Very nice looking work!
It's funny, I see similarities to my current monitors as well as the Focal's I re-worked the bass crossover on.
Some comments:
- My tweeter crossover is very similar to yours. I used Clarity MR with an Audyn 0.1 TC on the large cap, and Duelund 0.047 Silver in wax/paper. The Audyn did great things for the midrange, the Duelund I am not sure was worth it's spend. I hear better string definition and extension, but not $200/cap worth of those.
- Tweeter inductor - Just wondering, you have OK impedance without a resistor on the 0.2?
- Woofer Impedance - The Focal's bass section started up looking much like you have it, but I had issues with the bass impedance. What is the impedance in the bass like for you? Also, I'd be cheap here and use large value (20uF to 47uF) Axon's. If you are curious, check out my Focal 918 upgrades page and scroll down to the "Bogus Woofer Cap" discussion.
I notice the driver's still have '!' marks. I highly recommend DATS as very accurate and very useful for the impedance of the drivers. I measured in cabinet and used for my Xsim work, FR and impedance were spot on.
I'm sorry if this is the amateurish comments of a hobbyist compared to you. I just wanted to share my enthusiasm.
Best,
Erik
It's funny, I see similarities to my current monitors as well as the Focal's I re-worked the bass crossover on.
Some comments:
- My tweeter crossover is very similar to yours. I used Clarity MR with an Audyn 0.1 TC on the large cap, and Duelund 0.047 Silver in wax/paper. The Audyn did great things for the midrange, the Duelund I am not sure was worth it's spend. I hear better string definition and extension, but not $200/cap worth of those.
- Tweeter inductor - Just wondering, you have OK impedance without a resistor on the 0.2?
- Woofer Impedance - The Focal's bass section started up looking much like you have it, but I had issues with the bass impedance. What is the impedance in the bass like for you? Also, I'd be cheap here and use large value (20uF to 47uF) Axon's. If you are curious, check out my Focal 918 upgrades page and scroll down to the "Bogus Woofer Cap" discussion.
I notice the driver's still have '!' marks. I highly recommend DATS as very accurate and very useful for the impedance of the drivers. I measured in cabinet and used for my Xsim work, FR and impedance were spot on.
I'm sorry if this is the amateurish comments of a hobbyist compared to you. I just wanted to share my enthusiasm.
Best,
Erik
Last edited:
Thanks for the comments. I never got around to properly measure the response and impedance of drivers in the B&W 802S3. My first intention was to recreate the original crossover with high quality passive components without compromising the original design. I might look into DATS though, it's not that expensive.
As to my capacitor choices, I draw the line at Mundorf Supreme level. I value good quality parts, but you can only go so far until the cost to improvement ratio is hard to digest. My experience with Russian caps on bypass has been previously positive so this is what I went with. I'm not after the last bit of detail. As a matter of fact, probably many modern designs outperform my old B&W in this respect and that without using boutique parts. I like my sound rich and smooth.
As for my missing resistance in the tweeter network coil, I do have a Mills wire wound resistor at hand to compensate, but I didn't try it.
I took SPL measurements with a Dayton UMM-6 and this is what I got:
Overall, I see no problems related to my crossover design choices. There is, however a bump around 3.5kHz which I will try to compensate. This may mean I need a different coil anyway and perhaps some other small tweaks.
As to my capacitor choices, I draw the line at Mundorf Supreme level. I value good quality parts, but you can only go so far until the cost to improvement ratio is hard to digest. My experience with Russian caps on bypass has been previously positive so this is what I went with. I'm not after the last bit of detail. As a matter of fact, probably many modern designs outperform my old B&W in this respect and that without using boutique parts. I like my sound rich and smooth.
As for my missing resistance in the tweeter network coil, I do have a Mills wire wound resistor at hand to compensate, but I didn't try it.
I took SPL measurements with a Dayton UMM-6 and this is what I got:
Overall, I see no problems related to my crossover design choices. There is, however a bump around 3.5kHz which I will try to compensate. This may mean I need a different coil anyway and perhaps some other small tweaks.
Very sorry about the caps, I meant to share with you my experience, to see if you had found any benefit in your bypass caps. 🙂 By sharing exactly what I did I thought it was easier to compare. I'm sure your caps are ideal for your drivers.
Since you are working with XSim, you can examine the impedance for the virtual drivers. It won't be exact, you probably have less than 8 Ohm drivers, but it will give you an idea of what, if any, dangers exist. The problem I'm worried about isn't the frequency response but the impedance dropping and making the amplifiers work too hard. At worst this could cause premature amplifier failure, but more likely it can be no effect to a soft bass and reduced dynamic range.
If you had exact in-cabinet impedance measurements for the woofers you could put them into XSim and simulate changes. I may do this later, I'm really worried how low that woofer and tweeter impedance is getting. Maybe you have Boulder room heaters (i.e. amplifiers) and don't care. 🙂
Best,
Erik
So I copied your schematic and plotted the impedance so I could show you what I was a little worried about. Of course, this can't be accurate because I don't have actual driver impedance files.
This assumed purely resistive and 8 Ohm drivers. Any reduction from this will make the impedance charts lower.
The woofers go up to around 200-300 Hz. There's a dip around 300-400Hz below 4 Ohms. There's also a really severe peak at the mid/tweeter crossover frequency. I'm not sure how much this matters with solid state, but if you were driving this with a tube amp an impedance equalizing circuit should be considered. A 10uF cap with 30-40 Ohms at the inputs would do.
Also, I'm surprised there's no Zobel anywhere. If you can't measure the drivers it would be interesting to measure the woofers alone (through the crossover) to see if there's any HF bleed through caused by their inductive impedance rise.
This assumed purely resistive and 8 Ohm drivers. Any reduction from this will make the impedance charts lower.
The woofers go up to around 200-300 Hz. There's a dip around 300-400Hz below 4 Ohms. There's also a really severe peak at the mid/tweeter crossover frequency. I'm not sure how much this matters with solid state, but if you were driving this with a tube amp an impedance equalizing circuit should be considered. A 10uF cap with 30-40 Ohms at the inputs would do.
Also, I'm surprised there's no Zobel anywhere. If you can't measure the drivers it would be interesting to measure the woofers alone (through the crossover) to see if there's any HF bleed through caused by their inductive impedance rise.
No reason to be sorry. I'd love to have some, if not all, Duelund caps in my XO 🙂. Putting them in would probably cost as much as I've paid for the B&W and the new XOs.
I did some tweaks in Xsim and looks like exchanging the HF coil to 0.19mH 0.92 Ohm (Jantzen 0.4mm diameter) should help me almost completely get rid of that bump around 3.5kHz.
Another bump that is visible in both Xsim and the measured SPL is around 650Hz. I'd love to reduce that as well. Any tips? I'm attaching my updated Xsim data for reference.
View attachment matrix802s3 mod + new coil.zip
The impedance curve according to Xsim seems to go down to around 3.5 Ohm. This is not far from official specifications (3.7 Ohm). I'm not worried, my amps have the juice necessary.
I did some tweaks in Xsim and looks like exchanging the HF coil to 0.19mH 0.92 Ohm (Jantzen 0.4mm diameter) should help me almost completely get rid of that bump around 3.5kHz.
Another bump that is visible in both Xsim and the measured SPL is around 650Hz. I'd love to reduce that as well. Any tips? I'm attaching my updated Xsim data for reference.
View attachment matrix802s3 mod + new coil.zip
The impedance curve according to Xsim seems to go down to around 3.5 Ohm. This is not far from official specifications (3.7 Ohm). I'm not worried, my amps have the juice necessary.
So it seems if I reduce the MF inductance (L2) from 1.5mH to 1.4mH and at the same time increase the MF series resistance from ~1.72 to 2.7 Ohm, I should be able to reduce the 650Hz bump by 1.6dB while keeping the rest of frequency response in tact. Worth a try? My Mundorfs are baked varnish coils so it may not be an easy thing to do and then bring back to current state.
Last edited:
Well, my question and I'm sorry for not being clear, is that I could not find a very valuable reason to have the Duelund's there. I can't tell a difference, certainly not enough of a difference to justify the cost. 🙂 I was wondering if you found the 0.047uF cap in your speakers very useful.
I'm glad you've checked the impedance curve! Since you have the ability to capture an FR, have you recorded each driver for XSim in cabinet?
Best,
Erik
Can we see the FR in log scale please?
It's really difficult to do these adjustments without having the FR of each driver in cabinet, and no crossover. It's also much easier to get accurate with the impedance data. DATS V2 is $99. If you can swing that, we can do some crazy good things about this speaker. 🙂
At the very least, I'd strongly suggest getting the mid and woofer, then mid+woofer. All without crossover. This way we can set distance and get the right FR output from XSim. Well, it wont' be exactly right due to lacking impedance, but it will get us half way to great. 🙂
Erik
It's really difficult to do these adjustments without having the FR of each driver in cabinet, and no crossover. It's also much easier to get accurate with the impedance data. DATS V2 is $99. If you can swing that, we can do some crazy good things about this speaker. 🙂
At the very least, I'd strongly suggest getting the mid and woofer, then mid+woofer. All without crossover. This way we can set distance and get the right FR output from XSim. Well, it wont' be exactly right due to lacking impedance, but it will get us half way to great. 🙂
Erik
No, I never measured my drivers separately. I did my XOs before I got the Dayton UMM-6 mic, and now that everything is soldered together, I guess I'm just too lazy to do that.
My only concern are the XO transition point and it seems I should be able to do some minor improvements with my basic Xsim model and SPL measurements.
I didn't do a separate listening session for the 0.047uF PIO capacitor. I built in both the PIO and Teflon together. My experience (previous and current) is that they work well together and seem to fill up the space with ambiance and mellow details. Not that bare Mundorf Supreme are worse per se. They just sound more direct without the bypasses.
My only concern are the XO transition point and it seems I should be able to do some minor improvements with my basic Xsim model and SPL measurements.
I didn't do a separate listening session for the 0.047uF PIO capacitor. I built in both the PIO and Teflon together. My experience (previous and current) is that they work well together and seem to fill up the space with ambiance and mellow details. Not that bare Mundorf Supreme are worse per se. They just sound more direct without the bypasses.
Don't be lazy. 🙂 You have a lot more power to change things precisely and you'll get there faster. I've done lazy and "not lazy" and "not lazy" is a lot less work.
Ahhh. I know Magico uses Mundorf's, but after listening I didn't think they were right for me. I have Mundorf AMT tweeters, and was planning to go 100% mundorf caps until then. I like mellow and space filling though a lot!
Best,
Erik
Ahhh. I know Magico uses Mundorf's, but after listening I didn't think they were right for me. I have Mundorf AMT tweeters, and was planning to go 100% mundorf caps until then. I like mellow and space filling though a lot!
Best,
Erik
I would not implement this without the driver FR and impedance files, but here is the kind of work you could do if you had them. Now you can completely rethink the crossover, not just tweak here and there. Much faster IMHO.
First, the schematics I used. I cheat and create two drivers to compare them side-by-side in XSim. Once close, I will use the new schematic in the original system schematic. The second chart is the comparative FR of the two drivers.
For the HP I used a knee of 600 and Q of 1.1. For the LP I used 2kHz and Q of 0.7. For the L pad I used a padding of 6DB. All of this is so much easier when you can wipe away all that came before I think, and impossible to do accurately without the right measurements to start with.
First, the schematics I used. I cheat and create two drivers to compare them side-by-side in XSim. Once close, I will use the new schematic in the original system schematic. The second chart is the comparative FR of the two drivers.
For the HP I used a knee of 600 and Q of 1.1. For the LP I used 2kHz and Q of 0.7. For the L pad I used a padding of 6DB. All of this is so much easier when you can wipe away all that came before I think, and impossible to do accurately without the right measurements to start with.
Last edited:
Of course, another approach is to attemt to add a series notch filter around 800 Hz, but really, that's kind of the camel's nose under the tent. Soon after you try tweaks like that the rest comes down. 🙂 Still, feel free to attempt it with the Circuit Blocks.
One more observation. The crossovers slopes are too ideal. That is, they appear to be made without consideration for the driver behavior, and yet seem unnecessarily complicated.
Again, measurement is key.
Best,
Erik
One more observation. The crossovers slopes are too ideal. That is, they appear to be made without consideration for the driver behavior, and yet seem unnecessarily complicated.
Again, measurement is key.
Best,
Erik
Last edited:
So there .. I swapped the 0.2mH coil to a slightly higher ohmic one and surprise - barely anything changed. This basically proves what you've been saying about simulating without real life data.
However, the thing is I don't want to redesign these XOs from scratch. I spent a good amount of cash on parts and it would be a waste to ditch them just for some minor corrections in SPL.
My current course of action is to get a couple of 1Ohm resistors and 1uF caps to check what I can fine-tune without a complete overhaul.
Well, my other suggestion was to use the XSim Circuit blocks and put in a notch filter. 🙂
We also have no idea how well/poorly phase matched everything is without those measurements. It really wouldn't kill you to measure the drivers you know. 🙂 At least put that in, you'll find XSim will become dramatically more accurate and useful.
Best,
Erik
We also have no idea how well/poorly phase matched everything is without those measurements. It really wouldn't kill you to measure the drivers you know. 🙂 At least put that in, you'll find XSim will become dramatically more accurate and useful.
Best,
Erik
Glina,
I went down exactly the same route you went through with my Focal's, though I only spent a ton of money on coils, so I totally get where you are coming from. I learned too much about them and am now parting them out. 🙂
If you are spending that much money on parts though, why not spend the effort to measure what they are doing first? XSim can save you money that way. 🙂
Best,
Erik
Another try today and I put an additional 1 Ohm resistor in series with the 0.2mH coil. The 3.5kHz peak is no more a problem. Funnily enough, measured further away from the speaker (listening position) the 650Hz peak is also neglible.
Anyhow, you know how this goes. I first wanted an XO refresh without electrolytic caps (by now 25 years old) and cored coils, so I did some calculations and ordered compatible parts. I built the first version and listened - sound great, but what if my calculations we wrong? I got a proper RLC meter and measured. How about that, I did good. I then discovered Xsim and created the model. Worried with the peaks I saw on the model, I ordered the UMM-6 mic. It's nowhere near as bad as Xsim shows, but I know I can do better. This is where I am now. Should I start with all the knowledge, I would do some things the other way around. I'll leave that for the next project.
Anyhow, you know how this goes. I first wanted an XO refresh without electrolytic caps (by now 25 years old) and cored coils, so I did some calculations and ordered compatible parts. I built the first version and listened - sound great, but what if my calculations we wrong? I got a proper RLC meter and measured. How about that, I did good. I then discovered Xsim and created the model. Worried with the peaks I saw on the model, I ordered the UMM-6 mic. It's nowhere near as bad as Xsim shows, but I know I can do better. This is where I am now. Should I start with all the knowledge, I would do some things the other way around. I'll leave that for the next project.
801s
I'm wondering how the xover mod worked. Do you notice better sound. If so what did you notice? I'm thinking about messing with my 801s but I really don't have experience in this and don't have knowledge about crossovers.
I'm wondering how the xover mod worked. Do you notice better sound. If so what did you notice? I'm thinking about messing with my 801s but I really don't have experience in this and don't have knowledge about crossovers.
I have some Totem 2 way speakers and notice they seem to use the oil filled motor capacitor and they sound really good. I will look into using them in future.
It did work out fine and the sound did improve in overall clarity. The general character didn't change due to careful calculations and matching of all components.
I wouldn't go to such great length again to be honest as the expense was significant and improvements fairly mild.
What I would certainly do is exchange the old electrolytic bipolars to new ones (also electrolytic type not to have to worry about equivalent series resistance) from Mundorf or Jantzen, and the tweeter capacitors to a modern premium capacitor, such as the Mundorf Supreme, Jantzen Superes, Clarity Cap CMR or my new favorite, the Jantzen Alumen Z-Cap.
- Status
- Not open for further replies.