New DIY Mid High (90deg) - AKA PM90

Re: New DIY Mid High

Peter Morris said:
I tried a new (rough) crossover setting use 24dB octave slope at 650 Hz to see if this idea would work ... looks quit good now :)~:-)~:smile:

The plots are - 0, 10, 20, 30 & 40 degrees below the axis.
Click to expand...

Hi Peter,

its getting better and better.
Are you familiar with the overlapping FIR filters technique, which was used with GAE Director and implemented in DSC28,
if not a little bit of information is on the Four Audio HP like this:
http://www.four-audio.com/data/ALMA2009/ALMA2009_FourAudio.pdf


Uwe
 
Re: New DIY Mid High

Uwe Riemer said:
Hi Peter,

its getting better and better.
Are you familiar with the overlapping FIR filters technique, which was used with GAE Director and implemented in DSC28,
if not a little bit of information is on the Four Audio HP like this:
http://www.four-audio.com/data/ALMA2009/ALMA2009_FourAudio.pdf


Uwe
Click to expand...

Hi Uwe,

I was aware of the Four Audio stuff … I have been wishing I could to do all of those tricks for years.

When I did the FIR settings for Turbossound's Flex array the problem was controlling horizontal directivity through the HF and VHF crossover region. What you have described is exactly what I wish I could have done for the Turbo's; and this box. Unfortunately the Lake’s interface does not allow you to do this. The Flex took me a week to solve the directivity problems, but I eventually got there. The result of using FIR filters and the new crossover settings was a very noticeable improvement in sound quality.

http://www.turbosound.com/docs2/latest_news/EFEFZuEykpMTVAGkNQ.shtml

What’s happens off axis is really about the energy in the reverberant field.

With respect to this design, I’m not sure what happens when you use a dipole to control directivity. I suspect there is more energy in the reverberant field than the initial pattern would indicate – just a guess.

Here is a plot of Turbosound’s Flex that has its two 10's in a dipole, note the two lumps of energy around 500Hz. I suspect this is what is happening with my box in the vertical plan and that the energy in the reverberant field is better balanced than my frequency response plots would initially suggest.

BTW – thanks again for your input.
 

Attachments

  • flex pattern.jpg
    flex pattern.jpg
    121.6 KB · Views: 107
Last edited:
Re: New DIY Mid High

I think I have finally managed to get some really excellent directivity behaviour ... just needed a few more tricks with the crossover.

I can make that dip at 700HZ on the vertical plot a little bit better, but I'm concerned about load on the HF driver.

The FIR time is now only 2.5ms. The impulse response is still almost perfect and the phase response is almost the same above 300Hz. Because I'm now using an IIR crossover for the subs there is another 180 degrees of phase shift in the low frequency end compared to the first plots I posted.

Top plot - vertical

Bottom - horizontal
 

Attachments

  • DBL12 directity.jpg
    DBL12 directity.jpg
    181.7 KB · Views: 165
Last edited:
Re: New DIY Mid High

Peter Morris said:
The FIR time is now only 2.5ms. The impulse response is still almost perfect and the phase response is almost the same above 300Hz. Because I'm now using an IIR crossover for the subs there is another 180 degrees of phase shift in the low frequency end compared to the first plots I posted.
Click to expand...

Hey Peter,

How are you implementing the FIR filtering and the DSP tuning in general?

Ciao
Simon
 
Re: New DIY Mid High

Simon Coffin said:
Hey Peter,

How are you implementing the FIR filtering and the DSP tuning in general?

Ciao
Simon
Click to expand...

The first plots were all “brick wall” and used 12.5ms of FIR time.

The last plots used 2.5 ms FIR time; they were as follows:-

- HF 6300Hz - brick wall
- MID 700Hz - brick wall. Only about 20dB per octave with that FIR time. You could use a 48 dB LR linear phase simulation but the directivity plots are not as good.
- LOW – 100Hz IIR LR 24 dB per octave.

The LOW section of the BMS4594 uses 4 all pass filters to flatten the phase and some out of band filters. The 12 inch MIDS uses four filters, two out of band. The VHF driver is polarity inverted.
While all bands have some in- band EQ, apart from a reasonable large cut at 400Hz as predicted by Hornresp simulation not much EQ is used to get a flat response.

There are also a bunch of small raised cosine filters applied to the overall response to make the frequency response look like it was drawn with a ruler.

The first, all FIR settings required two Lake LM26s, one for each channel. The most recent settings allows the MID – Hi box to run with one LM26 and any other DSP for the subs. An all pass filter is used on the subs so their phase response is identical to the low mid through the crossover region,but it’s not critical.

Having described this complex crossover, the BMS driver has a very flat frequency response without EQ. BMS can supply a passive crossover for the driver so the box could be run two ways. The double 12” mid only requires two or three filters to produce a good response, so most DSP’s could be used to achieve a good response.
 
Last edited:
Re: New DIY Mid High

Hi Peter,

Nice work and excellent Lake tuning!

Curious about this HE version of the BMS driver, what is it? I have few DIY projects going/gone with the basic 4594ND and it would be interesting to get some other versions of that driver.

-Riku
 
Re: New DIY Mid High

It has a different phase plug for a smoother VHF response and shorting rings on the magnet structure for reduced distortion
. It sounds stunning.
 
Re: New DIY Mid High

Have you done this modification by yourself? If so, nice!

I have also experimented with few different (slightly modified) DIY versions of the phase plug, to get the driver and different horns/waveguides mate better.
 
Re: New DIY Mid High

Riku Ylinen said:
Have you done this modification by yourself? If so, nice!

I have also experimented with few different (slightly modified) DIY versions of the phase plug, to get the driver and different horns/waveguides mate better.
Click to expand...
No it's a standard driver, although I think it's quite new. It does not have the dip in the response at 14KHz. My main interest was the lower distortion.
 
Re: New DIY Mid High

No they don't have it listed on their website - don't have the internet at the moment, just my phone..... A large cable has been damaged, when I'm back online I can post a picture, The phase plug extends beyond the mouth of the driver..... Have a look at the impulse response - it's due to the 4594 and it's better than anything I have ever seen!
 
Last edited:
Re: New DIY Mid High

Peter Morris said:
I used a Lake LM26. The system with subs was running 4 way so you need two of them.

BMS however make a nice simple passive crossover for the HF/VHF. The drivers all have excellent frequency responses and need very little DSP correction. The mid horn has a peak at around 400Hz and needs a little boost at 85Hz; provided you don't want to make square waves I think you will get away with a simple three way DSP crossover.

The BMS 4594 nd HE is probably the best compression driver money can buy. It sounds fantastic. Being able to go down as low as 600 Hz or lower solves so many design issues; it allows the 12's to operate fully in their piston range and you have a true point source above 600Hz.
Click to expand...

Disclaimer: I sold the drivers to Peter.
Proclaimer: I also use the product.

One clarification. I probably don't need to say this here, but on some of the more dimly lit corners of the internet there is some confusion about the passive crossovers. All involved in the conversation seem to understand this, but i don't know who all is reading this. The passives are designed to make the two diaphragms act as one driver. Nothing more. No EQing, or zobbeling, or matching the driver to the horn. I have had complaints about the crossover not fixing all the woes one encounters. To me the passive is a great way to plug and play, usually the need for more headroom is not there, and the crossover point, 6.3khz, is out of the vocal range. As Peter points out, it is very simple. 6db one way, 12 the other.

The driver is also designed so that any lag from the passive is picked up physically, EG, it will be in time at the crossover point, and if you do go active, will need to add a slight bit of delay to the HF, to keep in time at the crossover point.
 
Re: New DIY Mid High

Primoz Vozelj said:
BMS coax CD is great. Does anyone know why none of "big" speaker manufacturers isn't using it in their products? The only two I know are Coda and JTR and both are relatively obscure.
Click to expand...

I am not sure what I am allowed to say, or not.
I have signed a couple of NDA's, but also assume that BMS has as well, and don't want to overstep them.
So it's better for me to not say all that I know.
 
Re: New DIY Mid High

Uwe Riemer said:
Thanks Peter,

IMO the vertical directivity behaviour becomes more obvious when flying the speakers.
More overlap in the crossover region is one tool, but puts more stress on the BMS driver. One of my 4592 midranges was killed with 24dB HP at 1k, and my low-mid section ( 2x 15") is way less powerful ( about 100dB sensitivity ).

Did you try to build the low-mid section the other way round, so that they become one horn ?
Click to expand...

Something out of the ordinary has happened to lose a midrange with these settings.
They are rated to cross at 300hz, and I use them to that point regularly.
 
Top Bottom