) experience I do think you can get some idea of the sound from the electrical response, and AndyR also agreed with me on this. I acknowledge it does not account for many other factors though (as you say), but obvious things like the big presence dip in the electrical response of the stock 3.3R crossover, or the presence peak on the MGIIIa or Stock Tymapni IVa are certainly audible to me.
My favorite currently is the miniDSP 4x10Hd. An extremely versatile unit that offers the ability to go 4-way, control your system volume, provides digital inputs, and a host of other features.
However, Satie would reject this unit out of hand. :) There's no satisfying some people. :)
Dave.
We have competing goals in mind. These have to be weighed against each other and I (not so Davy) can't stand mediocre DACs such as those in DSP devices. Using them makes my subjective evaluation useless as I focus on the irritating sound they produce - all I can do is say "I don't want this". I can easily tolerate wow, noise, hum, non linear FR and peaky mids, bloated bass, or overcooked highs, I can take lots of phase issues, I shut down with bad digital.
Though I suggested all along that Davy use REW to do the characterization of the drivers (I did it on spreadsheets a decade ago with manual measurements for my drivers) and work things out on the computer, but he did not want to redesign the speakers from the drivers up, but to tweak them. I knew that is what I would have to do from the get go, Davy did not and still doesn't intend to.
Davy is very sensitive to coherence and has gone back from multiamping to single amping. Your solution is a design tool instead of a tweaking/fine tuning of the FR and phase issues. I don't think that within the limited scope of Davy's goals he would benefit from creating a DSP and multiamp setup he can't listen to. Not to speak of the learning curve in using the measurement and dsp software which would not be in place for his final product.
Neither Davy nor I intend to be professionals in this field but to tweak what we have. I kept an adjustable PLLXO/hybrid XO so that I can change things on the fly for recordings and experimentation with positioning (Limage and Rooze setups that do better with tweaking of the XO to complement them because of timing issues with something so large as the T IV).
I had done the mod TIV/Neo8 with Marchand's XO and had excellent results. I liked what I heard with the naked Neo8 array and attempted to use it that way for a couple of years, but the practical limitations drove me to rework the XO again with the constraint of having a 1st order XO for the midrange.
Davy is trying to iron out a peakiness in the top of the midrange that is the flaw in an otherwise fine speaker that he likes. An all out redesign assault on the speaker is not what he wants to do.
Hi Satie,
Sorry didn't see your post until now.
I hear what you say about the DSPs if they do degrade the sound somewhat. I know the older Behringer DXC2946 didgital crossover is very dated compared to modern DSP crossovers but I don't think I could have used that on the Maggie mid and tweeter to make subjective evaluations.
You are right, I am just tweaking what I have right now, and I freely admit that quite a bit of it is a shot in the dark, but I do quite enjoy that and eventually getting to something that works.
I do like single amp drive (being sensitive to coherence as you rightly say). Part of the trouble trying to get the drivers to blend well with biamping and triamping is having to do lots of experimentation with different power amps, crossover setups etc. That can take lot of time to get it to work well. For sound quality above anything else, PLLXO on the mid and treble is better than any active crossover in my view (and also better than single amp drive at speaker level), though it can be harder to blend PLLXOs with active bass due to delay issues etc. I might eventually go back to biamping or triamping in the future and really go for it and try to get something coherent, but for now I'm just tweaking the basic sound I've got with these MGIIIas in single amp drive. Ultimately I'm not sure if you'd ever quite the same degree of coherence of single amp speaker level passive drive with any multi amp setup, but with careful tweaking and matching I am sure you could get very close and also get the big advantages of biamping/triamping such as enhanced clarity and less distortion. I could likely also live with the latter.
Davey is likely right in that it could be easier and quicker to iron out the issue I am having with now, but only if I had the system set up already and knew what I was doing (and know how to transpose it well to passive components). The expensive of buying the DSP and more power amps along with the learning curve and time needed don't really make it viable for my situation right now (especially with single amp drive), but thanks for the advice Davey. It's certainly something I will bear in mind to try out in the future, especially if I do go back to a multiamp setup.
I will try flipping the tweeter polarity with the 18 6 (6 12) setup as per your suggestion now to see how that fares. Then think about the mid LP inductor change.
Yeah, I'm well aware of your sensitive ears and point-of-view on this, but if you can't listen past the sound of some digital circuitry to make relative evaluations it REALLY makes this sort of evaluation tedious and difficult. That's all I'm saying.
I can appreciate the learning curve of measurement, DSP-software, etc, etc, but eventually you get to a point with this hobby that you're forced to learn/utilize this stuff. You two guys are at (beyond) that point and seat-of-the-pantsing experimentation is a real double-edged sword.
If you're going to be a master auto technician, you need a complete set of tools......you can't do everything with a crescent wrench. :)
Cheers,
Dave.
This is true. We also need to know what kind of frequency response is prefered. Most of the DSP-systems available seems to think it is a straight line from 20-20000 Hz at the listening position we are aming for. It is not!
Thanks for the advice. I've asked Roger to kindly run a simulation to check how 0.74 mH on the mid LP looks. I am hoping this is going to be an easy change to sort out the peakiness at 1-2 KHz. That inductor is 0.7 mH in the original mg20 (but with a 7uF parallel cap) and goes up to 1.12 mH in the MG20.1 (with a 12 uF parallel cap) though I guess the drivers are a bit different. The 20.1 (in particular) simulation does have quite a broad suckout centered around the prescence region. I can always unwind some inductor if 0.74 mH is too much and measure with my meter.
You are right it's v easy to change the external HP from 2nd to 1st order so I can do that later. The 18 12 bass to lower mid transition is quite good in the stock MGIIa XO. Better than it sounded with the stock bass HP in 18 6 so I think I'll try stock external first with this internal inductor change and it from there but I will definitely try both and see which one blends better.
The Neo 8 array must have great ability to be able to be run full range, but yes I guess it's sensible to protect it from bass and reduce the beaming problem in the upper mids/lower treble. I guess when it was run full range the lack of phase etc made things sound really good?
I think the broad suckout may not play out that badly in the presence region since I suspect the sensitivity of the mid is possibly 1 db higher than the bass. So long as the depression on the gann chart is not deep it may not sound like it looks.
Re the Neo8, it is phenomenal in the range it covers and yes, the lack of phase is really a great plus. But doing things that way just leaves me without the ability to play as loud as I need.
Post a Followup:
FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
This post is made possible by the generous support of people like you and our sponsors: