New FAQ - How to design without measurement gear

[MrkCrwly]

Re: New FAQ - How to design without measurement gear

Outstanding writeup. I'm working on an Amiga design and I think I will "play" and do some reverse engineering as a learning experience.

[PWR RYD]

Re: New FAQ - How to design without measurement gear

Very nice write up. This is the exact procedure I used to design my newest creation (tentively named the RUshers), thanks to Wolf's blog on preparing files for PCD.

[Paul Carmody]

Re: New FAQ - How to design without measurement gear

Very nice write up. This is the exact procedure I used to design my newest creation (tentively named the RUshers), thanks to Wolf's blog on preparing files for PCD.

Yes, Wolf's writeups are also a great asset to the community. This is merely a supplement to them. Jay Kim also had a similar kind of writeup on his site once upon a time.

This type of "pure sim" modeling is very useful, and its usefulness goes way beyond just being a learning tool. With a set of good, patient ears, the sky's the limit! I created the Aethers with it, Maynard created his Tzu-Jans with it, Pete Schumacher uses it all the time... Then there's Wolf, who has basically perfected this style of XO design!

[parodielin]

Re: New FAQ - How to design without measurement gear

This is great!!! Would you like to continue the Saga into the PCD?

[Wolf]

Re: New FAQ - How to design without measurement gear

http://techtalk.parts-express.com/blog.php?b=74

That will give you some things to try and look for/use to get better results.

I guess I need to do a writeup for setup using Series Xovers. It seems not a lot of people know how to use PCD for that. I suppose I could further augment the PCD setup for Parallels too. Maybe that's my next blog....

I hint a bit at it here:

http://techtalk.parts-express.com/blog.php?b=81

Paul- "Perfected??" I don't know about that....

Later,
Wolf

[PWR RYD]

Re: New FAQ - How to design without measurement gear

Paul, I hope you didn't take my post the wrong way. Your write up was fantastic! Love the graphics and examples. Wish I would have read it a year ago. I guess I was trying to say thanks to everybody here (Jeff B. for the Sim tools, Wolf, Paul, Curt, Pete, etc.). You guys have helped me learn so much so fast. I would have most likely gone in several wrong directions trying to simulate without Wolf's blog on using RM and PCD. Pete S. has helped me numerous times (the G-Dogs and the now in process RUsherer). This is a great learning process/adventure Just saying thanks guys! Maybe one day my wife will understand :p

[MrkCrwly]

Re: New FAQ - How to design without measurement gear

Sorry for the really stupid question but I'll ask it anyway.
In your step 2 graph (RS150) there are two lines. How do we know which one to trace?
On the Vifa DX25 graph there are three lines for SPL: Blue Green Red. I assume I use the Blue?
Thanks

[donprice]

Re: New FAQ - How to design without measurement gear

Very nice write up! I'm not sure I've learned more in <10 minutes in a long time.

You would spoil us if you (or another volunteer) continued to follow the design process through PCD. Or add links to where we can find additional info on PCD.

Thanks!

[diy speaker guy]

Re: New FAQ - How to design without measurement gear

You would spoil us if you (or another volunteer) continued to follow the design process through PCD. Or add links to where we can find additional info on PCD.

Thanks!

PCD already has a pretty decent guide created by the author. Also, Wolf has now published 4 excellent blog entries wrt general PCD usage. Fortunately, PCD is also extremely intuitive so it's fairly easy to learn once you get into it.

[Jeff B.]

Re: New FAQ - How to design without measurement gear

This is what is contained inside the Response Modeler. I don't know if anyone reads it or not....

"The Frequency Response Modeler" User Guide
Version 3.0 12/1/2009 by Jeff Bagby

System Requirements:
Which ever version of Excel you use you will need the Analysis ToolPak and the VBA Analysis ToolPak loaded through the Tools Add-ins menu for some functions to work correctly. When the file opens you must say "Yes" to "Enable Macros" for the file to function correctly.

Purpose and Function:
The purpose of this spreadsheet is allow the user to create and save modeled frequency and impedance files (.frd & .zma) for use in other software, such as crossover design programs. It also allows for the user to import existing frequency response files and modify them in various ways such as splicing box response data, adding in baffle diffraction data, raising or lowering a level, adding or subtracting delay, applying various types of equalization, applying standard crossover functions to the data. The program can also extract the phase data from a modified frd or zma file as well. I have attempted to make the function buttons as intuitive and self-explanatory as possible, but I will still give a few tips on using this program. The program is broken into four sections, The Response Modeling section, The Impedance Model, The Box Response Model, and The Baffle Diffraction Response Model. I will comment briefly on the main features of each section.

Main Response Modeling Section
This section of the program is where everything comes together. Here you will find it broken into various subsections. Using the yellow section below the graph you can import in a file that came be modified using the program. Or you may operate from the pre-installed 90dB line and modify it to match a driver's response. You can use the various sections below to add high or low pass roll-off, peaks, dips, modeled bass response, and even baffle diffraction data in order to shape a response to match a particular driver in a specific box. With these tools you can easily model the response of a woofer or a tweeter and save the final response out as an frd file for use in other software. Below is a brief summary of each of the subsections.

Minimum Phase Equalization Module Subsection
In this main section you will note the section titled "Minimum Phase Equalization Modules". Modification or modeling of the frequency response function can be made by using the five "Minimum Phase Response Modules" found at the top of the screen to the left of the graph. You may enter any frequency you choose in line labeled Fc, which will be the center frequency for the module. You may also use the spinners to scroll your chosen frequency up or down to fine tune your response. The frequency chosen will be the center point for the response equalization that will function similar to a parametric equalizer. For module to have an effect you must enter a dB level in the box in the "Gain" line. This dB will represent the magnitude of the peak or dip in the response that you are adding to the response. For example a positive 6 will yield a 6 dB rise centered at your selected frequency, whereas a -6 will yield a 6 dB drop at that frequency. The line marked "Q" determines the bandwidth Q of the equalization. A high number will produce a narrow, sharper spike or trough. A lower number will produce a broad rise or dip in response. A "Q" value of 1 will yield a bandwidth of approximately one octave centered around the Fc. Using these functions is quite easy to add or remove response peaks from a frequency response, model baffle step, or model a bass boost function, then save this response for use in a another program. Although there are only five modules, once you have fine tuned these you can use the "Save EQ Curve to Register…." button to capture the EQ at this point and then begin again with five modules you can effective have as many EQ modules in the response as you want this way.

Adjusting Level and Delay
Below the Minimum Phase Equalization section are two boxes to allow the user to further adjust both dB and Phase of the modeled or modified response function. You may type any value you choose into the dB box. If you are using imported data then the value will need to be in the form of 2 (to raise the level 2 dB) or -3 (to lower the level 3 dB) for example. If you are simply modeling the response the default will be 90 dB, from there you can enter any value you want such as 87 dB or 95 dB to set the response to a new level. In either case you can use the spinner controls to adjust the level up or down in 0.5 dB increments per click if you desire to do it that way. Next to the dB level adjustment is a box to add or subtract delay to the phase response. This delay is expressed in millisecond (mSec). For conversion 1mm of offset would be equivalent to 0.0029 mSec of delay and one Inch would be approximately 0.0737 mSec. By adjusting the value in this field you can adjust the phase response as you desire (as far as adding or subtracting constant delay is concerned). Again, the spinner controls allow for rapid, real-time adjustment of this feature.

Baffle Step and Box Splicing subsections
I will cover the function of this area more completely below when I cover the Box Response and Baffle Diffraction Subsections and how these features in the Main section interact with them. However, it should be noted that you can add or subtract baffle step from the response without doing a complex model of the baffle edge diffraction using the subsection at the bottom if you choose. You can use these buttons in the main section if all you want is a simple step function. You can even choose to add only a percentage of the full step if want, for example if you only want a -3 db step you might select 50% Step.

Response Smoothing Button
At the bottom of the main graph you will see a button for the response smoothing function. This button allow you to import an frd file and then smooth in various level s from 1/48th Octave to 1 Octave if you desire. You may also turn the phase response on and off, which will do more that it off on the graph it sets it to zero for the output if it off on the graph.

High Pass and Low Pass Transfer Function Subsection
Below the Baffle Step and Box Splicing area you will find a Low Pass and High Pass section. You may model the low pass and high pass roll-off characteristics using the Coefficients for a standard Fourth Order polynomial. There are several standardized textbook responses that are available from "No Low Pass" or "No High Pass" to several fourth order varieties. These responses will be applied to both modeled responses and modified frequency response files. (An example here would be to apply a standard active filter to a driver's response curve and then save it out for use in another program). To load the response of your choice; first select that response shape using the spinner under the blue cell for "Selected Textbook Response". Once the response shape is selected click the "Load Coefficients" button. This will load the Coefficients for that response curve to the page. The turnover frequency for the response selected is in the first cell in the line (with blue text). You may enter any frequency you choose and you may adjust it up or down using the spinner under the value. (Note: All boxes with spinners operate this way. You may enter values to the box manually or you may "spin" them up or down using the spinner. Or both.) You will note that there are spinner controls under the Coefficients as well. Even though the program will automatically load the Coefficient for the standard response curve selected, you may still adjust these Coefficients to shape the knee of the response in almost any way you choose using the spinner controls

Adjustable Frequency Response Extension
Similar to the High Pass and Low Pass transfer functions this allows the user to dial-in any corner frequency and any slope desired for roll-off. With this function you can extend the response at any slope rate, or use it as an additional high pass or low pass section that is adjustable to any slope in dB/oct. This allows you to adjust a roll-off to a rate between orders, such as using 2dB/Oct, 7dB/Oct, 21 dB/Oct, or even 200 dB/Oct, as desired.


(Next post for part 2)

[Jeff B.]

Re: New FAQ - How to design without measurement gear

Part 2:

Phase Extraction and Other Features
To the far right you will see the Equalization Transfer Function graph. This graph simply shows the sum total of all EQ and modeling being applied to the imported file. Below this graph you will see buttons that allow for other functions including the ability to import and external filter or equalization curve into the program that will be applied as well. Below this section you will find the Hilbert-Bode FFT Transform that allows you to extract the minimum phase from an frd or zma file. I chose to make this work with external files due to the way the equalization function work in the program. Some functions like the EQ curves add the correct phase already, other function like splicing the box response do not. Because of this is was safer to have this feature work this way: Once you have made all modifications to a file that you want to do then save the file as an frd file and then click the button for "Auto Extract Phase from FRD File", find that file, let the program extract the phase, and then save as the same file name (or another if you want) and the new file will contain the correct minimum phase data. It is always best to let this be the last step in your process.

Impedance Modeling Section
Although they look like two separate sections the Impedance Modeling and Box Modeling Section are tied together; both using the same data. All but the vent size and number is entered into the Impedance Modeler and the Box Modeler simply show calculated box response results. Since both frequency response and impedance data are needed for accurate crossover modeling and simulations it was important for this modeler to include a tool that could create a very accurate impedance model that could then be saved out as a zma file. The Response Modeler's impedance modeling tool is among the most accurate available anywhere. Since voice coil inductance does not behave as a true inductance simply using Le in a model will not generate an impedance curve that is accurate enough for most crossover design modeling. Because of this I included two entries that you may be unfamiliar with: The Le Coefficient and the Le Exponent. I have built a complex expression for a driver's rising impedance with frequency that allows the user to shape the curve by altering a coefficient factor and exponent used in this formula for complex impedance. Based on comparisons the default values of 6 and .565 yield the best curve fit for most drivers where Le was measured at 1kHz. You may, however, change these values if desired to curve-fit an impedance curve if the rising impedance is known. (If values of 1 are entered into each box then the curve will simply match what you get if Le was a pure inductance of that value.) Like with the Frequency Response above I have included a zma file of the Parts Express RS150 Driver. This data is free air impedance data. However, you can use the "Overlay Imported ZMA" button to overlay this response onto the modeled response and then enter the T/S and box parameters into the program. This will effectively model the low frequency impedance of this driver in the desired box and tuning. Now by adjusting the Le Coef. and the Le Expon. You can shape the higher frequency impedance until the impedance and phase line match the imported measured data. At this point the modeled impedance data will be highly accurate and can be save out as a zma file and imported into crossover modeling software. (Values for this sample driver are already included in the program as defaults as an example of how this works).

Impedance Modeling Modes
There are three different impedance modeling modes. 1) Modeling using full Thiele/Small parameters, 2) Modeling using only the Voice Coil Parameters and the Minimum Phase EQ modules, and 3) Modifying an imported curve using the Minimum Phase EQ modules only. You can also select one driver, two drivers in parallel, or two drivers in series as modes as well. And you can choose between vented box and sealed box models also.

Box Response Modeling Section
Once the driver and box parameters have been entered and the impedance modeling is complete you can move down to the Box Response Section and see the "Calculated Box and System Parameters". The graph will also show the modeled low frequency response of the driver in this box. You can now save this response out as an frd file, or you can move back up to the Main Response Modeling section and prepare to splice this modeled bass response onto the frd file that you are working with. The first step in this process is to click on the "Overlay Box Plot" button and using the spinners to the right adjust the box level up or down until it overlays on top of the frequency response data for a smooth splice. You can choose to splice at 100Hz, 200Hz, or 300Hz. When you feel you have the level adjusted click the appropriate "Splice Box Plot" button and the new frequency response will now include the modeled bass response. You will note that the phase line goes to zero at this point too, as splicing loses the phase information. Any plot that is spliced will need the minimum phase response extracted from the saved frd file as described above in the Phase Extraction paragraph. You can also unsplice the bass response if you don't like the result and adjust some more. This unsplice feature is a useful function at times.

Baffle Diffraction Response Modeling Section
A very useful feature in modeling frequency response is the accurate model of baffle edge diffraction. Below the Box Modeling Section you will find the Baffle Diffraction Response Modeling Section which will do exactly this. This section offers some features you will not find in many baffle edge diffraction programs. For example, this program will accurately model the diffraction taking into consideration baffle edge radius and driver piston directivity, which are both selectable by entering the value or via the spinner controls. The program can calculate the baffle response of an open baffle as well. The one limitation of this section is that you are limited to a four-sided rectangular baffle. You can model the diffraction as a loss function or a gain function Once the baffle diffraction is modeled you can save it out as an frd file of its own to use in other software or you can transfer it to the Baffle Diffraction registry and it will be included in the modeled response of the driver. If you already have baffle data saved the program is also capable of importing a BDS file using that instead. One note: It is usually best if you are going to model the box response too, to splice the box response curve to the driver curve BEFORE adding in the baffle response. I placed the Baffle Diffraction Response Model at the bottom because the best results are obtained when this is the last step, or near the last step in the modeling process. (You can still use the EQ and the Highpass and Lowpass section afterward without any errors or problems occurring in the model.).


Thank you,
Jeff Bagby

[Wolf]

Hey, Jeff-

Not that this is a bad place to post this, but maybe you should post them to your blog at your profile as well. That way you have several places on the internet that people can find the info. I know you download it with the User Guide, but if someone found it at your profile too, they wouldn't have to DL it.

Just a thought,
Wolf

[parodielin]

Re: New FAQ - How to design without measurement gear

PCD already has a pretty decent guide created by the author. Also, Wolf has now published 4 excellent blog entries wrt general PCD usage. Fortunately, PCD is also extremely intuitive so it's fairly easy to learn once you get into it.

I am looking for more. For example, what are design considerations when using these files in the PCD. So it's more of a design tutorial rather than using PCD tutorial. I think that's one thing that falls short both in Ray and Vance's books.

[Jeff B.]

Re: Hey, Jeff-

Not that this is a bad place to post this, but maybe you should post them to your blog at your profile as well. That way you have several places on the internet that people can find the info. I know you download it with the User Guide, but if someone found it at your profile too, they wouldn't have to DL it.

Just a thought,
Wolf

It's inside the file. There is no additional download. They just have to look at it.

[Jeff B.]

Re: New FAQ - How to design without measurement gear

I am looking for more. For example, what are design considerations when using these files in the PCD. So it's more of a design tutorial rather than using PCD tutorial. I think that's one thing that falls short both in Ray and Vance's books.

I guess I don't understand what you are asking for here. Can you offer a little more insight into what you mean?