I was recently gifted with a pair of the Dayton B652 Air bookshelf speakers.
They sounded like they have potential, but the treble is at dentistry levels of brightness. Plus cone breakup, etc etc...
So I looked for various crossover mods. This is the one I thought was the smartest way to go:
Fun with the Dayton B652-AIR -
Techtalk Speaker Building, Audio, Video Discussion Forum
So I ordered the parts.
I figured it should still work since the B652s still come with the same crossover components as the modder reported.
However it seems this mod is out of date, as well as the data from the Noaudiophile review. When I measured the speakers with and without the tweeter reversed, the result was almost identical (although the Noaudiophile DSP correction file has the cone mode filter in the right place).
The B652s have been revised in the past, IIRC the woofer cone depth was changed and the crossover was also changed.
In any case, the mod seems to be a reasonable starting point. The first thing I had to do was add a 2.2uF capacitor in parallel with the 4.7uF in order to shift the resonance filter to the new cone mode frequency.
However I'm left with a notch at the crossover frequency. I figured I would post this before finishing in case someone has already got this sorted out or takes one look and knows what to do.
Here is the original BSC mod:
First line is stock B652 with the BSC mod.
Second line is with tweeter reversed.
3rd line is with 2.2uF added in parallel with the 4.7uF cap to get the correct resonance frequency.
So, any thoughts on the best way to fix the notch? It seems changing the 2.2uF cap is the first thing to try...
They sounded like they have potential, but the treble is at dentistry levels of brightness. Plus cone breakup, etc etc...
So I looked for various crossover mods. This is the one I thought was the smartest way to go:
Fun with the Dayton B652-AIR -
Techtalk Speaker Building, Audio, Video Discussion Forum
So I ordered the parts.
I figured it should still work since the B652s still come with the same crossover components as the modder reported.
However it seems this mod is out of date, as well as the data from the Noaudiophile review. When I measured the speakers with and without the tweeter reversed, the result was almost identical (although the Noaudiophile DSP correction file has the cone mode filter in the right place).
The B652s have been revised in the past, IIRC the woofer cone depth was changed and the crossover was also changed.
In any case, the mod seems to be a reasonable starting point. The first thing I had to do was add a 2.2uF capacitor in parallel with the 4.7uF in order to shift the resonance filter to the new cone mode frequency.
However I'm left with a notch at the crossover frequency. I figured I would post this before finishing in case someone has already got this sorted out or takes one look and knows what to do.
Here is the original BSC mod:
First line is stock B652 with the BSC mod.
Second line is with tweeter reversed.
3rd line is with 2.2uF added in parallel with the 4.7uF cap to get the correct resonance frequency.
So, any thoughts on the best way to fix the notch? It seems changing the 2.2uF cap is the first thing to try...
Attachments
Did you flip the tweeter polarity after adding the 2.2 cap? Also in my measurement there was a big dip in that area for the tweeter so maybe that is what you are showing, albeit a bit larger than what I had.
BTW have you listened to these at all? As I mentioned in that thread I never actually built the modded version, just seemed like a fun exercise for anyone interested.
BTW have you listened to these at all? As I mentioned in that thread I never actually built the modded version, just seemed like a fun exercise for anyone interested.
Last edited:
A tweeter that is best exchanged for a better one but then you'd need a new filter, and then the woofer would have appeared also ready to be thrown away, and then is doesn't make sense anymore to bother at all.
This is the B652 Air, it's not the same speaker!
No, I don't expect a miracle after flipping the tweeter several times already in other tests. The dip responds to crossover, it is definitely not part of the tweeter response.
I knew what I was getting into. It was frustrating that the most pertinent mod was untested.
I've listened to them for a while over the last few months. With the BSC mod I prefer them over stock despite the 6KHz notch. At least I can stand them now.
Here are graphs when changing the 2.2uF cap. From top to bottom, 1.5uF, 2.2uF, 3.7uF.
No, I don't expect a miracle after flipping the tweeter several times already in other tests. The dip responds to crossover, it is definitely not part of the tweeter response.
I knew what I was getting into. It was frustrating that the most pertinent mod was untested.
I've listened to them for a while over the last few months. With the BSC mod I prefer them over stock despite the 6KHz notch. At least I can stand them now.
Here are graphs when changing the 2.2uF cap. From top to bottom, 1.5uF, 2.2uF, 3.7uF.
Attachments
Last edited:
Well, I broke down and measured both drivers separately in box. It turns out the 6KHz dip really is part of the AMT's response. I didn't get this in measurement with another pair of B652s. Maybe the tweeter is missing some damping material in the back cavity?
Attached is a Boxsim file with everything loaded, if you want to give it another go.
Attached is a Boxsim file with everything loaded, if you want to give it another go.
I pointed the speaker left and right and the measurements were so similar to on-axis they almost fully overlapped.
Also, note my microphone is 7db down at 20KHz, that is not the speaker.
Last edited:
Getting the final values pinned down is a bit difficult. The 3.45KHz spike has come up again. It seems the 40R resistor need to be increased. The second line is with the 40R resistor increased to 50R. The Boxsim simulation doesn't reflect this resonance realistically.
The tweeter capacitor should probably be 1u or 1.5u depending I guess on what happens when I stuff the box.
The tweeter capacitor should probably be 1u or 1.5u depending I guess on what happens when I stuff the box.
Attachments
With the first order crossover and the microphone, the resulting level response is a very good outcome.
I would measure just the woofer when you make those crossover changes. Notch filters cause a resonant "rebound" of the response outside of the notch. So you could end up with more woofer energy at say, 10khz, then before, but it might not be obvious when the tweeter is also being measured. But might sound worse to the ear. For that reason also I would tend to aim these offaxis so the woofer rolls faster. At $14 this design was the best compromise IMO, but isn't at all a "perfect" crossover. After you settle on improvements let us know it sounds!
I noticed the rebound effect with the notch filter. The inductor could be tapped off leaving some turns not bypassed by the capacitor.
The crossover frequency increased with the changes, it would probably be good to lower it if possible.
The crossover frequency increased with the changes, it would probably be good to lower it if possible.
The tweeter has a pretty steep natural rolloff, that value cap was chosen as much to protect the tweeter from trying to frequencies it can't, as it was to shape actual response. Which is to say you really don't have the ability to lower the crossover if that is what you were thinking.
So I increased the notch resistor to 80R and decreased the notch capacitance to 6.2uF. Green is previous, yellow is new. Which one do you think is better?
The spike at 4.8KHz got larger and the 8-9KHz region seems a bit hot. So I guess the question is whether the spike at 3KHz is worse than the spike at 4.8KHz?
BTW, I am listening to the speaker as I go. It's already much smoother to listen to, the bass and the mids and treble come together evenly, I just want to see if I can trade some harshness for clarity.
The spike at 4.8KHz got larger and the 8-9KHz region seems a bit hot. So I guess the question is whether the spike at 3KHz is worse than the spike at 4.8KHz?
BTW, I am listening to the speaker as I go. It's already much smoother to listen to, the bass and the mids and treble come together evenly, I just want to see if I can trade some harshness for clarity.
Attachments
Here is the current measurement including phase. I thought it was interesting that the phase is getting close to zero. The phase being positive through most of the range would seem to indicate gradual rising response towards the low end. The phase in treble gets lower the more I tilt the speaker up to match the acoustic centers.
Attachments
Neither you nor Noaudiophile measured the ENORMOUS notch in the tweeter response at 6KHz. I can only figure it must be a recent design change or quality control issue.
Attachments
Last edited:
So the frequency response is relatively flat now, at least when compared with before. It is slightly biased towards the low end, although there is a raised section at 600Hz-1.3KHz.
I've been listening and and it sounds nicely neutral compared to before. The low end is slightly boxy (I haven't stuffed the box yet). The treble is not outrageously loud and irritating. Another person mentioned the treble seems a bit muted, and I can see why, but I think that is partly due to being used to typical speakers with wild treble. But I might increase the tweeter cap to 1.5uF. Someone else actually paid for this experiment, so I will probably cater to their taste.
Something is clearly not right with the 6KHz notch and that is a shame. I'm not sure what I need to investigate to find out the cause, but I suppose I could open up one of the AMTs, maybe try putting some felt inside.
Any thoughts on where to go from here?
I've been listening and and it sounds nicely neutral compared to before. The low end is slightly boxy (I haven't stuffed the box yet). The treble is not outrageously loud and irritating. Another person mentioned the treble seems a bit muted, and I can see why, but I think that is partly due to being used to typical speakers with wild treble. But I might increase the tweeter cap to 1.5uF. Someone else actually paid for this experiment, so I will probably cater to their taste.
Something is clearly not right with the 6KHz notch and that is a shame. I'm not sure what I need to investigate to find out the cause, but I suppose I could open up one of the AMTs, maybe try putting some felt inside.
Any thoughts on where to go from here?
I think the components are pretty much settled after stuffing the box. There is a drop at 300-500Hz that I'm not sure I can do anything about.
Listening should be below the plane of the woofer, above that it gets pretty wild. At about 1 meter distance the lower half of the dustcap should be at about eye level.
I'm not a speaker guru so if anything I say sounds silly it probably is.
Listening should be below the plane of the woofer, above that it gets pretty wild. At about 1 meter distance the lower half of the dustcap should be at about eye level.
I'm not a speaker guru so if anything I say sounds silly it probably is.
Attachments
Any advice on how to mount the crossover in the box? I just soldered everything together and then stuck it to the bottom with duct tape... Then it made funny noises on certain tracks, so I put an old burnt sock under the crossover. It still makes noise at certain volumes but perhaps I'm just getting too loud for these speakers.