HiVi/Swans DIY 2.2A (crossover modification/upgrade)

(An update of this OP is available as post #99)

Hello everyone,

I’d like to share with you my latest project, namely building a pair of speakers from a commercial kit - Swans DIY 2.2A and the modification that I’ve come up with. As the name suggests it’s one of the two official kits from Swans and you can find more details about them here: swans diy speaker kits. As Swans Speakers appears to be a division of HiVi Research you can find the same kit under the name HiVi DIY 2.2A instead. Sometimes the cabinets and the components (including the drivers) are sold separately and not as a complete kit (depending on the vendor).

Being a commercial kit, these speakers are not very well documented in terms of technical data and design. That’s understandable - building speakers from a kit is supposed to be a straightforward process where we don’t really need a technical drawing of the cabinets and the schematic for the crossover network. This is the very nature of any kit after all. That being said, on the website above you can find a product manual with instructions on how to build the speakers and some graphs like frequency response and impedance, which appear to be from measurements.

The kit is basically a two-way system employing the following drivers: HiVi D6.8-IIB mid-woofer and HiVi Q1R tweeter. While the latter is easily recognisable as it’s part of the official range of HiVi drivers, the former is a bit of a mystery because there isn’t any official information, let alone T/S parameters, that can be found. It seems that it is some kind of variation of D6.8B which looks identical, but apart from that it’s not very clear. There are some speculations on various audio forums that D6.8-IIB drivers are basically D6.8B not meeting the quality requirements of the manufacturer and they are getting rid of them trough the DIY market as part of a kit.

Anyhow, I’ve decided to share my two cents and hopefully spark a discussion as I find these speakers holding a great potential, but only if you don’t use the crossover that comes with the kit. Despite the many positive reviews I’ve read, there are some people who were not happy with their purchase. In particular, they find the treble too “hot” or “bright” and I have to admit that they have a point. Others think that the bass is weak and I was not impressed either. The issue with the treble from the negative reviews is confirmed by my measurements. The problem is that the frequency response is tilted towards the upper end with a steady increase in the amplitude from around 9-10 kHz to 15 kHz where it gets even worse with a huge hump in the area above 15 kHz. As for the poor bass, part of the reason could be the resistance of the 1.5 mH coil in the low pass filter which is 1.1 ohms! In my opinion, this is unacceptable in a good crossover design, especially when we are dealing with not very sensitive drivers.

In the end I developed a crossover network based on a already existing design involving the same tweeter (Q1R) and D6.8B in almost the same enclosure (slightly different volume/dimensions, but the same baffle width). In addition to that I’ve included an impedance compensation network to cope with the tweeter’s overly enthusiastic response in the top end. As a final touch, I included a LCR network, which is operating as a notch filter aiming at the resonant frequency of the tweeter.

You can find the details of this project in the file below. Hopefully, other people can contribute to the discussion and share their experience with this speaker kit. I would love to hear whether you find the modification as something that is worth it or not. Maybe someone can suggest another crossover solution. Mine is not spectacular and it’s definitely not a benchmark for a good crossover design, I’m aware of that. Not to mention that there is a slight phase shift as well, but I don’t believe it’s audible. On the other hand, there was the same phase shift with the original crossover, so it’s not like I made it worse with my modification. On the contrary, I definitely think that the sound is better now with the suggested crossover upgrade.

- @llex
 

Attachments

  • HiVi Swans DIY2.2A - Technical Drawing & Crossover.pdf
    117.1 KB · Views: 1,221
Last edited:
  • Like
Reactions: 1 user
More info about the project and some pictures.

After my initial post about the modification of HiVi DIY2.2A, I’ve been thinking to elaborate further and to provide more details about the project, including some links and a measurement. Hopefully, this way I can shed more light on the matter for those who want to build these speakers with or without the modification.

A brief history of the project:
I’ve been interested in a variation of this speaker kit for a very long time - an unofficial kit which is supposed to be a replica of Swans D2.1SE. I once had the opportunity to audition these speakers on a High End show and I really enjoyed their performance, but financially speaking they were out of my league. However, the Swans dealer who offers them also sells a kit with all the parts that you need (including the drivers) to build a clone of the original model. You can find more details about this kit here: Exproject
- Тонколони - Направи си сам - Swans - D2.1SE+ KIT
. According to the website the crossover provided with this kit is an exact copy of the one in D2.1SE. My crossover modification, which I suggested in my previous post, is based on this one, but with an additional work of my own to tame the emphasis in the high frequency range. More on that later.

At some point I made a decision that in foreseeable future I would like to get my hands on this kit. Then alI I’ll need is to build the cabinets and I am ready to rock. Unfortunately, I’m currently residing in Denmark where almost everything (especially the services) costs an arm and a leg plus two skins and a kidney. Definitely not a place to order custom made speaker cabinets. Out of curiosity I went to a joinery here in Copenhagen and they quoted me a price of 400 EUR (~490 USD) just for a pair of speaker baffles, not for the whole cabinets. So, I was quite happy to see last December that the website where I usually buy components started selling the official HiVi kit. On top of that, all speaker kits were on a discount of 15%. There was even an option to buy the cabinets (which are assembled by the way) separately from the component set. Pure bliss for the average DIY-er! As you can imagine, I’ve got the complete kit. For those of you who are in Europe I strongly recommend this site: SoundImports: DIY audio components and accessories - SoundImports

The cabinets:
The build quality is quite good, they’re braced and the veneer is applied smoothly. The finish might look cheap to some people, but it’s all about taste. I’m not that picky. The drivers fit perfectly and even the holes for the screws match the driver’s openings. The recess for the tweeter seems fine with a minor variation of the outer edge on one of the cabinets - not a big deal. To sum it up, the cabinets are not exactly perfect, but very close. Given the price, I simply couldn’t expect more.

The parts:
The component set includes everything you need to finish the speakers - drivers, crossovers, ports, binding posts, poly-fill, etc. I already discussed the drivers in my first post and can’t think of anything else to add. The ports are shorter and with smaller diameter than the ones from the unofficial kit that I mentioned earlier. However, my WinISD simulation using the dimensions of the ports confirmed the manufacturer’s claim of F3 = 43Hz in this particular volume. You can see a screenshot of the simulation on the first picture below. The second picture is showing the port’s air velocity, which doesn’t seem theoretically great as the graph is based on only 15W of input power. I cannot give you any feedback whether the ports are chuffing or not as I haven’t cranked up the volume to test it yet. Sorry, we are still under pandemic lockdown here and I don’t want to “entertain” my girlfriend with very loud music while she is in a Zoom meeting for instance :)

The crossover:
At first glance, the low-pass filter in the original crossover (picture 3) seems a bit odd. It looks as if it’s 1st order with impedance compensation network, but it’s actually quasi-second order. We can see the same structure in the crossover from the D2.1SE clone (picture 4), but with a different inductor (2.6 mH). In my opinion, this is a well working design for these drivers and after some experiments and simulations I couldn’t come up with a better solution. I chose the second crossover variant over the first one by comparing them sonically, but even then I was not completely satisfied with the sound. I could still feel some extra energy in the high frequencies and my measurements only confirmed that. I added an impedance compensation network in the high-pass filter, which smoothed down the tilted response of the tweeter, including a nasty bump in the top end. I also added LCR network to address the hump in the rolloff section of the tweeter’s response. It’s basically a broad notch filter aimed at the resonant frequency. Its effect is hardly audible, but it’s an additional safety for the tweeter as it dampens the resonant frequency even more. It also helps to bring back the phase response to what it was before introducing these two networks. As I mentioned before, the phase alignment is not great, but in my opinion it’s fairly reasonable. On picture 5 you can take a look at the crossover simulation, where the phase shift is also visible.

In conclusion:
A measurement of the frequency response of the system with the modified crossover is shown on picture 6. Unfortunately, I don’t keep the measurements made with the original crossover and I can’t demonstrate the difference between the two. On axis measurement doesn’t tell the whole story anyway and describing the sound of a particular set of speakers is futile. What is “bright” for me could be perfectly fine for others; a good bass by my standards might be too weak for your taste.
However, there is something interesting revealed by this measurement - the dip between ca. 250 and ca. 700 Hz., which appears to be a baffle step. The online Baffle Diffraction Loss calculator that I usually use suggests a baffle step with -3dB at 563 Hz. This seems to be consistent with the graph, but I’m not very experienced and can’t tell for sure that this is definitely a baffle step. What makes me doubting is the response curve of the woofer (D6.8B) from the manufacturer’s data sheet (the last picture below) where we can see almost the same dip in its response. My assumption is that this is an inherent feature of the driver’s response and I decided to leave it as it is. My listening sessions during the crossover design didn’t give me any reasons to pursue an attempt to tackle it as an issue.
One more interesting observation is the increased area at 42Hz, just before the rolloff, which seems to confirm the F3 claimed by the manufacturer.

-@llex
 

Attachments

  • 01. HiVi DIY2.2A - Transfer Function (WinISD simulation).jpg
    01. HiVi DIY2.2A - Transfer Function (WinISD simulation).jpg
    215.8 KB · Views: 1,289
  • 02. HiVi DIY2.2A - Port Air Velocity (WinISD simulation).jpg
    02. HiVi DIY2.2A - Port Air Velocity (WinISD simulation).jpg
    118.8 KB · Views: 1,453
  • 03. HiVi DIY2.2A - Original Crossover Network.jpg
    03. HiVi DIY2.2A - Original Crossover Network.jpg
    18 KB · Views: 1,641
  • 04. Swans D2.1SE Speaker Kit - Crossover Network.jpg
    04. Swans D2.1SE Speaker Kit - Crossover Network.jpg
    52.5 KB · Views: 1,589
  • 05. HiVi DIY2.2A - FR & Phase (XSim simulation).jpg
    05. HiVi DIY2.2A - FR & Phase (XSim simulation).jpg
    152.6 KB · Views: 1,509
  • 06. HiVi DIY2.2A (modified) - Frequency Response.jpg
    06. HiVi DIY2.2A (modified) - Frequency Response.jpg
    101.3 KB · Views: 690
  • 07. HiVi D6.8B - SPL vs Freq.png
    07. HiVi D6.8B - SPL vs Freq.png
    46.8 KB · Views: 604
  • Like
Reactions: 1 user
Interesting. I just bought the 2.2A kit, and there’s not nearly as much info on it as the 3.1. What is the effect of just replacing (just) the 1.5mh cap? Even if replacing with the same value for mh, like you say, 1.1 ohms seems a bit excessive. I guess that would impact the bass output, as you say.

I guess I could try to run a sim, but 2.5mh is going to lower the crossover point. Why doesn’t it lower the tweeter? Seems to raise the crossover point in the tweeter as well? Not sure about lowering the crossover being a good idea, unless the woofer is playing into where it resonates, but I’m no expert on crossovers.

Anyway, I’m unconcerned with peaks at 15k, given my hearing (it’s probably a benefit), so if I did mod anything, I’d want to only make modest changes, I think.

I guess what I’m trying to say is, I’d be interested in more modest mods, given my lack of comfort in working with crossovers and not being sure that all of that effort around the tweeter would be a benefit for me. Plus wanting to most out of the existing parts. Is the Swans 2.1SE crossover a reasonable upgrade by itself? In theory, I could just buy two parts... maybe 3, if including the cap to go from 8mf to 8.2, but not sure if that value change is significant.
 
Edit: found discussion which suggests that decreasing the DCR affects the midrange, not lower bass, more. Maybe it wasn’t the DCR change that helped the bass for you as much as lowering the crossover point. Maybe it makes it more like the BSC you were wondering about? I should just stick with the stock crossover and measure it and see what it’s like.
Does it improve the bass response if I replace the lower DCR woofer inductor
 
Hi Vidgamer,
Yes, you’re right - the 3.1A kit is much more popular and quite often discussions about the 2.2A kit are highjacked and end up with people talking only about the 3.1A kit instead.

As for the crossover, my modification is based on the disappointment from the 2.2A speaker kit and its original crossover network. Maybe I should have given them more burn-in time, but I doubt it would make a significant difference. The first thing I did was to change the crossover with the one from the D2.1SE clone and it was a noticeable improvement in terms of more lively and open sound. The difference between the baffle width of 2.2A kit and the D2.1SE kit is only 1 mm, so I was not concerned about using a crossover from “different” speakers if you know what I mean.

Your observations about the crossover modification are correct - the woofer seems to be crossed lower (somewhere in the vicinity of 2000 Hz) and the tweeter higher (around 3000 Hz). At least, these are my calculations based on the measured impedance of the drivers, but I’d like to draw your attention to the picture below where you can see the XSim simulation for the electrical response. I don’t know if that’s helpful or even more confusing, but what I can see there is a crossover point of around 2600 Hz.

All of this reminds me of another HiVi kit employing the same mid/bass driver (D6.8B), but different tweeter. In the pdf file below you can see the details about this kit and you’ll notice a similar thing - woofer crossed at 2000 Hz and the tweeter at 2500 Hz.

That being said, I understand your concerns about the modification and the only thing I can suggest is to build the speakers and assemble everything except the crossover. Instead, connect a wire to each driver and take it out trough the bass port and then make the connection with the crossover outside the cabinets. Then connect the speakers to your amplifier and try to live with them for at least a couple of weeks (maybe even more). Don’t rush into any conclusions about the sound before that. If you can take measurements, that would be great. Compare your impressions of the speakers with the measurements and you’ll have a solid ground for an evaluation. Meanwhile you can experiment with a better coil for the low-pass filter - same inductance, but lower resistance. That would be a good starting point.

You might end up without the need of a modification after all, who knows...

- @llex
 

Attachments

  • hi-vi research - melody kit.pdf
    177 KB · Views: 312
  • hivi diy2.2a - electrical response.jpg
    hivi diy2.2a - electrical response.jpg
    135.6 KB · Views: 285
Ok, I got the kit, and have ordered a measurement mic, so I’ll try to test. Will probably go with stock. I might break it in a bit. I know break-in usually sounds like a scam, but there are enough posts about some woofers needing some break-in that I’m inclined to give it a bit of time. But how much time? Most breakin surely occurs within minutes, or maybe hours. I’m a bit suspicious of claims that some speakers need 100 hours! I’ll give it 2 hours. ;-)

If I had to guess, the alternative crossover sounded more lively due to maybe having a slightly more V curve to the frequencies. Which is not a bad thing, it’s just down to personal preference. I did something similar in modifying my old bookshelf speakers. I had trouble hearing the tweeter, so added a baffle-step-compensation type of circuit. I’m pretty sure it’s “less accurate” the way I have it, but it sounds good to me, or at least more interesting.

So, my strategy is to measure the stock crossover, and listen to see if I like it. Based on where I see any deficiencies, I’ll model the crossover in a circuit simulator to see what any changes do. With modeling, that change in the resistance should show up, not just the inductance.

A lot of fixes to the 3.1 crossover seem to be for having a peaky tweeter, but I’d like to confirm that’s a problem for 2.2a before getting too crazy. I might like it a bit peaky. :)

Thanks for your comments. Always interesting to see what others are trying!
 
I should mention... the BSC wasn’t the best improvement to the crossover. I also added a Zobel to the woofers and that seemed to tame the harshness on the upper-end of the woofer. The response graph looked pretty reasonable, I think, with that added. I got the idea from reading an article about designing crossovers. So, that might be something else to consider. So, if there’s a peak at the upper end of the woofer, that might be the first thing I’d change. Well, along with that inductor. But I need to measure to be sure what’s going on. And build the things. Might take me a while. :)
 
Thanks for your comments. Always interesting to see what others are trying!

You are welcome. At least now you have an alternative option to choose from when it comes to the crossover design. As for the break-in, that was not entirely the premiss in my previous post when I suggested to wait a few weeks in order to decide about the need of modification. I’m not entirely convinced that breaking the speakers in would bring a significant change in the way they sound. What I meant was to give them some time to reveal themselves in full - by playing different genres of music for instance. Our perception of sound is affected by a lot of factors, some of which psychological. The way we judge about something, including sound, could be a function of our current mood as well. Listening to a new pair of speakers is also tricky, because we are comparing them unconsciously with what we are used to, which is not necessarily a good thing.

Below you can find the files from my measurements that I used for the simulations. You might be interested to use them until you get your microphone. The file “HiVi D6.8B-IIB & HiVi Q1R.frd” represents a measurement from both drivers in the speaker cabinet without any crossover. I used this file to determine the relative delay between the drivers - something that is quite often neglected. The ZMA files are taken with DATS v3 as a “free-air measurement”. I hope that would be of help.

-@llex
 

Attachments

  • HiVi DIY 2.2A (FRD & ZMA files).zip
    37.2 KB · Views: 283
Did a test-run with one speaker, with things not fastened down. Wow, I think these will be great! Reviews have suggested that the DIY 2.2 is the same as a $800 Swan speaker, and maybe competes with more expensive speakers. Initial listening, things were very clear, but maybe a bit bright. I didn’t have it sealed up enough to judge the bass. Will do more detailed testing when completed. For now, I need to finish the cabinet- I plan to paint. Then start the 2nd box. (I wanted to do 1 box now and see if there were gotchas.))
 
Ok, one speaker completed. Didn’t have much time tonight, so just listened to a few things, and, yeah, sounds bright. I didn’t want to get into test tones just yet, but I do have that calibrated mic, now, so I just used it on some music. Worried that it was weak in the bass, I grabbed the nearest bass-heavy CD near me - Leftfield. Sounds like I hear the bass hits, so it seems to be getting low.

See attached. The noise floor was about 20db in the bass region, down to near 10 for treble. I was trying to keep it quiet to not disturb the household. It doesn’t look overly peaky, at least not obviously so. I mean, it’s playing music, not a test tone, so it’s going to vary, but I mostly wanted to confirm that it’s able to play low bass and that the high frequencies weren’t too crazy. Will test with test tones soon.

I think painting the speakers was the hardest part of this project. Took me all weekend, and it’s imperfect, but good enough.
 

Attachments

  • E0E08FE3-06FC-4B80-93E8-57E2368E1D5A.png
    E0E08FE3-06FC-4B80-93E8-57E2368E1D5A.png
    467.9 KB · Views: 407
Hi first time to post here. I am interested in building the Hivi 2.2A Kit. My question is will it sound much better if i use the Jantzen Z Cap superior vs the Hivi caps? I have no experience in building a different x-over for the kit like the OP has designed but i can replace the caps easily.

Thanks
Joey
 
.... My question is will it sound much better if i use the Jantzen Z Cap superior vs the Hivi caps? I have no experience in building a different x-over for the kit like the OP has designed but ....

The HiVi caps look good to me. I wouldn’t bother changing them for the same value. The resistors seem like cheap ones - maybe they’d be worth upgrading, but I’m not excited about changing those either.

We’ve been discussing upgrading the inductor for the woofer. If you go with one that has less resistance, it will modify the midrange, not just the bass, so a different value is needed for mH. This should make a big difference to the sound. The question is, would it be better or worse? This is the part I’m stuck on! I might have to try it to decide.
 
Ok, I tested with pink noise from an old test CD. It seems like the area from about 1khz to 2khz is a bit raised, and the lower mids to bass is a tad down, until this big boom at 50hz. It doesn’t look like much, but if it averages, say, 3db down, that’s noticeable.

Based on this, that inductor mod is starting to look good. I was pricing them and the cheaper 18awg ones aren’t a whole lot better in DCR. The thicker wire ones are pricey - $20, $30... but the DCR is noticeably less. But this might explain why this kit comes with what it has - there’s no room for $60 worth of inductors in a kit that already is super cheap!
 

Attachments

  • 2994E311-A428-48DE-A523-1A93F77D6D16.png
    2994E311-A428-48DE-A523-1A93F77D6D16.png
    464.7 KB · Views: 407
Joey, I was thinking, if you went with the Swans 2.1SE crossover mod, it would be a good excuse to do your cap upgrade. At a minimum, for each crossover, it would need a 4.7uf cap and 2.6mH inductor, I would think. And just use the stock parts otherwise.

Allex, I’m having trouble finding 2.6 or 2.7mH inductors. I’m thinking maybe a 2.5 would be good enough, combined with a slightly higher DCR (than your 2.7 above). Probably would work out about the same. Starting to be a pricey experiment, particularly when I’m trying to save money on a DIY kit! �� Still, a very affordable kit, so no worries. YOLO
 
While I was hesitant to rush into mods at first, after testing with the stock crossover, it seemed apparent to me that Allex was right about the bass being weak. The bass was there, it just made the speakers sound bright or thin, I guess. Also, it seemed like the efficiency (loudness) was reduced compared to a different speaker, and I really wondered how much that 1.1 ohms was dragging it down! Now that I was looking for the bass, I wasn’t going to be happy until I fixed that. As mentioned before, I decided to go with the 2.1SE crossover that Allex provided, particularly after Allex incorporated that into his redesign and seemed happy with that part of it. While it seemed like a good idea to play with a simulation, I got impatient and didn’t have the time recently to set one up, and went for it.

Here’s a graph with the changes. Indeed, the bass is now brought up to match the midrange. It looks pretty good to me.

There is still this dip at 3k. I also tried it with the original 6.8 mfd caps, only changing the woofer’s inductor, and that dip was still there. I couldn’t tell much difference between the two, really, but decided to stick with the Swans 2.1SE design, and ended using the new caps. Will listen for a while, but happy to have the missing bass.
 

Attachments

  • 97DF8257-819F-41C6-8358-3F76BC83DD94.png
    97DF8257-819F-41C6-8358-3F76BC83DD94.png
    450.6 KB · Views: 350
Hello again,
Sorry for the delayed response and not answering some questions, but my focus has shifted significantly these days, as I’m looking for a job and trying to learn new things. Anyway, glad to see that other people have similar experience with the stock crossover and my suggestion was helpful.

...will it sound much better if i use the Jantzen Z Cap superior vs the Hivi caps?

Joey, in my opinion Jantzen Superior Z Cap in series with the tweeter is a good choice in therms of better detail and less “grainy” sound of the high frequencies. However, it wouldn’t provide a significant, like night and day, improvement of the sound. I would suggest to avoid using expensive caps until you’re sure that a certain design is THE crossover. Think of it as a final touch, using it only when you decide you’re happy with the overall sound of the speakers and won’t change it.

Vidgamer, I’m curious about the coil in series with the woofer. What value of inductance and resistance did you use?
 
Allex, that's perfectly fine -- you gotta get your priorities straight.

I think your observations and suggestions were correct, and it was extremely helpful to have those schematics to compare! You were definitely right about it being light in the bass and high in the treble, and I think my test measurements, although imperfect, only confirm that.

I used a 2.5 mH coil with 0.59 ohm resistance. While selection seems to be limited, I probably could have found something with less resistance, but it would have been more expensive (and larger!). As it is, these were pretty pricey, compared to the cost of the kit -- it's like 25% more just to modify the crossover!
Sidewinder 2.5 mH Air Core Inductor 16 AWG

It seems apparent that with both of the Swan/HiVi kits, the crossover can be improved. It's just that the 2.2 kit doesn't get much attention.
 
Hi,

I got a kit as a Christmas gift, the version without cabinets. I have made my own cabinets and used some of the knowledge mister Gravelsen has shared on his website.
The holes for the d6.8s are chamfered, chamfering driver holes
The cabinets are damped with felt on all sides except the baffel, Cabinet-damping
I have made the bracing of the cabinet a little bit more sturdy than the one in the kit.
The cabinets are made in 19mm mdf with 0.7 mm walnut rot veneer glued to the surfing with epoxy. The surface finish is made with about 0.7 to 1.0 mm of clear epoxy covered with 3 to 4 layers of polyuretan lacquer.

I don't have any measuring equipment, but I can't hear that the speakers lack in bass response. They beat my modified B&W 601 s3 by far in bass and clear sound. But interested in the modifications you guys have suggested I tried to replace the coil for the midwoofer to a 2.4 mH 0.51 ohm coil and I think the sound got a little more laid back in the midrange.

At rather high levels I think there is some kind of distortion in low treble or high mid, so I have ideas of changing the cap for the tweeter to a smaller value, 5.6 uF or 4.7 uF as in the 2.1 se. The data sheet for the tweeter say that the tweeters frequency response is 2500 Hz and up but in a another part of the data sheet say flat responce from 1500 Hz and up. And in some forums people say that this tweeter shouldn't be used lower than 2500 -2700 Hz, do you guys have any thoughts on this?