New DIY Mid High (90deg) - AKA PM90

I am still trying to get this D.I.Y. to sound sweet, and starting point being some good measurements, would anyone comment on these?

They are BMS 4594HE 16 ohm passive filter and RCF MB12N351 4 ohms on Powersoft K series amps.

When comparing to BMS FR graph I wonder if I am doing something wrong or ?

The measurement was taken at 60cm on axis appr. 1 meter from ground, indoor. I have made many outdoor measurements and not much changes in the results for the HF part, which is my main concern.

I think the dip in the 4.8Khz area is a bit strange and so is the sudden level rise at 9Khz.

Thanks in advance.
MH90_60cm.jpg

I
 
Andy Turner said:
I am still trying to get this D.I.Y. to sound sweet, and starting point being some good measurements, would anyone comment on these?

They are BMS 4594HE 16 ohm passive filter and RCF MB12N351 4 ohms on Powersoft K series amps.

When comparing to BMS FR graph I wonder if I am doing something wrong or ?

The measurement was taken at 60cm on axis appr. 1 meter from ground, indoor. I have made many outdoor measurements and not much changes in the results for the HF part, which is my main concern.

I think the dip in the 4.8Khz area is a bit strange and so is the sudden level rise at 9Khz.

Thanks in advance.


I
Click to expand...

I think that's OK - have a look at my measurement - the pink trace. I have some EQ suggestions in the double 10 thread which uses the 4594 + passive crossover with the RCF HF950 horn....

- BMS 4594 with crossover on RCF HF950 (not HE version)
- Crossover – Linear phase 48 dB LR simulation @ 630Hz
- If you don’t have a Lake I would use 24 dB LR @ about 675Hz

654Hz
Q 0.96
+4.1dB

1K343
Q 0.69
-3.6dB

3K22
Q 0.33
-2.7

4K6
Q 0.68
+3.3 dB

8K49
Q .82
+2.1 dB

19K06
Q 3.1
+5.3

AP 1K895 Q 3.4 2nd order
AP 12K35 Q 0.34 2nd order

On top of that there 12 more raised cosine filters - 1 to 2 dB each for minor adjustments. The most noticeable would have been a dip - 2 dB .... 5K to 8K ish.

 

Attachments

  • spl comparison HF.jpg
    spl comparison HF.jpg
    1.2 MB · Views: 36
Last edited:
  • Like
Reactions: Klaus Zimmermann
Peter Morris:
Thank you very much.
It is very helpful to get some kind of consensus that your baseline is not all wrong.

I have learned that there are many ways to get a flat FR and that they all sound different. Different strategies depending on whether you go for primarily lowering peaks, or rising dips, - how much gain, - ignoring the very sharp variations or going for +-1dB. etc.

In my current EQ setting I have a +7.1dB @ 4.861Khz Q=3.6(0.4oct.) to make up for the dip on my measurement?? Some recordings can sound a bit harsh, but without the EQ, the sound is not 'right'.

- Further I have been having a hard time making the 4-12Khz area sound smooth. The hard bit for me has been determining where the problem is. I learned that even small variations in the very high frequency area can cause it. The solution for now, has been to focus on getting the HF measurement smooth and with a small rise from 5Khz. While chasing the smooth VHF - attention to mic placement in front of the horn is very important.

Jason Joseph:
Thank you very much for sharing. Resolutions on our graph is 20dB vs. 5dB on mine. Could you adjust and post again?
If you can find the time to share your DSP settings, that would be interesting also.

Discussing measurement approaches:
- Another thing is measurement-distance and off-axis measurements. What is the 'correct' distance when measuring a horn, and even a coaxial CD?
I heard someone talking about coaxial compression drivers beaming at short distances, - which can give some strange VHF artifacts.

I tried off-axis-average measurements to pursue a 'power response' like approach. Interesting work which can explain some of the mismatch between the on axis FR and your perception of the sound in a room.

Here are my current DSP settings which reflects the FR in post 501:

DSP, MH90 IIR version 10:


HF: BMS 4594nd HE 16/16 Ohm

Passive BMS crossover

Powersoft K6 DSP amp

Delay: 1.8ms

Gain: 32db

TypeFREQQdB
Highpass LR24702
Peaking7603+4.3
Peaking8591.6-1.5
Peaking1.1224.1-1.5
Peaking2.8704.3-2.7
Peaking3.2442.9-0.9
Peaking4.0886.9-2.0
Peaking4.8613.6+7.1
High Shelving1.25 oct.7.772-2
Peaking8.4761.7+2.1
Peaking9.86314+2.7


MID:

2 x RCF MB12N351 8 Ohm.

Powersoft K8 DSP amp

Delay: 0ms

Gain: 32db
TypeFREQQdB
Highpass LR24100
Lowpass LR24702
Peaking1161.1+6
Peaking3643.8-7




 
Andy, by the look of that plot i think you are doing something wrong with the polarity at the BMS coax. There is a dip at the lower cut off (650hz or so) which should not be there. Could you please try reversing the polarity on the MF secion on the BMS and see what will that do? Only the MF.
 
When i get home I will see if I can change the resolution and resend... I think if you make it too fine then you will be chasing your tail attempting to make it too perfect. A straight line doesn't always make for the best sound.. Also we had to do a -2.5 on the outputs of my DBX Venue 360 on BMS. Like Marjan said check your wiring again and make sure you have everything connected positive to positive negative to negative...

Ill post some basic settings later..
 
Marjan:
Thanks. I have tried reversing the polarity on the VHF section only on the coax, and the dip between the HF and VHF got much worse(red)
.

DIYMH90_indoor_VHF_inverted.jpg

Further, - the dip in the 650 Hz range is a bit strange also. When I use the same crossover settings on a DBX drive rack PA+ and a 4 channel powersoft amp, I have no dip at the crossover section. I am using 1.5ms delay on the BMS with the DBX and 1.8ms to get the best possible results on the powersoft dsp.:?: There must be something different with the way they translate the filters.

Jason:
Input to DSP settings would be very much appreciated :)~:)~:smile::)~:)~:smile::)~:)~:smile:. Do you also have the HE version of the BMS?

You're saying that you have -2.5db on the coax, and is your mid section also 2 x 8 Ohm in parallel?

I get your point about chasing the straight line FR. I tried it and as you I found that it didn't sound good.

In the FR that I have demonstrated in my latest posts I have aimed for smooth HF/VHF with a small rise. The smooth HF/VHF - with no sharp dips or peaks I found was very important. If something goes wrong up there, - Hi-hats, metal etc. start sounding sizzly, shrill.
The dip in the 4.8Khz is also important to make up for.

MH90_IIR_version10.jpg
 

Attachments

  • MH90_IIR_version10.jpg
    MH90_IIR_version10.jpg
    190.8 KB · Views: 21
DSP, DRIVERACK VENUE 360

HF: BMS 4594nd HE 16/16 Ohm
Passive BMS crossover

Powersoft M50q amp

Delay: 1.96ms
Gain: -2.5db at crossover point
TypeFREQQdB
Highpass LR24473
Lowpass LR618k


MID: 2 x B&C 12NDL76 16 Ohm.

Camco Vortex 6 amp

Delay: 0ms
Gain: +2.0db at crossover point
TypeFREQQdB
Highpass BW1895
Lowpass LR24500

Of course i forgot when i got home. So ill send a reminder to myself for the actual PEQ settings...
 
  • Like
Reactions: Andy Turner
Marjan Milosevic said:
Andy, differences between dsp units are normal. But that still does not explain that dip between hf and
vhf. I am extensively using 4592 which is not that different from 4594 and i get to this response with very little eq
Click to expand...

Hi Marjan,

The 4594 is not as flat as the 4592 ... the smaller throat has an advantage for pattern control especially with the 90 degree box, but its not as "flat" as the 4592. The 4594 was selected as there were suitable 60 and 90 degree size horns available that would work with this project.

BTW that's a great looking response from your box :)
 

Attachments

  • 4594 vs 4592.jpg
    4594 vs 4592.jpg
    157.9 KB · Views: 40
Andy Turner said:
Marjan:

I get your point about chasing the straight line FR. I tried it and as you I found that it didn't sound good.

In the FR that I have demonstrated in my latest posts I have aimed for smooth HF/VHF with a small rise. The smooth HF/VHF - with no sharp dips or peaks I found was very important. If something goes wrong up there, - Hi-hats, metal etc. start sounding sizzly, shrill.
The dip in the 4.8Khz is also important to make up for.

Click to expand...

FWIW, my problem when I started this project was determining what the target response curve should be. Many people argued it should be flat but to me that never sounded nice. The conclusion I came to was that flat is probably correct measured at 1m in an anechoic chamber ... but how we are measuring it, it sounds better with some HF roll off ...
 
Last edited:
Peter Morris said:
FWIW, my problem when I started this project was determining what the target response curve should be. Many people argued it should be flat but to me that never sounded nice. The conclusion I came to was that flat is probably correct measured at 1m in an anechoic chamber ... but how we are measuring it, it sounds better with some HF roll off ...
Click to expand...

Evan and I both also found this to be true.
 
It all depends how the box is used. A box on a stick close to your ears will probably need some roll-off because the 7-12khz regin will be a bit painful. But hang the same box on a truss, you might find the VHF is lacking.
My personal liking is pretty much inline with Andy's graph. A bit lowered 2-5khz region, and a small boost at 12khz.
 
Peter Morris said:
Hi Marjan,

The 4594 is not as flat as the 4592 ... the smaller throat has an advantage for pattern control especially with the 90 degree box, but its not as "flat" as the 4592. The 4594 was selected as there were suitable 60 and 90 degree size horns available that would work with this project.

BTW that's a great looking response from your box :)
Click to expand...

Peter, BMS graphs are a bit misguiding. 4592 and 4594 graphs are not made on same horns (which is normal because they are different throat) so it is hard to compare or use those graph because none of us is actually using the BMS horns those graphs are made with.

The main difference is that 4592 plot is made with a 90x55 degree horn, and 4594 with a 60x40 horn. No idea whay they did that, but for sure it is barely usable if you want to make any conclusion by simply comparing the plots.

 
Andy Turner said:
So how much HF roll-off have you settled for?
And can someone explain why this is the case?

My experience so far, has been that unless I target for a gentle raise from 5 kHz to 10 kHz by 2-3dB, it sounds like I get the sensation of way too much VHF 9-12 kHz.
Click to expand...

We didn't really settle on a set amount of roll off. I just remember that we both agreed that out of the box, the flat tunings sucked from 9k-12k, and let our ears inform our tuning choices.
 
+2 to HF/VHF roll-off needed.

But I like to still tune speaker processing to flat. (subs included)
And then use sets of shelving filters at the mixer, for highs (and lows), .........to dial in tone preference, "house curves", etc.
 
Top Bottom