2Way Bookshelf Project - 1st Project

[biyer]

Hi All.

I just completed my first diy 2-way bookshelf speaker project. Please could you share your opinion and what I can do better next time.
I would like to thank FM @arunkvivek for guidance provided.
Have been looking for a project to build for some time. I got an idea after I came across the post in ASR.

Disassembly of psb Alpha B1 bookshelf speakers

It's cold and windy outside, I'm bored, my hands are itching to do something, so here I am. The psb Alpha B1 are my first "decent" speakers, purchased at a retail price of $299 CAD/pr. Needless to say that I have long outgrown them in the ten years since, and I'm fed up with trying to offload...

www.audiosciencereview.com

The above driver is still being manufactured in India (by Peerless Audio which is currently not related to Peerless Fabrikkerne but bought over assets of peerless's Indian subsidiary)

Peerless SM13-PCT-8 5.25" Hi-Fi Woofer Titanium Coated PP​

Peerless SM13-PCT-8 5.25" Hi-Fi Woofer Titanium Coated PP Overview: This 5.25 inch 8 ohm driver is a member of the high performance HDA (High Definiti

diyaudiocart.com

Then came found another tweeter from peerless which was apparently used by the Energy Connoisseur series.

https://electrocarts.in/product/25mm-silk-dome-tweeter-low-fs-neo-system/

That's how project code name Alpha/Energy was conceived.


1. I started by measuring the TS parameters myself by using test jig (https://audiojudgement.com/measure-thiele-small-parameters-using-free-software/)

2. Then designed enclosure in WinIsd . The volume is approx. 9.14 liters with port diameter 2.25 inches, port length 6.2 inches tuned to 65hz

Enclosure was built using 18mm HDF wood which was scraps found at local CNC shop.

Dimensions are 7.5W x 10D x 13H inches.

3. Measured the Drivers in box and captured FRD and ZMA. I also captured the woofer in different axis's to find cone breakup. Used Dayton imm-6 mic.
4. Used xsim to design crossover.

5. Put everything together and measured the response (All measurements are with 1/48th smoothing and 5db step size)

On axis 30.40.60 and 90 degree response

Near Field port (RED) , Woofer (BLUE) and Tweeter (GREEN)

the cabinet/ port resonance around 800hz needs to be addressed. Also need to measure with mic closer next time.

Pics

Finished Painting using Granotone Speaker Paint.

The Total Cost of project was under 17000 INR

Overall I am very happy with the project. They sound better than my Wharfedale Diamond 9.2. I am caught up with some family responsibilities will add additional measurements once I find time.

I am currently powering these using a TDA1521 board based audio amplifier which I built for testing.

 

[rwnano]

Good job Sir

 

[sadik]

Good Job... Welcome to the DIY world.

 

[arunvenkats]

Nice! Congratulations

 

[arunkvivek]

@biyer - excellent finish. Waiting for a listening session! Dainty subwoofer next! Also post some pics of the testing amp!

 

[Vineethkumar01]

Hi @biyer
For a first DIY project, you have done really good.. Congrats for that..

But since you asked about what can be done better, here are some suggestions based on different aspects of the build:

1) Midbass driver bass alignment

If your measured T/S parameters are accurate, this is a relatively high Qts midbass, which might like a closed alignment better than a bass reflex. This is probably evident in that 4dB hump around 100Hz in your alignment sims above, when I model the same in VituixCAD, I get similar response curves with probably not accurate assumptions on the kind of damping you have used inside cabinet






Note that you are close to running out of excursion around 40Hz with a power exceeding 5W into the woofer. Also, even though generally 4dB humps in bass sounds very nice in the beginning, it may become too much and fatiguing over time. So generally, people go for 1-2 dB bumps in the bass. Again all depends on your room situation as well. If there are room modes killing everything near 100Hz, you may/may noe percieve the differences.

Now, if you had used a closed box alignment and used a series capacitor on the woofer of value around 370uF (good quality electrolytic caps are fine I think), you could get a much more balanced response like below and push more than 10W of power before excursion limits come into play. You can have some DC protection on the woofer as well

 

[Vineethkumar01]

2) Speaker box shape & baffle step compensation
You have not stated anything about baffle step compensation above. I assume that you have taken it into consideration.. Even if not, your bass hump and the series inductor of 2mH might play a role in flattening the response eventually. But it is always take its effect into consideration and then tune the overall response.

My simulations show the following profile for baffle diffraction based on your baffle dimensions. This is pretty significant and good to account for in the crossover sims



You seem to have a mild waveguide on the tweeter and a phase shield (uasually helps with the aluminium dome break-up at the top, but I dont know why they have used it if it is a silk dome tweeter). While the waveguide might help you in managing diffraction to some extent, having rounded/chamfered edges starting immediately at the tweeter edges is among the best options to terminate a waveguide on a baffle at least relatively properly. In general, shoe box shaped boxes are never the best shapes for speakers. But sometimes it is all we can do in a relatively low cost build

If you have been wondering where that bump in tweeter response around 4kHz is coming from, it could probably be from the diffraction response of your baffle to your tweeter. Below is an example of the diffraction sim for the tweeter based on dimensions you have given





Among the better options to use this tweeter in a baffle may be a baffle shape as shown below

 

[Vineethkumar01]

3) Frequency response measurements & sound balancing via crossover

I really appreciate you take off axis measurements on a first build.


However, it does not confirm much by itself other than that there is some diffraction impact or crossover impact around 2.5kHz. I would suggest that you learn to take proper polar measurements in future and use VituixCAD fr processing. It will give you much more insight into how sound energy is radiated into 3D space. For example, here is a sample plot and the kind of information it can give




You can see the state of axial response and other associated curves such as Listening window response and power response.
The further away from the speaker you are while listening, one of these curves might become more significant in telling how much balanced the speaker response is. For nearfield listening, the axial response might dominate perception, while if you sit about 1 or 2 m away, the listening window response and power response might have their say. Accordingly, you can optimize the response in the crossover based your goals

 

[Vineethkumar01]

4) Monitoring the impulse response/timing aspects in crossover design

Your speaker being a small 2 way speaker has less potential to screw up timing of the speaker. But on a more elaborate build/speaker concept, it is always good to have a check on timing aspects via plots showing ETC/RMS of the step response as well, while striving for a smooth set of frequency response curves. While the ASR (audiosciencereview) crowd says spinorama is the be all, end all of speaker design, more experienced designers who have actually designed hundreds of speakers repeatedly say about the importance of not underestimating the timing aspects along with smooth frequency responses.
Here is an example set of timing-related curves for the speaker responses I posted above.


Also good to keep an eye on the excess group delay around 100Hz and trying to not exceed about 2.5ms at 100Hz


In fact, there are more discussions and examples of how good/bad timing aspects can look like

VituixCAD v2 - HTGuide Forum

For information, VituixCAD speaker simulator has been updated from version 1.1 to 2.0 with the following additions and changes (so far): Crossover schematic changed from formal block ladder to free form supporting unconventional topologies. Predefined passive blocks replaced with individual...

www.htguide.com

(The above thread is a treasure trove of speaker design information from one of the most experienced speaker designers I have ever known. So learn everything from it, while you are at it.. )

 

[Vineethkumar01]

End of the rant..
Hopefully, it is useful to someone..
In general, start using the latest tools (VituixCAD at the moment, compared to XSIM & Jeff's PCD) if one might not want to be limited by the handicaps of the design tools. Learn the proper way to take measurements and use the equipment required. And then have a go at it.
It is 2025.. The time for single-axis frequency response optimisation-based speaker design is long gone..

 

[biyer]

Hi @Vineethkumar01 , Many thanks for the detailed response. Really appreciate you taking time for giving such a detailed response.
I agree with all of your points.

The TS parameter measurements are relatively accurate tested them over 5-6 driver whose ts parameters are known. Didn't get exact values but were in close ballpark to help with enclosure design.

Initially wanted to use Virtuix cad but due to commitments and to wrap up things quickly for the time being shifted to XSIM and then in future redesign if needed.

Original plan was to take off axis measurements of both drivers and then use them for crossover design in Virtuix Cad.

I started with the idea of replicating the PSB alpha B1 for the enclosure so never considered sealed enclosure. A lesson for me to not neglect other enclosure types in the future while designing. I was also looking at similar alternative drivers like

HiVi L5-8R 5" Woven Kevlar Cone Woofer 8Ohms - Pair

HiVi L5-8R 5" Woven Kevlar Cone Woofer 8Ohms - Pair Overview: Rigid woven kevlar cone Long throw linear excursion design Hi-Vi Symmetric Mot

diyaudiocart.com

Dayton Audio DC130A-8 5-1/4" Classic Woofer Speaker

Dayton Audio DC130A-8 5-1/4" Classic Woofer SpeakerOne of the main design goals at Dayton Audio is to create loudspeakers that can achieve high levels of performance at reasonable prices. The Dayton Audio DC130A-8 5-1/4" woofer is a great example of this approach. It features a non-pressed...

www.parts-express.com

the above were touted as replacement drivers for PSB alpha B1.



1) Midbass driver bass alignment

Regarding the 4db hump I took some inspiration from LS3/5A and Diamond 9.1 , Also the extended plan was to construct a subwoofer and cross it around 80-100 hz using diy opamp based analog HPF and LPF which also takes care of cone excursion.

Read somewhere that excursion below the response spec of driver doesn't matter. But I will rather take caution and add a series Capacitor as an HPF before the woofer.





But I really liked your idea of a series Capacitor I am planning to add one, would like your opinion on which I can go.


Let me discuss point no 4 before others

4) Monitoring the impulse response/timing aspects in crossover design

1200 usec (1.2ms ?) is group Delay value in XSIM in hte current crossover config.



I am considering 3 options for the HPF cap before the woofer.

Option 1.

Took Nominal Impudence or Rdc as 6.5ohm.

250 uF https://diyaudiocart.com/product/bevenbi-250uf-100v-electrolytic-non-polarized-crossover-capacitor/

This has a cut off of around 100hz (considering ) which smoothens the response (max 1.5db) and dampens the excursion to below max of 4.5mm when provided 25W power . (max amplifier power I am aiming for but power handling should be higher due to other losses and tweeter.)






But I get a group delay of around 2900 usec (2.9 mS)




Option 2:

400 uF Capacitor with a Fc of around 63hz but dampens poorly max 2.5 db



Also cone excursion is still a little high




But group delay is improved with max 2390ish usec (2.4 ms)




Do you think its a good idea to add one?
Below is the ETC





I have also found the z-offset timing using PCD which I used in XSIM.

Now onto point 3

3) Frequency response measurements & sound balancing via crossover

Yes the crossover is at 2.4Khz. I measured the off axis response so that I can create polar map etc at a later time.

Using this link as guide https://www.audiosciencereview.com/...ents-spinoramas-with-rew-and-vituixcad.21860/


2) Speaker box shape & baffle step compensation

I originally wanted to round the edges but again had to wrap up. I will try this at a later time.

For baffle step compensation I thought I can handle it with FRD measurements in crossover design software.

 

[biyer]

@biyer - excellent finish. Waiting for a listening session! Dainty subwoofer next! Also post some pics of the testing amp!

Hi Arun . I have used TDA1521 based boath which provides updo 12W of power with 0.5% distortion.

The IC requires minimum external components so I thought the changes of screwing this will be minimal.

I used the below board which comes with a rectifier circuit builtin. Only think I needed the manage some grounding to handle hums. But now its silent.

Just note nobody should run an amplifier without a fuse at least after the transformer. I will add fuse later.

https://robu.in/product/tda1521-audio-power-amplifier-board-module-2-0-dual-channel-215w-with-rectifier-filter-circuit-diy-audio-speaker-modification/

 

[biyer]

Hi Arun . I have used TDA1521 based boath which provides updo 12W of power with 0.5% distortion.

The IC requires minimum external components so I thought the changes of screwing this will be minimal.

I used the below board which comes with a rectifier circuit builtin. Only think I needed the manage some grounding to handle hums. But now its silent.

https://robu.in/product/tda1521-audio-power-amplifier-board-module-2-0-dual-channel-215w-with-rectifier-filter-circuit-diy-audio-speaker-modification/

The cabinet is made of https://www.amazon.in/APPUCOCO-Cabl...pcontext&ref_=fplfs&psc=1&smid=A1HDE05ESXIWOC

 

[Vineethkumar01]

Hi @biyer

Regarding the bass alignment, I was suggesting a series cap for a sealed box alignment for the midbass. Not the ported version that you have now.
With a high Qts woofer tuned with Q > 0.7, the series cap will interact with the impedance to flatten the bump and extend the response to low frequencies. It is a very old trick that has been in use from the 70s. I hear that some of the philips high Q woofer used the trick to extend bass a little bit more in a sealed box. (Here is an example scenario where this trick is being used: https://www.humblehomemadehifi.com/download/Humble Homemade Hifi_Black Box.pdf)

With the ported alignment, the high pass will just act as a high pass filter, which may be helpful with subwoofer integration but not for the above-mentioned purpose.

Regarding group delay, I think your measurement has not been centered to impulse peak in REW, hence the large group delays I think. Not sure though. Also, you need to merge nearfield and farfield response along with baffle step to get actual response
Also excess group delay measurement/graph is what you want to look at and not the group delay itself. These are available in proper places (impulse view) in VituixCAD. So, no need to use REW for it..

If you are willing, you can post your measurement REW .mdat files here and I can take a look.

 

[arunkvivek]

View attachment 93092View attachment 93093View attachment 93094
View attachment 93095

The cabinet is made of https://www.amazon.in/APPUCOCO-Cabl...pcontext&ref_=fplfs&psc=1&smid=A1HDE05ESXIWOC

Great jugaad of wire management bin!