Announcement

Collapse
No announcement yet.

Help with 3-way design - Klipsch Heresy III clones!

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Help with 3-way design - Klipsch Heresy III clones!

    hi all, i'm currently working on a 3-way design on a budget! I like the look of the Klipsch Heresy III and i think their shape would fit perfectly in my space. What I don't like is that the original design would most likely require a subwoofer and they're beyond my budget ($300) when new. I'm a pretty experienced cabinet maker/woodworker so I feel pretty confident in my ability to recreate them in walnut veneer. What I don't trust are my speaker design skills...hence why I am reaching out to the community for help with my project. I will be posting regular updates for those interested in the build. with the intro out of the way, this is what i'm thinking:

    Drivers:
    Peerless by Tymphany D19TD-05 3/4" Poly Dome Tweeter
    HiVi C3N-III 3" Paper Cone Full-Range Driver
    Dayton Audio DC250-8 10" Classic Woofer

    Filters (see attached image for calculated values and actual inductor and capacitor values in red):
    Tweeter: 2nd order butterworth HPF @ 5200 hz, + L-pad attenuator
    Mid: 2nd order butterworth band-pass @ 750-5200 hz
    Woofer: 2nd order butterworth LPF @ 750 hz
    Click image for larger version

Name:	image_82504.jpg
Views:	401
Size:	148.4 KB
ID:	1428614
    Enclosure:
    15.5" W x 21.5" H x 12" D; 3/4" walnut plywood carcass, 3/4" MDF front and rear baffles. Solid pine internal bracing
    Sealed 1L enclosure for mid + tweeter
    Ported 40L enclosure for woofer. 2.5" dia x 6" port tuned @ 35 hz for a +2db peak in mid 50hz and an F3 of 35 hz...according to my modeling.

    Questions/concerns for the community:
    does the logic above seem sound or do you see any major red flags? I don't have much experience in speaker design so I'm hoping some more experienced community members can QC my work. My main concern is my choice of mid-woofer. It has a stated sensitivity of 84 dB 2.83V/1m which is much lower than the DC250-8's sensitivity of 88 dB 2.83V/1m. However, i'm not sure how accurate these measurements are. The woofer seems highly efficient given it's price range (i was able to snatch 2 of them for $15 each on a clearance sale). Does anyone have any real-world experience or measurements? Is there any way to design around this? I think an L-pad for the woofer would be a waste of power and the HiVi drivers are only $6 each so I thought they would be worth trying. Any other tips / comments are appreciated!

    Project requirements:
    Less than $300 in parts
    F3 must be no higher than 35 hz
    Max SPL 105 dB or higher
    Must fit the enclosure above in order to keep to the original Klipsch Heresy "look"
    I already have the 10" woofers and tweeters lying around so I would like to keep those in the design.

    These will be for casual listening / party speakers so I don't think sound quality is critical but they should sound full and not distorted....and most importantly they will need to LOOK good

  • #2
    The results of that crossover calculator likely won't sound good - they rarely do. It's missing some key elements - like BSC, proper (measured or simulated) frequency and impedance files, and phase information.

    Start here - https://sites.google.com/site/undefinition/diy

    On the right hand side you'll see some articles Paul wrote many moons ago. They still hold true today!
    Co-conspirator in the development of the "CR Gnarly Fidelity Reduction Unit" - Registered Trademark, Patent Pending.

    Comment


    • #3
      hi, thank you for the response. I will look into it... I too am a big fan of Paul's site and have previously built a couple of his designs with great results. The issue I'm having is that I can't find much information on the HiVi C3N-III...the dayton stuff is usually pretty good with documentation. And yes, I know it is much easier and I can probably get better results by building someone else's design but that's not the point of this project...

      I wanted to start from the ground up because I am on a budget and already have some of the drivers.

      Comment


      • #4
        This is a great place to learn the ropes and I applaud your willingness to dive right in. But I will say most people start with a 2-way. There are so many things to juggle with your first design - a 3-way system really compounds things.

        ABout your mid choice - I would not expect much from a $6.00 driver, but I have been surprised in the past. The fact that they don't publish any data would disqualify it for me. You can find great sounding mid/full range drivers for under $20. Look at some of the Peerless - the TC9FD18 is a nice one. But, I would urge you to invest in a calibrated mic before buying anything else. They are not horribly expensive - $75 for the UMM6. Measurement software is free - ARTA or Room EQ Wizard. You can design a speaker without measuring, but there can be problems gathering files from various sources. There will likely be slight differences in measurement distance, time of flight, and SPL levels. All that adds up to data that won't reflect what you are modeling & eventually hearing.
        Co-conspirator in the development of the "CR Gnarly Fidelity Reduction Unit" - Registered Trademark, Patent Pending.

        Comment


        • #5
          No offense meant, but the Klipsch Heresy III is a 12" 3 way with horn loaded compression driver mid and tweeter. I don't see much in common with your concept besides the general box shape.

          Comment


          • #6
            Originally posted by djg View Post
            No offense meant, but the Klipsch Heresy III is a 12" 3 way with horn loaded compression driver mid and tweeter. I don't see much in common with your concept besides the general box shape.
            None taken and you are correct, it is not my intent to have similar sounding speakers -- I just like the way the heresys look (call me vain). I've heard klipsch heritage speakers before and they are not my cup of tea...but i do very much enjoy the way they look.

            with that said, my original plan was to put the Paul's Amigas in this cabinet since the parts are within budget however, the enclosure is much larger than what the amigas call for. Plus i'm not sure what effect the wider and shorter baffle would have on their sound. As I mentioned in my OP, i got a good deal on a set of DC250-8 so I wanted to design around them. Their FR graph shows a pretty large peak at 2000 Hz so I would think they would have to be crossed no higher than 1500 hz. I thought it would be just as difficult to find a tweeter and design a 2-way that can cross over so low...but i'm sure its possible. The Dayton DC28F-8 seems like can be crossed pretty low but I already have the D19 tweeters, which i think will sound better. I have not been thrilled by the way dayton tweeters sound in the past but I must admit that I have not heard the DC28F-8.

            I picked the HiVi for the mids because i've had very good luck with their drivers. It's unfortunate that the data for it is not available since their other drivers (B3N, etc) have a wealth of documentation. Still, I trust the brand and at $6, I thought I can give them a shot and see what happens. Their other 3-4" drivers are even less efficient so I would have to use multiples, which raises the cost and complexity. I wanted to see if anyone has any experience with them because theres a chance that they will not sound great without additional filters.

            Comment


            • #7
              That woofer likes another cu.ft. to hit the mid 30s. I'd shoot for 2.5cf w/a 3"id x 7" long Precision Port.
              It can only take 20wRMS @ Xmax @ 104dB.
              YOUR (smaller) box can take 34w (@ 107dB - not accounting for baffle step loss), but I only see an F3 in the mid 40s w/a +3dB hump around 80.

              To design an XO that works, you need (at a minimum) a plot of the driver's freq. response. I don't believe HiVi has any published for that 3".
              "Textbook" (or "online") calculators don't (generally) work well at all.
              Your 750Hz filter on that woofer will actually roll it off about an octave higher (1500 Hz) AND it leaves it with a +6dB "hump" (bad) from cone breakup right below 1kHz.

              Your 5k filter's transfer fn on the tweeter shows it @ -3dB @ 5.5kHz (pretty close) and -6dB down @ 4k.
              It also leaves the tweeter running about +9dB too "hot" (in relation to the woofer).
              Last edited by Chris Roemer; 01-07-2020, 10:51 AM.

              Comment


              • #8
                Most/All of your requirements are met by Paul Carmody's Tarkus

                Within budget too!

                F3 of 29 Hz and can pump out 106 dB!

                https://sites.google.com/site/undefinition/tarkus

                That's a proven and well designed 3 way. If you are not after the Klipsch sound, I don't think you'll be able to design a better 3 way speaker than what is done by experienced designers, such as Paul Carmody and others. Don't take this as an act of discouragement, if you want to try it on your own still, you really should!

                If I were you, i'd consider the Tarkus and see the builds others have done at the bottom of the page, and see whatever suits you! Put some grills on them maybe!

                Comment


                • #9
                  Also, something more similar shape and characteristic would be Pi Speakers. Check them out if you don't know about them.

                  http://www.pispeakers.com/catalog/pr...0a4f6b47462815

                  http://www.pispeakers.com/catalog/pr...0a4f6b47462815

                  I doubt they get down to 35hz, but if that's your choice of cabinet, I think these are pretty close!

                  Comment


                  • #10
                    Originally posted by Chris Roemer View Post
                    That woofer likes another cu.ft. to hit the mid 30s. I'd shoot for 2.5cf w/a 3"id x 7" long Precision Port.
                    It can only take 20wRMS @ Xmax @ 104dB.
                    YOUR (smaller) box can take 34w (@ 107dB - not accounting for baffle step loss), but I only see an F3 in the mid 40s w/a +3dB hump around 80.

                    To design an XO that works, you need (at a minimum) a plot of the driver's freq. response. I don't believe HiVi has any published for that 3".
                    "Textbook" (or "online") calculators don't (generally) work well at all.
                    Your 750Hz filter on that woofer will actually roll it off about an octave higher (1500 Hz) AND it leaves it with a +6dB "hump" (bad) from cone breakup right below 1kHz.

                    Your 5k filter's transfer fn on the tweeter shows it @ -3dB @ 5.5kHz (pretty close) and -6dB down @ 4k.
                    It also leaves the tweeter running about +9dB too "hot" (in relation to the woofer).
                    Thank you, this is very insightful. What simulation tool do you use? I have a Mac and it's harder for me to find good ones. I tried Jeff Bagby's Excel tool but it is glitchy and will not let me load the .frd files. Are there any others?

                    Why is there such a large discrepancy on the woofer crossover point from the "textbook" value? Is the tool you are using taking into consideration the actual impedance of the woofer at the crossover frequency or something to that effect?

                    Comment


                    • #11
                      Originally posted by Kustomize View Post
                      Most/All of your requirements are met by Paul Carmody's Tarkus

                      Within budget too!

                      F3 of 29 Hz and can pump out 106 dB!

                      https://sites.google.com/site/undefinition/tarkus

                      That's a proven and well designed 3 way. If you are not after the Klipsch sound, I don't think you'll be able to design a better 3 way speaker than what is done by experienced designers, such as Paul Carmody and others. Don't take this as an act of discouragement, if you want to try it on your own still, you really should!

                      If I were you, i'd consider the Tarkus and see the builds others have done at the bottom of the page, and see whatever suits you! Put some grills on them maybe!
                      the tarkus woofer section alone calls for a 60L enclosure...that's much larger than my space allows. I'm sure they can shake some walls though

                      Comment


                      • #12
                        Those filters you got (off the net?) WOULD work OK (not accounting for SPL discrepancies between drivers) IF all speakers had FLAT freq. resp. plots AND flat impedance curves. Fact of the matter is that NONE of them do.

                        Grab PE's .zip file for that woofer (295-315), unzip it, and look at the F/Z (fr & imp.) graphs.
                        Most woofers have an impedance that rises from a low (around 200Hz, like 6n(ohms) for a nom. 8n driver) up to a high MANY times higher up in the top octave - like maybe 40, 80, or even > 100 ohms. Your coil/cap choice COULD roll the woofer off OK if it stayed at 8n, but the more it rises, the less of an effect your filter has. You'd have to have a coil that kept gettting larger and laargeer as the frequency rose - OR ... use a "Zobel" network across the woofer's leads which can effectively flatten out that "rise".

                        Also, you've got to deal w/any FR irregularities w/the XO. MOST drivers (even very expen$ive ones) are only relatively flat over the range they're designed to be used in.

                        PCs were designed for use by engineers and scientists, Macs were for use by "artists".
                        I THINK you can run some kind of PC emulator on a Mac, then some of the speaker building tools would be avail. to you.

                        If you can't make THIS box large enough for a 10", there are some 8"ers that would do OK in 1.5 cf or so.
                        I've got data files for your tweeter and woofer. If you got a mid that also had data avail., I could design a proper XO for you.

                        Comment


                        • #13
                          Originally posted by Chris Roemer View Post
                          Those filters you got (off the net?) WOULD work OK (not accounting for SPL discrepancies between drivers) IF all speakers had FLAT freq. resp. plots AND flat impedance curves. Fact of the matter is that NONE of them do.

                          Grab PE's .zip file for that woofer (295-315), unzip it, and look at the F/Z (fr & imp.) graphs.
                          Most woofers have an impedance that rises from a low (around 200Hz, like 6n(ohms) for a nom. 8n driver) up to a high MANY times higher up in the top octave - like maybe 40, 80, or even > 100 ohms. Your coil/cap choice COULD roll the woofer off OK if it stayed at 8n, but the more it rises, the less of an effect your filter has. You'd have to have a coil that kept gettting larger and laargeer as the frequency rose - OR ... use a "Zobel" network across the woofer's leads which can effectively flatten out that "rise".

                          Also, you've got to deal w/any FR irregularities w/the XO. MOST drivers (even very expen$ive ones) are only relatively flat over the range they're designed to be used in.

                          PCs were designed for use by engineers and scientists, Macs were for use by "artists".
                          I THINK you can run some kind of PC emulator on a Mac, then some of the speaker building tools would be avail. to you.

                          If you can't make THIS box large enough for a 10", there are some 8"ers that would do OK in 1.5 cf or so.
                          I've got data files for your tweeter and woofer. If you got a mid that also had data avail., I could design a proper XO for you.
                          chris, thank you for your time and willingness to help. I think i understand what you are saying. I am an engineer (maybe not a good one) and the mac was a hand-me-down from my brother many moons ago. anyway, I got jeff bagby's tool to work on my work laptop and started playing with the low pass filter (i have not input any other data other than the woofer's .frd and .zma files) and the low pass filter info. I tried adding a parallel RC contour filter before the LPF (i'm guessing this is the same as a "zobel" network as you called out), which on the tool does flatten the peak. However I don't understand the purpose of the series capacitor in the RC contour filter...it doesn't seem to have a large impact on the overall response...is it necessary? What effect does the resistor have on sound level and what should the resistor be rated for? most audio-grade resistors seem to be ~10W...is this enough for this application? I tried to use inductors, caps, and resitors that are readily available and low-cost. The crossover point shifted to ~500hz but i think that is okay if an adequate mid-range can pick it up from there. please let me know what you think

                          Click image for larger version  Name:	dc250-8.JPG Views:	0 Size:	152.8 KB ID:	1428723

                          can you please share the peerless tweeter frd and zma files? i couldn't find them. I would like to play around with that filter as well and I'll input more data into this file once i have more time. if you have a mid-woofer that can be purchased for <$15 and has readily available .frd and .zma files, i'm open to substituting the HiVi out of the design. do you have any tools you can recommend for the enclosure and port tuning? the values you cited earlier in regards to F3 and peak are significantly different than what my basic excel tool is telling me (not sure i remember where i got it from; i can double check). if you'd like to take a stab at creating the XO for me that would be fantastic. I just want to make sure we understand the top priorities:
                          1. overall cost
                          2. physical size (limited by the geometry of the heresy III...15.5"W x 21.5"H x 12"D outside dimensions)
                          3. using the 10" woofers and peerless tweeters i already have
                          4. provide a reasonably low F3 with the 10" woofer in a 40L enclosure, i estimated the rest of the internal volume will be taken up by port, drivers, bracing, etc.

                          everything else is pretty flexible. i'm not afraid of putting together complicated crossovers, however i would like to avoid expensive XO components (see priority #1). remember, i just want a set of speakers that will sound "decent" ...not ones to compete with.

                          Comment


                          • #14
                            Off you go. (I use PE's part# to name my F/Z files - mostly).

                            264-500.FRD
                            264-500.ZMA

                            All you have to do is grab FPrawn's "FPTrace", grab a screenshot of the drivers PDF file, and have at it.
                            These files have NO phase data (phase is there, it's all just set to zero). You normally generate min phase in your process.

                            I could run a sim using the Peerless TC9FD18-08, okay?
                            Attached Files
                            Last edited by Chris Roemer; 01-09-2020, 03:19 AM.

                            Comment


                            • #15
                              thank you! i've been doing some research and decided to get VituixCAD & windows running on dual boot (i've held out long enough). I must say...i'm pretty hooked on this tool and it has an SPL tracer that works pretty similarly to the tool you suggested. I also rummaged through my garage to look at what i have on hand. I found quite a few drivers left over from failed attempts :( Here are some options I already have:

                              Dayton PC68-8
                              Dayton PC83-4
                              Dayton ND105-4
                              Dayton TCP115-4 (i have a whopping 10 of them)

                              I took a stab at using the PC83-4 since the last two are a bit too large and feel like would be a waste in a 3-way. From what i've read, they can pump out quite a bit of bass on their own so I'd rather save them for a 2-way project but that could be the hoarder in me talking. Anyway, here's what i put together after some tinkering:

                              Click image for larger version

Name:	3-way XO.JPG
Views:	159
Size:	21.6 KB
ID:	1429018
                              Click image for larger version

Name:	3-way FR.JPG
Views:	166
Size:	175.2 KB
ID:	1429019

                              there are quite a few peaks & dips in the summed response but I don't have enough experience to know how discernible they are to the ear in real life. On paper, the response seems pretty flat to me (+/-3db for the majority of the spectrum). Let me know if you see any red flags or where there's room for improvement. I'll keep adding info to the model such as baffle & enclosure dimensions but im not there yet...

                              I haven't really added up how much the XO are going to cost but it looks like in the $30 range assuming I use electorlytic caps.

                              Attached Files

                              Comment

                              Working...
                              X