From minidsp to dsp408

Kornbread

   Got the homework to a manageable point due to time off around the holiday and decided to play with the dsp408 for a while before heading to the lake for some fishing. 

   For the most part, physically copied the filter slopes, timing, peq settings from the minidsp to the dsp408.  It sounds different but the results are pretty darn close, at least for one on axis measurement.  I'm thinking off axis should remain the same since the filter slopes are the same.

   The hum is bothersome.  Probably a ground loop somewhere but have yet to figure it out.  There was a little bit of hiss/tweety bird with my ear near the tweeter with the minidsp in the system, but the hum with the dspp408 in the system just too much.

   Notice how the treble is down about 4db in the first trace (light blue).  To my ears it doesn't sound that dark but the room they reside in is about 24'x26' with a low and multi-angled ceiling, all sheetrock plus hardwood floor.  A clap of the hands tells just how live and bright the room is.

   Light blue trace is the baseline minidsp that I've been listening to for time.  The brunt orange and lime green are the newly constructed dsp408 filters.  Huge room nodes around 140 and 40/50hz.   

  

 

Kornbread

Another thing that the minidsp does that I haven't figured out with the dsp408 is a subsonic filter.  I like bass heavy music, deep synth dubstep type of music sometimes played at as loud as it will go levels.  Having the ability to start a steep subsonic filter under 20hz. is a bonus with those ported rs225s.  They make the awfullest sound when they unload and bottom.           

tajanes

Could be a coincidence, but the posted frequency response shows a bumping up in the sub frequencies, at/near the rs225’s fs @ 28.3.

FYI, I run an open-back design (with a dsp low-shelf bump to offset the dipole loss), and put in a very sharp / narrow notch filter at fs to attenuate the loss of control at this frequency.  While it cannot of course adjust fs, it certainly cleans up the woofer’s output by not stressing this run-away due to the driver's loss of control at this frequency.  Give it a try and rerun your sweep.  

Also, where is the box frequency Fb / port targeted at? And courious at to your box net volume and port diamater / length- thx.

dcibel

You're looking at room response @tajanes , not nearfield woofer response. The rest of your comments about Fs aren't really a good description of what goes on at Fs, or a need to notch it out with a filter. It will of course affect the roll-off of a passive filter, but there's really no need to address Fs when using an active filter.

Back to the topic, it would be a better comparison to see the two different DSP response overlapped, but it looks to be different response from about 2kHz and up between the two. Are you sure you have the same transfer function and driver levels between the two? You should be able to directly connect the DSP to your pc soundcard input and run the sweep in REW to view the measure the transfer function of the DSP directly. Just remember to disable any mic calibration.

Kornbread

@tajanes ; I think the hump at ~30hz is the room.  Other speakers have had that hump when measured in about the same location.  These speakers are too big for me to move without a lot of difficulty, and pain, so they get measured inside in their normal position instead of out on the patio like the smaller stuff.  Very live room. 

@d@dcibel ; not sure why the responses didn't overlap better.  Didn't change any volume settings or move the mic but the 2nd and 3rd measurement was way higher on the graph, but the actual output didn't sound any louder than the 1st measurement.  REW has done this to me before and the frequency response always seemed accurate, just not the db.

   Transfer function is crossover slope, right?  Well, both mini and 408 say it's the same. 

   Have got to get rid of the hum before doing any listening.  This may be a bear.  Not sure how much the hum affected the graphs.             

dcibel

Kornbread said:

   Transfer function is crossover slope, right?  Well, both mini and 408 say it's the same. 

    

  Yes, the output of the filter. You can run the output of your PC to the DSP, then plug the output of the DSP back into your PC input to physically measure its response. You know that the Dayton DSP doesn't create correct "textbook filters" when you ask for something like 2kHz 2nd order L-R. It would be interesting to verify that the filter coefficients that the software enters into the DSP are actually true. Put it this way, if it says its the same, but your acoustic measurements are not, that's a pretty good indicator that they're not the same.

Kornbread said:

not sure why the responses didn't overlap better.  Didn't change any volume settings or move the mic but the 2nd and 3rd measurement was way higher on the graph, but the actual output didn't sound any louder than the 1st measurement.

Select your measurement, hit the controls button, that's the gear in the upper right, then look where it says "offset" and "add to data". You can shift your data to whatever SPL you want it to be to overlap them here.

tajanes

dcibel,
Yes, I know what I was looking at (I use the same measuring system) but thanks for playing... 

And, specifically on topic regarding subsonic filter/ bass heavy music (per the 2pm post), while it could be room affect at 28Hz (hence my stating it could be a coincidence), it can in fact help to attenuate/address the issue around Fs, as when music passages sweeps through this range the lower level of control will affect the frequencies the woofer is contemporaneously reproducing. 

Kornbread, the notch is easy to give a go via dsp (at least via the minidsp systems I have used), especially if the dsp408 doesn't easily incorporate a subsonic filter.  It does add value in my system (and may clean up the balance- midrange- of the woofer's response). You had mentioned the woofer unloads and bottoms when you play loud heavey bass, reducing the level at 28Hz and/or revisiting your Fb tuning are both worth dealing with, and to possibly smooth out the 30 to 80Hz a bit. And if you do, I'd be interested in seeing you post a new frequency sweep- thanks and good luck.

Ed_Perkins

Kornbread said:

   Transfer function is crossover slope, right?  Well, both mini and 408 say it's the same. 

         

The phrase "Trust, but verify" comes to my mind, especially with some of the problems the dsp408 has had in the past.  It should be easy enough to directly measure the high and low pass outputs of each device.  I've done it with a few ultra cheap DSP boards to measure the harmonic distortion.

Kornbread

Same colors as before.  The orange and green just sound too hot for me but this is at only one position, off axis might tell more.

There is a pretty tight filter of about -4db@30hz. but really, 30hz is about flat with the mids.  Time to get to work.          

tajanes

thx for posting, 30hz ~flat with the mids ++

re 'time to get to work'

you'd not be alone for preferring a 'BBC dip' (slight reduction acoustic output ~1kHz to 4kHz) -maybe look to move the dip (in your latest post) that looks to center around 4.2kHz towards 2 to 2.5 with a low q, very shallow notch ? 

jhollander

I dislike the old BBC dip. Any reduction of 1K to 4K is in the sensitive listening range for me. Now move the dip to 500 Hz, I'm all in. Similar to the LS3/5s and the Continuum.  I think a reduction from 400 to 700 range cleans up the midrange.

Kornbread

Which leaves me puzzled as to what to do.  My ears do not like the area around maybe 1-4k hot at all (the off axis response might look different, or with this live room I'm sure the listening seat is way different, so maybe that explains it?).  First reflection looks to be @3.34ms.    

To me the dip sounds better; more refined not clashy.  How do people voice their projects at a meet? 

It's taking awhile to figure out what I'm hearing and what freq it is but slowly getting there.

amp=no hum, amp + dsp408= too much hum.  Changing receptacles and ferrite magnets are no help.            

PWRRYD

When I run live sound I usually like to scoop out a few dBs from 400 - 800 Hz on most singer's mics.  It really clears up their voices.  Plus there is already enough stuff that's playing in that range (piano, guitars, etc.).

I use to like the 1 - 4k BBC dip on my speakers.  Then I realized what I was missing and how bad my old car stereo systems probably sounded.

jhollander

The graph you are showing is heavy in the mid because the highs are pushed down.  If this is me I want the woofer to set the level. If that is 81 dB at 350 Hz set that as your goal and flatten the peaks (not average) to this level.  I'd tweak after you are flat.  YMMV

Kornbread

Was going to compare the two unites so I tried to measure the crossover slopes of the minidsp by feeding it from the headphone out and looping that back into the mic in. 

Played with a bunch of settings in REW but couldn't get it to work that way.   

Ed_Perkins

How about running the minidsp output directly to the mic in?  You may have to turn the level on the minidsp down quite a bit (and/or the mic level), but it should work.  I've done this, but I have an external USB interface that gives me more adjustments, including a 20 dB pad button.