MelodyMan

Help with rePhase and linear phase crossover

Recommended Posts

Looking for some help with using rePhase to create a linear-phase crossover.  I have read miniDSP app notes here but cant make sense of the first step ''Use the Paragraphic Gain EQ tab with minimum-phase filters to flatten the amplitude response'.  I cannot find a detailed explanation of the type and shape of filters to use to achieve this flattening. If anyone knows of a more detailed guide to using the tool or could guide me with some sample settings to fit the curves that would be most appreciated.  I include a photo of the left mid SPL and Phase measurement in REW and the same measurement imported to rePhase.  It would also be useful to know is phase something like the inverse of SPL and on the phase diagram what is the significance of the first peak in the chart between 50 and 100 hz before the vertical dotted line.

Thanks in advance.

left_mid_rew_spl.thumb.PNG.2c9823a799a9dcd5965072c1d1c76369.PNG

left_mid_rew_phase.thumb.PNG.8e9b40ca5e8fea85c1616a3ba55566af.PNG

rePhase


left_mid_rephase_import.thumb.PNG.455d2010290884539e0c3a57860ff5f1.PNG

Share this post


Link to post
Share on other sites

The phase is the timing of the different wavelengths produced by the speaker and is not related to the amplitude or SPL response as far as this is concerned. The phase and amplitude can be treated totally independently or they can be altered together, depending on the type of adjustment required.  

As a driver cone moves back and forward to create sound waves, the time it takes to create each frequency can be different meaning the phase is different on different frequencies.  The phase plot shows how much variation the driver is producing between the timing of all the different frequencies that you're asking it to produce.  The amplitude graph is obviously just how loud each frequency is.

The options in rephase can either adjust the amplitude of the chosen frequencies along with an equal change in phase, or they can just alter the phase of each frequency without changing the amplitude, or they can adjust the amplitude without changing the existing phase.  Which one you use & need is down to your specific measurements.

Minidsp suggest using the paragraphic gain EQ adjustments in order to get rid of those three peaks that you see in your first image.  In the bottom image (showing the rephase software), you aren't actually on the correct tab that you need in order to get past the first step of the tutorial.  Your image shows the minimum-phase manual filter tab is selected but it should be showing the paragraphic gain EQ tab.  The correct one (if following the mindsp guide) should have sliders on it that look like an old graphic EQ.

rephase-screenshot.png

This part is very easy to do, simply look at the amplitude response and move the corresponding sliders downward where you have those three raised notches in your SPL. Adjust until they have gone away.

If you follow the minidsp tutorial, you should be working on minimum phase filters to flatten out the SPL (removing those three peaks in your SPL without altering the phase) and then filters to just correct the overall phase without changing the amplitude. It's easy once you know how but it can seem difficult at the start. Have a play and try things out, you get a visual representation of most changes so you can see for yourself what effects the options would have and you can easy change them back if not correct. It's a good way to learn.

The flip at around 130Hz is just a visual representation to show that the phase has shifted more than 180 degrees.  At that point, it has moved so far out of alignment that the phase has inverted. A rear port will do this or it could just be that the mic has picked up a phase issue with the speakers interacting with the room.  You can flip it back in rephase but I'd take some more measurements from different positions to confirm it's not just the room.

Edited by rv295
Added extra info about ports
  • Upvote 1

Share this post


Link to post
Share on other sites

If you'd like a quick starting point on how to get past the first stage then I'll add some instructions below. I'll start off explaining how to remove one of those humps in your SPL with the Paragraphic gain EQ function of rephase, as per the minidsp guide. 

The hump in the middle of your SPL measurement is the easiest to start with and learn from.

Open up rephase and load the measurement that you used in the first image.

Click on the Paragraphic gain EQ tab so you have the sliders on screen.  

Decide upon the frequency to start the adjustment, this is the frequency that corresponds with the very middle point of the area to be adjusted (the centre of the hump).  It looks to me that the centre of that middle hump is around 420Hz.  So...

Pick a slider, one that has a Hz figure somewhere near 420Hz and change the figure in the Hz box below that slider so that it states 420Hz.  We are going to be using that slider for this adjustment.

Now you need to decide how much attenuation is required to correct that middle hump in your SPL measurement.  The peak of the middle hump looks to be around 5dB higher than the average response from that driver.  So, I'd either pull the 420Hz slider downward until it states -5dB or just type -5 in to the dB box below the slider.

Now you need to specify the Q factor.  The Q factor is how sharp or how broad the adjustment is going to be.  A high Q factor (such as 15) will produce a thin, sharp adjustment that doesn't effect many frequencies.  A Q of 15 will effect about 1/10th of an octave.

A low Q factor (such as 0.404) will effect a much wider range of frequencies.  A Q of 0.404 will adjust 3 octaves worth of frequencies.

Looking at your 1st image, the middle humps covers the frequency range from around 250Hz up to 600Hz, this is roughly 1 and 1/5 octaves so would require a Q factor of about 1.25 

To calculate an octave in frequency terms, you need to double the starting frequency.  So if you start at 250Hz. One octave up is 500Hz and one octave up from that is 1kHz and so on.

Here are a few samples of different Q factors and how many octaves they cover. 

Bandwidth
in Octaves        Q factor
3.0                     0.404
2.0                     0.667
1.5                     0.920
1.0                     1.414
2/3                     2.145
1/2                     2.871
1/3                     4.318
1/6                     8.651
1/12th               17.310


Once you've set the slider to correct the middle hump, the boxes below the slider should state dB= -5  Q= 1.25 and Hz= 420.  You may need to slightly tweak these figures but I'd use this as a starting point.  Then take a new measurement with the filter applied. If it's not right, go back and tweak the figures.  All the other settings on the rephase screen should be as per the minidsp guide.

  • Upvote 1

Share this post


Link to post
Share on other sites

Hi Ross, thanks for the explanation and tutorial very helpful.  When looking at your rePhase screenshot I see a normal bell shaped SPL curve as a solid line but on my measurement imported to rePhase I only see phase as a dotted line inversed like a U shape.  Which means I would have to apply the opporsite to your recommendations to get correct movements. How would I see the solid SPL line in rePhase? Or have I measured something wrong. Screen shot attached.

thanks

left_mid_rephase_adjust.thumb.PNG.e08b3b921771e985c98a37d560576d3c.PNG

Share this post


Link to post
Share on other sites
13 minutes ago, MelodyMan said:

Hi Ross, thanks for the explanation and tutorial very helpful.  When looking at your rePhase screenshot I see a normal bell shaped SPL curve as a solid line but on my measurement imported to rePhase I only see phase as a dotted line inversed like a U shape.  Which means I would have to apply the opporsite to your recommendations to get correct movements. How would I see the solid SPL line in rePhase? Or have I measured something wrong. Screen shot attached.

thanks

Hi Dean, In rephase, the dotted lines is phase and the solid line is the SPL.  Can you not change your view settings to show both phase and amplitude?  Your image only seems to be showing the phase but you need to view the SPL graph in order to complete the first part of the minidsp tutorial.  I'm sure it's only your view settings or program options that need altering. 

Not that it matters, but that rephase image that I posted was from the minidsp tutorial that you linked to.

You could also get REW to do a lot of this first part for you... REW can calculate and generate these types of PEQ filters to flatten out a driver or speaker response.  All you would need to do is copy over the Q factor, dB adjustment and the Frequency for each filter that REW generates when calculating the best way to correct the driver response.   The automated REW filters would work in this scenario as you'd be using them for amplitude adjustments only.  The REW filters can be generated by opening your measurement in REW and then clicking on the big square 'EQ' button at the top of the measurement window.  It's pretty obvious how it works but google will turn up some useful guides on how to use the REW EQ function for generating correction filter details.   The only thing REW can't do is calculate the filters to correct the phase. You'll have to do that trial and error, manually.

Share this post


Link to post
Share on other sites

I hate reading my posts back, please forgive the many spelling and grammar errors.

I had forgotten to mention that I'm trying to help from memory alone.  I have loaned my laptop and microphone set-up to a friend so I am unable to open up rephase and work out how to switch the preview between phase and SPL or both.  If you are completely stuck, let me know and also send me the measurement file that you imported in to rephase. (you should be able to attach it to a private message, failing that, drop me a message and I'll send you my email address).

...

Edit:  Another thing to note early on, is that the minidsp unit can only run a certain number of taps due to hardware constraints.  Rephase could be producing a larger number of taps by default.  You may need to look up how many taps you can apply to each channel and set the maximum number of taps in rephase.

I don't want you to go through this whole process to find that you can't actually run the filters that you end up making because they need more taps than the minidsp can handle.

Edited by rv295

Share this post


Link to post
Share on other sites

Hi Dean, a couple of points when using rePhase and OpenDRC-DA8. I'm writing from memory so forgive me if detail is vague. 

rePhase is not very user friendly as you've noticed however you can set 6 different views and save these settings. Import of filters and measurements from is possible besides phase. 

I've approached the subject differently when designing crossovers. As what you're trying to achieve was incredibly difficult. So here's some pointers if practical:

- read about different crossover types paying particular attention to what happens with phase 

- choose crossover type based on the above and use time delay to model phase behaviour of both drivers 

- use FIR to flatten phase response particularly if you're dealing with bass. OpenDRC-DA8 will not have enough taps if you'll want to do both and since you can use PEQ to deal with amplitude it's only reasonable to use FIR for phase

- remember FIR filters introduce delay to the system don't forget to add time delay as it will mess things up if you don't

- if all else fails have a look at Eclipse Audio FIR Designer paid software which seems to address most issues you're having with rePhase. I've nearly pulled the trigger but back using passives at the minute

Share this post


Link to post
Share on other sites

Hi, got delayed on reply but  thanks for advice. The problem is that the magnitude or SPL line does not display in rePhase from my measurement file, if I use the same file in FIR Designer I see the SPL and Phase lines together as expected, image attached. I have tried exporting the measurements again and reinstalling rePhase and plyaing around with settings but nothing will expose magnitude/SPL. 

Ross, I will send my measurement file to your PM and would appreciate if you can open and see if it works for you.  I will post the resolution on this thread for anyone else with the same problem.

I do like the look of FIR Designer but this is a pro tool and being a novice I'm not ready to spend $100 on XO software until I can see some results with linear phase first which is why I would like to get rePhase working in first instance to give me some confidence to go forward.  Im also looking into Lin Exakt as it does look easier and Im already pretty sure the DAC inside the OpenDRC DA8 is not the sound I'm after.

59e9d57e5ea7e_FIRDesigner.thumb.PNG.91b0717641382bd1c17651808c127fe7.PNG

Share this post


Link to post
Share on other sites

Interesting, OpenDRC-DA8 uses AKM4440 and I've compared it to other DAC chips. Level matched. It really is a good sounding chip however it may not shit the rest of your system. 

In regards to showing magnitude from imported file in rePhase change the view on the right hand side tab. As default it's set to strange values something like -100dB. It will then show up. 

  • Upvote 1

Share this post


Link to post
Share on other sites

Luke was right, your SPL is simply not appearing because of the range displayed by default in rephase.  I imported your measurement in to REW first, to check that it loads correctly and there was no problem, both phase and SPL were displayed.

To view your SPL in rephase, simply adjust the gain offset as per the image below.

rephase-screenshot.jpg.1afbd12da05bc01fd4964cefbdf2a0c7.jpg

After taking another look, it is best to enter -95 rather than 90.   90 will allow you to see the SPL in full but -95 will keep it just below 0dB (full strength digital gain).

Edited by rv295

Share this post


Link to post
Share on other sites
1 hour ago, MelodyMan said:

Yes that was it, can see it now. Thanks

Good stuff... If I were you, I'd just use the Paragraphic Gain EQ (set on linear phase) in order to do minor correction on the driver SPL without altering the phase, and then use the Paragraphic Phase EQ tab to correct the phase.  Set the maximum number of taps that the minidsp can run and Bob's your uncle.

The problem with this method of correction is that the results can only be as good as the measurements used to create the filters. 

To start out, I'd most certainly ignore anything small or any unusual problems in the measurements that can't be explained.  These are most likely going to be caused by very specific positioning issues with the room & mic or speaker & mic so therefore shouldn't be corrected.  Take lots of measurements for each driver with the mic moved around very slightly, an inch or two.  If the issues you see in one graph appear in them all then it is safe to assume it is something that you want to correct.  If the issue only appears in some or just one measurement, then it best to ignore it. 

If the phase follows the SPL, ignore the little spikes and look at the average shape of the phase, then correct for that average curve.  What you'll get in real terms is a better match for the actual phase variation of the driver and therefore the crossover points.  Trying to correct for the little dips and peaks will just make things worse as they are a result of the room and speakers and not what the driver is actually producing if completely isolated on its own.

  • Like 1

Share this post


Link to post
Share on other sites

I think Im getting the hang of rePhase now, although looking at my bass driver phase line it wraps at around 107hz and Im not sure if I should try to even out the wrap or if this is not possible and not a problem. Seems all the examples have a nice flat line around -45 but mine sits around -15 and then wraps at 107hz.  Could this be a bad measurement?

left_bass_rephase_adjusted.thumb.PNG.9880e05d70d85e2d1ba2b5f3a3bdb883.PNG

Share this post


Link to post
Share on other sites

You've done pretty well getting a semi usable phase measurement from the bass while indoors, they are the hardest to measure as the bass wavelengths can be longer than the room you're measuring them in (or at least longer than the distance to the mic and the first reflection point) so the measurements usually contain a lot of phase flips and a lot of cancellations making them very hard to read accurately.  

I would expect that sharp rise in phase to either be an effect of the adjustments that you've made or less likely, a room and mic issue.  Therefore, the only sensible thing to do is to take more measurements (both fresh measurements and also with that filter applied to see the result).  The phase certainly wouldn't naturally be like that.

Are you able to take one of your bass cabs outdoors and measure it with more free space around, preferably raised up off the ground?   If so, it's the only sensible way of getting a really good bass measurement that you can be sure is right.

Share this post


Link to post
Share on other sites

@MelodyMan  How did you get on Dean?  I'm interested in an update if you spot this and I hope everything went well.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.


  • Recently Browsing   0 members

    No registered users viewing this page.