Home Theater Forum and Systems banner

1 - 14 of 14 Posts

·
Registered
Joined
·
18 Posts
Discussion Starter · #1 ·
Hi all, new here.
Hoping there is easy solution to this.
I have a sub that must have linkwitz transform for certain response.
Because I know the particular EQ parameters needed, (which I gather from simulation software - Win ISD).
I would like to synthesize the curve digitally as opposed to via analog circuit.

Keeping in mind measurement sweeps would need LW taken into account aswell.
My initial thought was to insert EQ prior to sound card, so that anything played via that but I'm unaware of such software, I could use EQ within playback software but the EQ would have to match the response used to measure with REW.

With REW - and I'm unsure If TARGET CURVE function can already otherwise do this -
Is there a way to do this? i.e. play measurements via linkwitz transform EQ?

I can't be the only one with this in mind, it does save implementing a physical Transform circuit (though it should be done, it saves a few bucks for time being).

So any ideas?

thanks. :wave:
 

·
Registered
Joined
·
84 Posts
J River Media Center will do what you want. You can program the lt in the PEQ, then measure the response to see if it's working. You would use the "loopback" feature in MC which would grab the audio sent from REW and process it through the DSP.
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #4 · (Edited)
Hi all

Thanks for the replies.

John,
Apo seems to be what is required thanks.

However WinISD does not have export function unfortunately and I'm not confident manually transferring desired response (quality factors et all) across to external program (to then create txt file to use with APO if you catch my drift).

I will say that I'm not relying entirely on digital signal processing - I have implemented an anlog 24dB/oct HPF and xo circuit(s)... and have taken this into account with winISD:
http://i.imgur.com/Z8LgkW2.jpg
http://i.imgur.com/hYddqeg.jpg

The latter (final response) I think is what I need to implement with PEQ (and keep in place at all times using APO).

So question now is;
what's easiest way of simulating and then generating LW transform to then export via text for use with APO?

I'll probably finalize a guide online when I'm done because honestly this project is taking longer than expected.

Thanks again
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #5 ·
I've yet to check this but iirc (no web access at home) REW does have fields where you can enter driver paramters, Qts etc, though not as accurate as WinISd, can REW be used to generate and compensate a LW curve in the first instance? I doubt it can...
 

·
Premium Member
Joined
·
1,562 Posts
We can enter any SPL target curve we want into REW and it will calculate filters for us. REW help has some info on entering custom target or house curves. We can provide additional help if needed. The calculated filters can be entered and used in APO, another EQ program, or an EQ device.

If we do enter a theoretical LF response shape and then measure the speaker/room response, we will find that the room impact will have a very large effect on the actual measured response. The well calculated theoretical response may not even be recognizable among all the room modes. More filters can then be calculated to meet the final room response target (house curve). This can result in the desired response.

I suspect that this method is likely to have more filters with some partially working against each other. This should not be real problem. It just may not be a very efficient use of filters.

An option is to just use the final house curve target for the calculation of the filters initially. That way REW will find the most accurate and efficient use of filters to correct for an actual speaker/room installation.

Either way will probably provide acceptable results. There may be less work, less filters for the same end result with the second approach.

The theoretical response shaping (targeting an anechoic response) is the only option if we do not have the actual room and SW placement established. This is why it is not done by speaker manufactures. A nominal target is their only option considering they know nothing about the eventual installation.
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #7 · (Edited)
jtalden, I think this is over complicating things. I would like to run all measurements through a Linkwitz EQ appended prior to soundcard. APO is a viable contender - it simply replaces a physical incarnation / analog circuit.
There is a HPF filter in signal path aswell - therefore the final response is flat to 25hz in the end - I can't see how this could affect measurements if a sweep with correct duration is used.

The question now is how to convert/transfer the below parameters required (derived from WinISD) to a txt file compatible with APO.

Since john's post, I discovered there is in fact a GUI version of APO EQ but this doesn't change situation.

Something tells me Linkwitz Transform shape is more than a simple low shelf as it appears.
I'm not mathematically gifted at all, so have no idea how to copy whats written below into parametric EQ form (even graphically).

In my case, driver in sealed 78L box without EQ:
f0 = 44.677
Q0 = 0.818

The linkwitz parameters needed (derived from WinISD):
fp = 10hz
Qp = 0.707

Once in place, all measurements and audio passes through APO with LW, just like an analog circuit.
Just don't know how to transfer to txt to be used with APO.

Hope this clarifies :bigsmile:

Any ideas?
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
I am probably misunderstanding your question, because it seems to me that it is you that wants to over complicate the situation. :scratch:

The type of filters implemented to achieve a final target LP response (house curve) is not relevant to the sound quality. REW will calculate a filter set to achieve the target response. Do you care for some reason that that an electrical LT is synthesized first using a shelving filter plus peaking filters, or are you just interested in just achieving the desired house curve at the LP? As I said originally, either way can work - your choice.

If you feel you want a shelf filter as a part of a filter set used to first implement an LT electrical shape then one can be entered manually into REW. You can still then use REW to calculate additional parametric peak filters for additional shaping to the final electrical LT if needed. The manually entered shelf filter will be a part of the overall filter compliment that REW lists.

I was trying to highlight that you have two choices for your process of achieving a target SPL response at the LP.

Easier way:
> Place the speakers in the room positions.
> Measure the room response at the LP using any analog filter or XO setup you have implemented.
> Enter the desired LP response into REW's house curve (flat, sloped or any shape your want).
> In the EQ panel allow REW to calculate the filters needed.
> Apply the filters in APO.
> Remeasure to confirm the predicted result is achieved.

More complicated way:
Step 1. [Create an electrical LT response]
> Enter the desired preliminary modified LT target curve into REW's house curve.
> Measure the soundcard loopback or create and enter a linear SPL response manually into REW. [Maybe easier/better to measure a loopback from your analog LPF/XO unit so any analog HPF is in the response?]
> Go to the REW EQ panel and manually enter a shelf filter if you want to force one for part of the filter set.
> Then allow REW to calculate any remaining filters needed to achieve the modified LT target you entered.
> Apply the filters in APO.
> Remeasure the loopback to confirm the target was achieved. You now have created the desired electrical LT.

Step 2. [EQ to House curve]
> Place the speakers in the room positions.
> Measure the room response at the LP using any analog XO/filters and the APO LT filter set implemented per step 1.
> Enter the desired LP response into REW's house curve (flat, sloped or any shape your want).
> Go to the REW EQ panel and allow REW to calculate any remaining filters needed to fit your house curve.
> Add the new filters to the original LT filter set in APO .
> Remeasure the LP response to confirm the target was achieved.

Both methods will create the same sound quality.

The APO thread should have all the needed info to implement the filters using APO. When you have your filters defined by REW, I am sure one of the APO users can assist you in how to if you are having difficulty with that step.
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #9 ·
Re: Measuring via Synthesized (digital) Linkwitz Transform?

I forgot to mention in OP I wanted to convolve what REW generates with preferred music player.
I see more clearly now how this may confuse things.

Unless with APO you can append two subsequent EQs? the latter being the LW?


I will try messing around with REW tonight and report back tomorrow.

I will save your text aswell for further reading.

Thanks for the help (though I should have been more clear). :wave:
 

·
Registered
Joined
·
84 Posts
J River, that I mentioned, has an lt in its' PEQ that will let you input all 4 parameters so it's tuned correctly. It has a loopback feature as well that will let you process any audio on your PC through its DSP. You could do a physical loopback connection on your soundcard to see if the lt circuit is performing correctly. If it is, you could use that measurement to create a Linkwitz transform target curve in REW.
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #11 ·
Back again,

I looked in REW. As has been suggested, the only way to input LT response is via REWs House curve feature. However, it still can only be imported by means of .txt.

My question is how to create .txt format EQ. I'm not aware of any programs that allow to create then export LWT as .txt.

There are various LT calculators online, most are spreadsheets and so I doubt any can export as .txt response.

Also,
@ nate - thanks for suggesting JRiver, but I intend to use an alternative for the time being
 

·
Premium Member
Joined
·
1,562 Posts
Entering a house curve into REW is easy if you know how to create a text file. The instructions for the file contents is found at REW "Help/Preferences/House Curve". Just enter the XY points of the LT curve you want to create into the text file. You can use a spreadsheet to create the XY points and expert them to the text file or just read them from the WinISD chart you created and manually enter them.

Load the resulting text file into "Preferences/House Curve". Be sure to set the "Preference/Equalizer/Slope" settings to "0" instead of the default. Otherwise the house curve you entered will be impacted by the slope settings; we don't want that.

To create the filter settings needed using REW I suggest a loopback measurement of the sub output that includes the LP/HP filters if that is possible. If the house curve is your final curve (#3) REW can calculate the filters needed directly from that. You could set a shelf filter manually and then let REW calculate the remaining peak filters, but looking at your target shape, I don't think a shelf filter is called for. If you can't measure a loopback with HP/LP filters active then the appropriate changes to the house curve should be made.

Even if you aren't doing any measuring with REW and instead just calculating filters sans the HP/HP filters you can do that also. It is just necessary to define the house curve appropriately and to import a flat measurement file for the REW for calculation.
 

·
Registered
Joined
·
18 Posts
Discussion Starter · #13 · (Edited)
Hi

I now realize APO EQ is incompatible with XP!
Looks like I shall have to use REWs House curve. I will get round to REW eventually but I'm stuck at this stage which is prior anything to do with REW:

Creation of .txt (for importing/use with REW).

I discovered the UniBOX spreadsheet floating about online. It has a .FRD export feature built in and these .FRD files can be opened with notepad. It has the following headers:

Frequency[Hz] Level[dBspl] Phase[Deg]
Example: 10.0000 57.5817 171.0299

I've attached an .FRD file to this post.
It's in .txt format because site wouldn't allow upload of .FRD - it's the same anyway.

jtalden, You say to export from winISD its a case of taking X/Y (Freq./Level) plots from a graph and transferring across to devise a curve. Is it really as easy as picking random plots from a graph?
Shouldn't considerations be made to the Q factor also (for example)?

Thanks again.
 

Attachments

·
Premium Member
Joined
·
1,562 Posts
Yes, it can be done manually that way. But...

If the attached curve were the house curve you wanted, it can be imported as is. Just load it into REW as the house curve. Below, I did that so show you the result. You must adjust the house curve settings as shown or the curve will be distorted as the settings dictate. Also, a house curves is normally entered with the SPL levels being relative to 0dB as shown in REW Help. Since your file is at ~93B you have to adjust it accordingly. I just told REW to reduce the target level to 0 dB instead of the default 75dB. The curve then fell right at ~93dB as shown. [An option is to edit the text file to reduce the SPL level by ~93dB. The third column in the file can be deleted if you want because the phase is not used in the house curve. REW just ignores it, so it can also just be left as is - no problem.]

I also imported the same file as a measurement. It is also shown on the chart. So the chart shows the same response for both the house curve and the measured response. Since I moved the house curve down 75dB from ~168dB to the same level as the imported response, the 2 traces overlap each other. The legend shows that both are being plotted. I maybe should have offset them a little. If this were the current actual vs the target house curve an REW filter calculation would indicate that no filters are needed.

EQ Shot2.png

Your question about the Q factor confuses me. We are not entering filters into REW (at least not yet). We are entering the house curve and the response that we want to modify (in you case a flat curve modified by the LP/HP filters should be measured/entered). REW then can calculate the filters needed to modify that "actual response" into the target house curve response.

It is very likely that I am still not understanding you situation though.

I finally surmised that you wanted to calculate filters needed to create an LT electrical response as modified to include the HP/LP filters, i.e. your curve#3:
To do that we could:
> Enter curve #3 as a house curve. [The target response of the voltage to the SW.]
> You have the HPF and LPF filters implement in an external circuit now so a loopback measurement using REW will give your current electrical response. It would be a flat bandpass with the the HP / LP filters applied the response. It would not have the peak shown in #3, the house curve.
> Allow REW to calculate the filters needed to modify the measured response to match the house curve response.
> Enter those filters into an EQ unit (APO or other) and activate it.
> Measure the loopback again to confirm the measurement now follows the house curve.

You have not confirmed if you intend to measure a voltage loopback of your system using REW, or are just using REW to do filter calculations based on imported files. Which is it? Either can be done.
 
1 - 14 of 14 Posts
Top