Welcome! Need support, you got it. Or share your ideas and experiences.
Return to Planar Speaker Asylum
24.125.205.20
In Reply to: RE: MMG MiniDSP and bi-amp Questions posted by acroy on February 23, 2012 at 06:23:49
You didn't mention whether you are using the "newer" MMG's or the original model. For the newer version the filters are 1st order - LP at 290 and HP at 3300. For the original the LP filter is 2nd order at 680 and the HP is 1st order at16602550.
If the phase of the tweeter and woofer is incorrect you will hear a drop in the output around 800-1000 hz.
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Edits: 02/24/12 02/24/12Follow Ups:
Hey Neo,
If the phase of the tweeter and woofer is incorrect you will hear a drop in the output around 800-1000 hz.
I had assumed that the 1st order slopes have the tweeter and woofer in phase. When I was messing with the allocator software to mimic the stock slopes I never could get the right graph. Davey showed me that even in this config the tweeter and woofer are wired out of phase so any crossover mod should keep that same orientation.
So "incorrect" in this case is in phase with each other???
For instance, here is the curve with the drivers out of phase:
and here it is with them in phase:
I guess I thought the drivers were supposed to be in phase on 1st order slopes. Where can I read more about that stuff??
Afterwards we discovered faith; it's all you need
If you have a symmetric first order crossover than the phase difference of the LP filter and and HP filter is a constant 90 degrees at all frequencies. It doesn't matter if the polarity is reversed since the difference is still 90 degrees.
However this does hold up with an asymmetric crossover and in that case changing the polarity may very well result in a different phase relation and a "dip".
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Neo mentions: "If you have a symmetric first order crossover than the phase difference of the LP filter and and HP filter is a constant 90 degrees at all frequencies. It doesn't matter if the polarity is reversed since the difference is still 90 degrees."
Neo, that's precisely what I thought at first. It didn't work quite that way in practice but I'd have to look up the charts to refresh my mind. (It's Friday at work and my mind is having a major dip at the xover point : - ))
The phase relations are determined by the math. Here's two graphs that show the difference between a symmetric 1st order crossing at 800 hz and an asymmetric 1st order also crossing at 800 hz.
With the symmetric the phase difference is a constant 90 and the output of normal polarity and inverted polarity are superimposed on each other.
With the asymmetric, the phase difference varies with frequency and the normal polarity creates a dip while the inverted polarity has a much shallower dip.
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Neo, I was wondering if we were talking about 2 different things. By asymmetric, I meant the original MMG xover with the mix of 1st/2nd order slopes. In this case, the current design would be symmetric for having 1st/1st.Perhaps the symmetry you refer to is with regard to another set of parameters? I believe you mean that the points for each driver are at the same frequency. This latter case is when one should expect to see no difference? I guess this is it. Which would explain why, by being asymmetric in this sense, I didn't see what I expected from a current design. Of course, MMGs don't want to use the same points because, hey, that would be boring and conventional. LOL!
So, both MMG xover designs share this kind of approach to frequency point asymmetry. In addition, the original design is further asymmetric in the slopes used.
All of which helps us have so much fun. : - ))
Edits: 02/24/12
It's my understanding that a symmetric crossover is like the typical L-R in that the fc of the LP and HP are the same as are the orders of the two filters. Anything else is asymmetric. Magnepan almost exclusivly uses asymmetric crossovers with the exception of the SMG(a)(series 1st order) and IIa (parallel 3rd order).
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Edits: 02/25/12
Hint. You should think polar response here. Horizontal polar response, to be more specific.
Cheers,
Dave.
I have the 'newer' crossover with 1st order slopes.
You state LP 290 and HP 3300
Maybe I am being dense - but where is the sound coming from in that huge hole from 290 to 3300? Is the rolloff so gentle, and the panel response curves so, um, 'rising-rate' (for lack of a better term) that the 290-3300hz range is covered even with the 1st order slopes in the way?
Many thanks as always. I am sure these questions ahve been answered a million times. I boy-scount-honor swear I'll help educate the next batch of newbies once I'm educated myself ;)
Ideally (perhaps) the filter's should look like step-functions or "cliffs" with sudden drop off at the cutoff frequency, but this is not possible.
Here's a diagram of what's going on,electrically. As Davey points out the actual acoustic response is different and that -6db dip @ 1000 Hz may not exist when you listen to the speakers.
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Edits: 02/24/12
excellent- thanks much. Graph helps a lot. What tool is that - a freebie simulator would be fun to play with. I understand the basics, i.e.crossover orders, but would have to pencil-whip the numbers.
I use a spreadsheet that I developed in Open Office. Open Office does everything that MS Office does and is free.
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
See my post below regarding filter frequencies.
There would be "huge hole" IF the transducers responses were flat and wide ranging. But they're really not. This is definitely not a "textbook" crossover design. It's the acoustic responses that are ultimately important. You adjust the electrical responses to achieve whatever the target acoustic response is......regardless of how funny the ultimate electrical responses look.
Cheers,
Dave.
Yes, also seen at my end, Davey.
When one measures the actual non-filtered output of the drivers, more so the mid/bass, it becomes very clear why Magnepan chose to do things in not-quite conventional ways.
Huh? Neo, I must have been smoking some very bad crap for far too long but could you recheck the LP spec for the original MMG?
Mine is the old design and came with the same 2.2mH coil as the newer ones. The only difference was the 2nd order caps (25uF). So, the x point would nominally be the same but with a faster slope.
An odd thing about the older tweeter is probably not widely known. The old tweeters in mine are 2.6 ohms (measured). The new tweeter is "probably" 3.5 ohms (extrapolated, not measured). With the caps that came with them, it would seem that:
- Old MMGs tweeter with 24uf@2.6ohm = 2550hz
- New MMGs tweeter with 12uF@3.5ohm = 3790hz
That 2.6ohm in mine caused me quite some anxiety when I tried Tazmon's capless HP mod and adjusted it for "flat" rsponse, years ago.
For a first order LP filter fc= R/(2*pi*L) which works out to 290 hz for a 4 ohm load with a 2.2 mH inductor. For a 2nd order filter the fc= 1/[2*pi*sqrt(L*C)], so for the original MMG this is 1/[(2*pi*sqrt(2.2*24*10^-9)] or 680 Hz.
I was unaware that the original MMG tweeter had an impedance of 2.6 ohm, in which case the fc is 2550 Hz as you stated. I will need to update my spreadsheet.
BTW, Tasmon reported the woofer inductor as 0.75 mH while the official schematic (both on the Tweaks page) is 2.2. I never understood why one would want to go to a capless crossover as the inductor is the least ideal of the passive devices -relatively expensive, poor tolerances, very real capacitance and resistance, limited values, etc.
On the violin: "Heaven reward the man who first hit on the idea of sawing the innards of a cat with the tail of horse."
Edits: 02/24/12
He either mis-measured the inductor at 0.75mH or made a decision to change (lower) the value for his "cap-less" crossover.Everything he did is highly suspect, IMHO. It really should probably be removed from the tweaks page.....or at least explained. Many folks have raised questions regarding this "mod."
I don't think I'd analyze these MMG crossover in terms of crossover frequency. Look at them as two separate filter sections and model/analyze based on a defined -3db (or -6db) point. It's less confusing that way.
Cheers,
Dave.
Edits: 02/24/12
100% in agreement on both counts.The Tazmon tweak worked for me because I had the feedback from mic measurements. It didn't take long for me to see that the low pass was not right. I used the 2.2mH instead and things began to fall in place. Pity that the HP has to waste so much energy as heat. Shy of a line-level xover, it sounds delightfully clean and extended. We should insert clarifications for this mod.
Later, as I went on to line-level filtering, once again the feedback from the measurements showed what you just said with regard to the MMG modeling.
Edits: 02/24/12
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: