B&W 803D crossover caps

I am considering an upgrade of the crossover capacitors in by B&W 803Ds, particularly the mid and HF coupling caps.

I took out the top bass driver to find out what caps were installed. It looks like for the diamond tweeter B&W uses a Mundorf Supreme silver/gold, 4.7 mfd 1200V. For the mid driver there are two; a 47 mfd Mundorf MKP 400V series coupling cap (in series with the driver) and a 10 mfd Mundorf Supreme siver/gold bypassing cap (parallel to the driver).

I was thinking about changing out all three, but have a few concerns.

I was going to replace the 10 mfd, 4.7 mfd Supreme silver/gold with Supreme silver/gold/oil. Would there be enough of a difference in these two types to justify the cost? I also do not want to make the upper end any brighter.

I am also concerned about the long term reliability of oil filled caps, as some failures have been reported in warmer environments. I wonder if B&W did not use the silver/gold/oils for that reason.

The biggest impact I suspect will come from the replacement of that series 47 mfd MKP. I would probably use either the Mundorf MCap EVO (Al metalization), MCap EVO oil (Al/oil), or the MCap EVO silver/gold/oil. All three are the same size for 47 mfd, and will fit to replace the MKP. Barring the issues about oil, which might be the best sounding? Again, I want to avoid too much enhancement of the upper midrange.
No it would not be worth it to stay in the Mundorf family. If you want a much better sounding cap top to bottom, then go with the Jupiter copper foil in bees wax cap. Nice step up for sure! I know as I did it. Unless to go with Jupiter or Duelund, stay put.

Yes the Evo is your best bet on the 47uf.

Thanks for your input. Reinforced what I was thinking. The cost to do the swap to the silver/gold/oils would be about $1000 for both channels. Serious money.

I think I will stick with Mundorf at this point, as I want to maintain the character of the 803. The Dueland are stupid money and there is no place to put them without adding a lot of wiring inductance to the network. The Jupiter are about $500 more than the Mundorf for a complete swap sans the 47 mfd, and the 10 mfd 100V part is just a bit too long to fit on the crossover pc board. But looks like a serious contender for the 4.7 mfd feeding the diamond tweeter if I want to do that alone.

I also discovered that all the bass driver mounting screws were barely tight. Torquing those up made a large difference in the bass response. I also torqued the tension bar on the mid-range driver and that also improved sonics quite a bit as well. Freebes that didn't cost me a dime.
I don't wish to sound to basic here but I believe more bypass caps may be the only sensible thing. Have you asked B&W-- or an authorized repair center? Are you up on any crossover design changes for the newest series? ?Have you spoken to Mundorf? And lastly,seriously, have you had a lot of success modifying crossover in other comparably high quality intelligently deigned speakers

My experience with the Mundorf S/G/O was that they added too much emphasis in the treble. I do not have any experience with the S/G capacitors but I suspect I would like them even less than the S/G/O as I would expect them to be brighter and maybe a bit harsher. Given a choice between those two I'd go with the S/G/O, but as I'll describe below, at this level I'd go with the Clarity MR capacitors.

My crossover journeys have ultimately led me to Duelund CAST-Cu capacitors, however, along the way a capacitor that I liked a lot was the Clarity MR. They are fast, extremely quiet, dynamic, very well balanced from top to bottom, spatial and holographic. To those attributes the Duelund CAST-Cu adds a naturalness of tone, greater harmonic richness, and I think perhaps greater sustain although it's been too long since I made that change to likely recall that attribute accurately. Also, I would say that the Clarity MR beats the Duelund VSF-Cu in everything but tone and harmonic richness. I recommend the VSF or CAST ahead of the MR, but unless going with a Duelund crossover, my experience leads me to recommend the Clarity MR without hesitation. I would add, however, that Bill's experience with the Jupiter copper foil caps should not be ignored either, I just don't have any experience with them. Again, based on my experience, and considering budget, if it were my project I would use the Clarity MR's throughout, connecting 2x10micF + 1x27micF caps in parallel to give me the 47micF equivalent for that big value. I strongly prefer the Clarity MR over the Mundorf S/G/O, and as I said, I'm quite confident I would like the Mundorf S/G even less - but comments such as these should be placed in the context of one's system.

Also, depending on what internal wire is already in the speakers I would also consider rewiring the speakers. On my projects I have had very satisfactory results using Kimber TCX. The effect of doing that should not be underestimated. Kimber's Hyper Pure grade copper has no grain that I can hear. Wherever copper wiring is used in my system (interconnects, speaker cables, crossover wiring, speaker wiring) that is what I use. They system is very, very free of grain.

What resistors are in the speakers? My preferred resistors are Mundorf M-Resist 20W. I think Bill likes Mils? Others will recommend the Duelunds, though in my systems I preferred the Mundorfs. For me the M-Resist give a very nice, solid weight to acoustic strings - I don't like plucked acoustic strings to sound too wispy so have come to like the M-Resist a lot.

I do have a couple of questions though. In your original post you said that you wanted to upgrade your crossovers, and in a subsequent post you said that you wanted to maintain the character of the 803. So my questions are: (1) what are you hearing (or not hearing, as the case may be) that has you motivated to upgrade the crossover - what improvements are you seeking?, and (2) what part of the character of the speakers are you wanting to maintain? In what direction are you wanting to take your speakers and/or your system?

Also, what inductors are in the speakers?

Finally, as an upgrade example, I am currently helping a friend with a crossover upgrade to his Silverline LaFolia speakers. The LaFolia uses well-integrated tour-de-force of drivers including the Dynaudio Esotar T330D tweeter, so I have a pretty good feel for where we are headed. For that project we are using:

- Clarity MR 630 VDC caps
- North Creek 10awg inductors
- Mundorf M-Resist 20W resistors
- Kimber TCX internal speaker wire
- WBT Topline binding posts

Unless my friend were going to spend a lot more money I think this will be a very nice setup.



Do not worry. I have plenty of experience modding equipment. If you doubt it, see my post on mods for the Spectral DMC-10 in the WhatsBest forum.

BTW, my version is just the "803D". I have it on good authority that the later version "803 Diamond" uses only the Supreme Oil, not the SG or the SGO. This info was provided by looking at a replacement crossover for the Diamond version. Seems like that is a big step back, according to the only review I have found to rank all three Mundorf Supremes (humblehomemadehifi).

Call B&W regarding crossover caps? Are you kidding?

A simple swap out of a crossover caps is not rocket science. It will not damage the speaker or destroy my amplifiers. At this level of performance, the results will probably be very subtle, but that does not mean they are not worth doing.

Manufacturers often make design decisions based on cost as well as performance. We have an opportunity to re-evaluate that decision and remove cost from the equation, up to certain limits.

Thank you for your thoughtful comments.

My goal is to remove a bit of upper midrange brightness without effecting resolution in bells, symbols, vocal microphonics. The speakers are not at all bad as they are, but could stand a bit of improvement in these areas. I well know from my experience modding amplifiers that improvements in certain components can improve the sound in seemingly small but still significant fashions. I have no hard and fast goal for the end result, but will probably know an improvement when I hear it.

A few other parameters. The replacements need to fit on the existing pc board without kluges or modifications. No running feet of extra wire from a remotely mounted component to the board. They cannot reduce the reliability of the speaker by leakage of oil or wax. And finally there is cost. Given the value of the 803Ds I don't feel like spending $5000 on caps.

I was using the extensive review of humblehomemadehifi for initial evaluations. That review does not place the Clarity MR above the Supreme SG, but I realize there is probably a degree of subjectivity (to each his own). The Jupiter caps are highly regarded, perhaps above the Mundorf line.

As I posted above, it looks like B&W may have stepped back in SQ for the later Diamond series. They may have had a reason for using Supreme Oils (Al metallization) instead of the more expensive SG as in my D version, but I suspect it may have been more cost driven than SQ driven.

Regarding the other components, from my BOM documents on the 803D, the resistors are part number RNP50s, which are 100W, 5% power film resistors (TO247 package) that are clipped in to heatsinks on the CO board. They are made by MB Electronics in Germany. The resistor in series with the diamond tweeter is either 0.5 or 1.0 ohm (there is a contradiction between the value on the BOM and the schematic). The resistor in series with the midrange FTF is 100W 2.0 ohm RNP50s. My suggested replacement would be a Caddock MP900 power film resistor, which has a tighter tolerance and better tc, same TO247 package.

The inductors look like nothing special. Air core, conventional wire wound, wire gauge 1.2 mm (not flat ribbon).

It seems to me the most important cap is the 47 mFd midrange FTF coupling cap. But here we are limited severely by that capacitance. Mundorf has a Supreme EVO SGO that is 47 mFd, but it costs $700-800 and is too large to fit. The next closest are the EVOs I cited above. Bypassing is possible, but probably not that effective as this driver is crossed over at 4KHz, and a bypass would be kluge at best.

The next is probably the diamond tweeter coupling cap at 4.7 mFd. The Jupiter will fit as well as perhaps the Clarity

And yes, I considered re-wiring as well. But not for now, as that requires a complete strip down of all the drivers, and I want to avoid removing that FTF midrange if at all possible.
Hello again DH. I simply asked questions-that I thought may be helpful to you or others thinking of such mods. Re B&W; JBL Engineering was very (in fact surprisingly so) helpful re my 1985 250Ti's many years ago; even to provide me new ideas that they had come up with and would be implementing! I respect the investment B&W has in it's business and goodwill so anticipate good PR; why shouldn't you approach them? Are you perhaps doing a bit of 'assuming'? I believe Mundorf are technically expert. I could not find your post re the Spectral DMC 10 in the WhatsBestForum but am interested as I enjoy a DMC 20 with phono in my system. Might authorized repair centers have knowledgeable techs-perhaps with real interest in what they do?
Sorry if I misinterpreted your comments. Mine were based on an experience I had a few years back with B&W regarding a delamination problem I had with some 804s speakers I had. The finish on the top of the speaker (Cherry) was splitting and cracking and I felt this was a manufacturing defect but B&W would not agree. They were arrogant and condescending which left a bad taste in my mouth. They also tend to believe that their engineers have the last word in innovation and if they did not think of it, well then it will not be an improvement. Ergo I would be loath to contact them about any potential component replacement as I can pretty much predict their response.

You mention JBL and I was going to point out that I also re-built the crossovers in my JBL L212's, as well as the power amp in the sub. I did not use crazy audiophile components (just good Dayton polypro caps and air core inductors) and the sonic results were amazing. I had previously modded these by mounting the drivers on solid walnut plates to time align them. That mod is documented on the JBL Lansing Heritage site (under DHL or DHL93449 no sure which) a few years back.



The post on the WB forum is under the Spectral Manufacturer's section of the forum.


I wanted to thank you as well for your thoughtful reply. I will be interested to learn how your project goes and hope that you will post updates to this thread.

I wanted to make one more comment about the capacitors I discussed which were in a bit of disagreement with Tony Gee's observation. For me the Mundorf S/G/O tilted things upward too much whereas I found the Clarity MR to be very well balanced. In my system the latter was the superior performer and based on that experience I would not recommend the former to anyone wanting to avoid adding emphasis to the high frequencies. Face's and/or Undertow's comments on this topic a few years back on the Capacitor Log thread make for good reading as well. I agree with you too that it'll ultimately come down to system synergies and personal preferences, I simply wanted to relate my experience for your consideration.

All the best,


The more I look into this and the more reviews I read, the more opinions differ.

Looks like the 47 mFd will probably be the Mondorf EVO. Al/oil or SGO I am not sure yet. Some say (ie Tony Gee) that the Al oil are better for midrange which is the application here. On a purely material basis, I prefer the use of silver metallization over Al due to it's conductivity. No one else makes a cap this size that even comes close to fitting.

For the Diamond tweeter coupling cap, I could use the Clarity MR, as it is very reasonably priced. I cannot find anyone with stock on the 630V part, only the 400V part, which Parts Connexion carries. Do you know of any stocking dealers for these? Any opinion about 630V vs 400V? I have heard one should use the highest voltage rating is a series (Tony Gee again).

I would also consider the Jupiter for this application if it weren't for the very high cost of these ($520 pair). If I don't like them, I may not be able to re-sell them if they have been soldered in a circuit. I was speaking with (Jerry??) at Jupiter and he was very knowledgeable, but had no direct experience of the copper foil/wax units being used with B&W diamond tweeters. He also made mention that a source he knows (that likes to cut apart caps to see what's in them) tells him Mundorf "oils" actually have no oil in them. I find this surprising, but perhaps they use extremely small oil films that are not easily detected.
Dhl: I have a bit of recent experience, although with a different speaker. This summer I threw caution to the wind and did some nice crossover upgrades to my Tyler Woodmere speakers. I wanted to share my results in a dedicated thread, but I wanted to give you my thoughts now, so that you could have a bit more information to help you.

I swapped out the SEAS Millennium tweeters for the new SEAS Beryllium tweeters (exact size and fastener swap). I already had the Mundorf Silver/Gold/Oil tweeter cap. I am very, very pleased with the result. I did, naturally, swap resistors also, to match sensitivity as required. I won't go into a long rant, but this is super nice now. Multiple high freq instruments now are played simultaneously with realistic separation and space of each. Nothing harsh or tinny. And the presence is superb.

The tweeter mod was so nice, it made me long for better-defined midrange, especially upper mids where they are approaching high frequencies, as the mids did not measure up to these great high freq notes. Midrange caps are 120 uF, and I bit the bullet and bought six Mundorf M-Cap Supremes which total that capacitance. Yep, they are a big bunch of caps that take a lot of space. Well....was it worth it?? Hell Yes! A good bit of break-in was required, and right off the bat I only noticed more separation in upper mids. But as they broke in, I am getting a ton more definition, air and bass. Much more lower midrange/upper bass coming through now. The midrange drivers are 7", and in the big Woodmere enclosure I get gobs of bass. That is the biggest adjustment for me. I keep asking myself- is it realistic, or just bigger? Those big midrange caps are nearly broken in now I reckon, and I think it is both bigger and more realistic. I accomplished exactly what I was hoping to do in keeping up with the new tweeters.

Best part is that these improvements, while significant, do not ruin the overall sonics of the Woodmere speakers. They have simply been woken up. I could not be happier. Hope this gives you a bit of confidence to try your luck.

It's funny how these reviews go. I think at this level of capacitor it must come down to system synergy and personal preference.

I have experience only with the 630VDC version of the MR and that's the one with which I would stay. For my friend's crossover project a few weeks ago I ordered a set of MR's from PartsConnexion. They didn't have the ones we needed in stock but it only took them a couple of days to get them from their distributor. If PartsConnexion is indicating a longer delay for the values you need then you might give MadiSound a call. HifiCollective in the UK might be another source.

You mentioned the various conductive materials in your post. As I know you are aware, the MR is an aluminized polypropylene cap. I prefer the tone of the copper film caps more, but as you also mentioned in your note the cost of admission to those caps is high. To me, short of making the jump into those caps, the MR is the best out there.

I reread your original post and it looks like you need a 4.7mfd for your tweeter? I have a pair of MR 630VDC in that value that I don't use anymore since moving the crossover to Duelund CAST. I'll send you an email offline to discuss further.



Based on your experience, did you find a material difference between the Mundorf M-Cap Supremes vs the SGO? I understand the "Supremes" do not contain oil, so my quandary is still SG only vs SGO vs silver oil vs "no oil". I still do not quite understand the use/impact of oil in the first place.


I wanted to thank you publicly for generous offer. I will try those Carity caps when I get them.
Thanks DH for the link. I found it interesting but did not see any post mods pics?
Dhl: My project was really different than yours, and it is not a direct comparison at all. Sorry if I failed to make that clear.

Tweeter upgrade was only drivers. I had already started with the S/G/O, so my only variable was the driver.

Midrange upgrade was only capacitors. I originally had a single Solen cap and replaced it with six Mundorf M-Cap Supremes. No further comparison. I went about as far as I could go, given the large cap values of the mids. I think that set me back about $900.

Sorry I have no direct comparisons between other caps.
Giving it another try to help here, after re-reading your scenario. For my project, I read all the cap shootout opinions and took my chances. My beryllium tweeters really like the SGO, but I cannot say more. Absoutely zero objections.

For the mids, I read in more than one cap comparison report that M-Cap Supremes shine in the midrange. After a lengthy break-in, I can agree, although I hear more lower-mid pronouncement than upper-mid (in my rig and in my room- an extremely important qualification that cannot be ignored). I also seem to recall reading that M-Cap Supremes, in crossover applications, yield a better result than EVO, which led me to the M-Cap Supremes for my midrange caps. For about $200 per speaker, you can try two 18uF plus one 10uF, which will yield nearly your 47uF target. To me, it makes sense to try that one first. Took about 20 hours to sound decent, and 50 hours to sound good. After that, they just kept getting better, with more sound and detail coming through. But that was compared to the Solen cap.

Good luck, and please keep us posted.

If you are referring to the Spectral posts, there are three parts because WB forum had a limit on photos per post.

Here are links to all three parts:




I am kind of where you were at. I would like to make a change, but not really mess around with a lot of trial and error swapping stuff in and out. If I could have gotten a general consensus that a particular option was a no brainer improvement, I would have gone that route.

But now it looks like that is not the case so I will probably stick with two mods, the 47 and the 4.7. For the 4.7, I will try the Clarity MR in comparison to the stock Supreme SG. I would like to also try a Jupiter copper/wax/paper but the cost of entry is quite high.

Regarding the 47 I hear what you say but a stack of three parallel caps just will not fit. So practically I am left with the EVO or EVO/Al-Oil or EVO SGO. The EVO Al-oil was suggested by Tony Gee's review but I am actually leaning toward the SGO. Since the EVO Al-Oil is not that expensive, perhaps the reasonable thing to do is to try that first, then maybe the SGO. Then once the mid range choice is selected, try the Clarity MR in the tweeter 4.7 position.

Supremes have a real advantage technically over the EVO in that they are two series capacitors interwoven, which reduces the series inductance of the overall capacitor. The EVO is a conventionally wound single cap, but is more compact because it is not really 4 caps in one like the Supreme.

What I cannot rationalize from a fundamental POV is what difference the materials combos make. Aluminum metallization with and without oil vs silver metallization with oil vs silver/1% gold with and without oil. Clearly silver and silver gold are much better conductors than Aluminum, but the metallization thickness matters too. The silver/gold should have a better (lower) DF if the metal was a better conductor, but Mundorf's DF data do not support this. So maybe the Al films are much thicker than the Ag/Au, offsetting any advantage in bulk conductivity.

I suspect the gold is added to the silver to reduce any tendency for corrosion of silver. Silver is not really that "noble" in that it easily forms chlorides and sulfides (tarnish), and these may be formed by chloride or sulfide impurities in the oil (or in outside air contamination if it gets into the cap). Gold, even in small percentages, may reduce the tendency to form these compounds. That is why I prefer the silver gold alloys from a basic longevity perspective, although people's impressions of sonic performance are another matter.
Oh, and one other point. B&W went with a Supreme SGO (10 mFd) as a bypass/parallel cap across the mid driver, but an MKP in series with it. I always wondered why they did this, but now realize that the series cap is in series with a very large outboard inductor (1.25 mH). So why spend a lot for a low inductance cap like the Supreme if you are going to put a large inductance in series with it? The parallel bypass cap, however, is directly across the driver so low inductance will pay off there.
From my experience with both Clarity and Mundorph caps the Mundorph S/G/O to Clarity is not a big jump and not worth the cost.

My recommendation, based on many crossover updates, would be to change the resistors to Duelund Silver Graphite first. This will improve transparency and more importantly in the context of your comments result in a much smoother sound. Steen Duelund is on record as saying that when upgrading crossovers with Duelund components you should start with his resistors.

Of the caps in the standard crossover I would only do the 47mF in the mid - probably a Silver/oil Mundorph which is a little smoother than the S/G/O.

Another option if you are keeping the speakers long term is to remove the crossovers and place them externally. Again I have found significant improvements in lower noise floor and less grain by simply removing the crossovers from the box.


I would like to maintain the stock look if at all possible. Relocating the XO to the outside, well, you might just as well build the XOs from scratch then. No for me at this time. I agree that relocating them outside the cabinet would remove a lot of component vibration issues that XOs face while mounted inside.

The Dueland resistors will not fit the stock layout, as B&W uses radial lead (not axial lead) power tab type resistors with heat sinks. Their power rating of 100W is also way above any Dueland power rating.

Only an EVO 47 mFd SGO will fit. But I agree that one should be replaced first.
I have also used the Mundorph M-Resist (20w) with good results.
Have you considered redoing the crossovers on new boards with point to point wiring and better component layout to reduce interaction ?

In a word, no.

I had only planned to replace a few caps, not rebuild the entire cross over. I don't think that a rebuild is sonically warranted at this point.

I am hesitant to replace 100W rated components with 20W versions, no matter what the sonic results MIGHT be. Perhaps with a 0.5 ohm resistor a 20W might work, but I have not done the power dissipation analysis to confirm.
Dhl I did move the huge crossover in my 1985 JBL 250Ti's outboard to great effect. Perhaps you can consider ways of isolating your XO within the boxes, although I feel B&W are respectful of this benefit as previously they advertized the benefit of having the XO in the bases of top speakers. By the way what are your feelings re: bypass caps?
The best sounding resistors by far are Path Audio folks. Tried them all and these are special. Not cheap, but very special! Look into them and be impressed.

I can see the potential benefit to moving the XOs outside the box. My early JBL L212 had the crossovers outside of the enclosed driver housings.

Vibration from the drivers will definitely influence the inductors and capacitors in the network, simply "gluing" them to the pc board with lumps of RTV or caulk may not be that effective.

But practically speaking (no pun intended) moving them outboard of the 803D also has it's pitfalls, as you end up with longer speaker wiring between the drivers and the XO, with the accompanying increase in inductance and resistance. Not to mention the ugly appearance. The bass XO is huge, with monster 100 mFd Mundorf polypropylene bypass caps. Placing that outside would look butt ugly IMHO.

Bypassing caps...not sure what they buy you in a crossover. I know folks use them to "tweak" the sound but I have never played with them for that. I would rather buy a single cap with the right dielectric and low ESR in the first place if at all possible. Plus, from a technical viewpoint, things like dielectric adsorption cannot be remedied with parallel caps; only ESR or maybe ESL. The frequency range in a speaker crossover (maxing at maybe 30 KHz) is just not high enough to see a resonant frequency shift by using a 0.1 mFd cap in parallel with a 5 or 10 mFd coupling cap in a tweeter XO, for example.

So for example, the B&W diamond tweeter is crossed at 3500 -4000 Hz. The reactive impedance of the 4.7 mFd coupling cap at 4KHz is about 8 ohms. It would only be 0.8 ohms at 40 KHz! The 4.7 Mundorf has an ESR that is about 10 milli ohms or .01 ohms, still 10 times lower than the capacitive reactance at 40 KHz. Now what good is placing a 0.1 mFd cap in parallel with the 4.7? It would have a capacitive reactance of 50X the 4.7 (at 40 ohms) and who cares what its ESR is, because that would be swamped with the capacitive reactance.

If you are parallel connecting a series of lower value caps that may make sense if you cannot get the total value in a single cap (like a teflon or polystyrene, for example).

Hi, my experience might be of some help if you want to upgrade the 803D2, which is what I've done. I did buy a new set of Sonus Faber Olympica III, and wanted so much to replace my 803 (or to be honest, my wife wanted to replace my 803...) The SF midrange was to die for, but the same cannot be said about the tweeter and bass. To make a long story short, I sold the SF after upgrading the X-over in the 803.

The 803 suffers from; some light upper frequency enhancement with the tweeter, and some nasality and "in the box" sound from the Kevlar midrange.

When going through the x-over (B&W had kindly posted the schematic on their homepage), it was obvious that the 47uF MKP Mundorf and the industrial resistor in series with the midrange was a week point. I knew also that the 5,1uF Mundorf Supreme SGO combined with tweeters easily can highlight the upper frequencies, too much in a not optimal set up.

I first replaced the 47uF with the new EVO MCap Alu oil and put in a Duelund Silver CAST resistor. I also replaced the 10uF Supreme Oil in parallel with Supreme SGO.

It took some hundred hours to burn in the new setup, but the result was very satisfying. In brief, a much larger soundstage, the in the box sound was 75% reduced, the bass(!) was indeed punchier, 3D stage in all direction had grown. At the same time, there were some highlighting of the upper midrange that was not to my likings. I therefore decided to replace the EVO MCap with the new Supreme EVO Oil.

At the same time I decided to replace the 5,1uF with a Duelund Copper CAST to see if I could get more from the tweeter.

It now needed some 500 hours to sound OK, and after around 700hours, it was confirmed, not only was the midrange free from any kind of boxiness, it was also the best sounding midrange I've experienced so far in my home. The tweeter together with the midrange now sounds like a Stradivarius with all I can wish for regarding holography and texture.

I have listened to the new 805D3 in two different setups. The first did impress me deeply. They had the speed and holography like never before. In the second set up they did not impress me at all, just telling me that the surrounding is just as important as the speaker themselves. I did like the speed and holography, but they did not have the texture as the 803 now has.

With the speed from the 805 in mind, I decided to replace the MKP for the bass drivers as well, this time with the fantastic resistors from Pathaudio and Mundorf EVO Alu in oil.

This time the upgrade gave me some more macro dynamic, but not as much as I had hoped for. It seems like the driver themselves are the bottleneck here.

There is space for the upgrade behind the Kevlar driver, but you have to reorganize X-over a bit. For those how want to see how it can be done, just send me a mail, s-b-o@frisurf.no.

I'm very happy I sold the Sonus, the sound from the 803 betters them in every aspect, and this with a great margin. I've of course considered the new 804D3, but I'm not sure they will better the fantastic texture, bass and resolution from the 803. Be aware, they use MKP, electrolyte and not the top of the line Mundorf capacitors. They have also started to bypass the capacitors with smaller values. I've never succeeded with this little trick in my x-overs, for power supplies, yes, but not in loudspeakers.

The 803D3 are too expensive, and since I've now started to upgrade of the 803, I've ordered the new Continuum drivers to replace the Kevlar. Still waiting for the drivers to arrive.

My experience tells me, there is lot more to get from the 803 with better components, and it is a shame that B&W has not offered a signature model with these upgrades incorporated already!

(To bad I cannot show any photoes here in the Forum)


Interesting post. I have been looking at what B&W has been doing with these cross overs in the last few product iterations. Particularly with the capacitors, they have been all over the map, except with that 47 mFd MKP, which is consistently used from the 804S through the 803D2/3.

In my 803d (first diamond in the series), the tweeter coupling cap is 4.7 mFd Mundorf Silver/Gold Supreme (not oil). In the 803D2, they went to Supreme Oil (Al metallization?), not Supreme Silver/Oil or Silver/Gold/Oil. Looks like in the D3 they are using Supreme SGO at 5.1 mFd, not 4.7 mFd?

I have decided to try the EVO SGO 47mFd in the midrange. The Al/Oil is cheaper but I prefer silver metallization. I have also decided to replace my Supreme 4.7 mFd S/G with a Supreme 4.7 mFd SGO. If that is too strident, I will consider a Jupiter copper paper/oil, but I am nervous about the longevity of these paper/oil caps (along with the cost).

I also have a pair of 804S, with the Al tweeter. That product uses almost the same values as the 803d. The tweeter coupling cap is Mundorf 4.7 mFd Supreme, and the midrange (which is the same unit as in the 803) is coupled with that MKP 47 mFd, and bypassed with a Mundorf Supreme 10 mFd. Like the 805, the bass drivers are bypassed with an electrolytic parallel with a film (150 mFd electrolytic parallel with a 47 mFd MKP film). In the 803D, the bass drivers are bypassed with two 100 mFd MKP films.

I agree with you that bypassing an electrolytic with a film in a cross over is not a good idea. I think that is because in a XO, the caps see significant AC voltages. Any dielectric adsorption effects in an electrolytic cannot be reduced by a film in parallel, and may therefor may be audible.
I use the Jupiter copper caps and have never had an issue. You will never have an issue in a speaker crossover as there is no heat. Heat is the only factor of concern with these caps. I use them in tube amps with heat and still no problems. 

I have had them in my speakers and gear since they were introduced perhaps two years ago. So much better than any Mundorf caps based on my considerable cap rolling experience.  They are far more quiet, they remove noise better than the Duelund CAST. They are SOTA for micro details and exhibit a nice airy, natural tone. I prefer them to Duelund as they are more neutral through the upper mids. Very special cap. 
Over the last few days I have had a chance to complete the mods to my crossovers. Thanks to all for their suggestions.

I had to remove the upper bass driver and the mid range FTF driver to get at the crossovers and the wiring. I also removed the tweeter but I will get to that challenge below.

The XOs are mounted on plastic standoffs that are a bear to release the PC boards from. B&W slops glue on top of the barbs that hold the boards to the standoffs, and getting all the glue from the barbs usually means destroying the barbs. I contacted B&W and their CS was kind enough to send me back ups (they are not listed in their parts lists). I ended up pulling the two front standoffs and replacing them with a threaded insert designed for wood. My local hardware store had some that fit the .280" holes perfectly (use 5/16", for #8 screws). With .25" tall spacers, these mounted the boards perfectly.

The HF XO board is mounted on the roof of the midrange chamber, the MID XO is mounted on the floor. To remove the HF XO you have to release the board from the standoffs, then cut a number of tie wraps to release all the wiring. The HF XO is fed with two 12 ga wires (red/black) terminated in gold plated tab connectors (nice). Another pair of 12 ga black/red are soldered to this board and are terminated in a Molex connector that feeds the tweeter. I did find many of the tab connectors to be very loose, so I had to correct that before re-connecting.

I ended up swapping out the Supreme SG 4.7 mFd for an identically sized Supreme SGO. I also pulled the 0.5 OHM tab power resistor (no name on it) and replaced it with a 0.75 OHM 1% 100W MP9100 Caddock thick film resistor that mounted to the OEM heatsink perfectly. I used 0.75 OHM instead of the 0.5 OHM to reduce the tweeter output slightly. B&W originally spec'd a 1.0 OHM which they changed mid production to 0.5 OHM.

For the MID XO, I also swapped out the 2 OHM resistor with a 2.0 OHM 1% Caddock MP9100. I pulled the MKP 47 mFd Mundorf film cap and replaced it with an EVO SGO 47 mFd. This cap also fit quite nicely. I had PartsConnex match a couple to under 1%.

After re-mounting the XOs, I soldered all the gold plated tab connectors to the wire they were crimped to. Unusually, some were soldered by B&W and some were not. B&W also used some sort of grease on the connectors (not necessary for gold connectors IMHO) so I removed that as well. I pulled all three bass drivers to tighten and solder connections.

Now the tweeters. I strongly suggest most users leave these guys alone. They are extremely fragile and you must not touch that diamond diaphragm if at all possible. In mine, I felt a little too much front to back play, so I thought something might be loose. The motor unit/diaphragm is coupled to a bar that runs rearward through the end of the casing tube. At the end of this bar, a soft, gooey gasket material (looks like Sorbothane with the consistency of a gummy bear that has been on a hot sidewalk) surrounds the bar and isolates it from the casing. When you pull the motor and bar forward, you inevitably destroy this gasket as it sticks like crazy to the bar, and turns itself inside out as you pull forward. So order a backup from B&W before you disassemble the guts of the tweeter.

In my tweeter, the plastic coupling nut that holds the bar to the motor was loose. And you cannot tighten this nut unless you pull the guts of the tweeter out.

After all was put back together, I had a chance for some preliminary listening. I have to say the total end result was more pleasing than I anticipated. If I have to use a single word, it would be "smooth". Edginess on horns and vocal sibilants was gone. Bells and symbols are now slightly recessed in volume, but still very clear in position, with more 3d space between them. The overall sound fill left to right is vastly improved, with sound spilling outside the boundary of the left and right speakers (something I have been struggling to get with these 803d's for some time). The midrange seems to be separated better from the bottom end in clarity (this is hard to describe), so the bass transients no longer "mask" some of the midrange, as they had done before.

All in all, I am very happy with the changes.
Um, why not go to active crossovers and get the most control of the drivers and how they sound?  A lot easier than the "black arts" of trial and error using different caps.

Because that route is fraught with just as many variables and complexities. I was going to do that with another previous set of speakers and ended up buying the B&Ws instead.

To do what you suggest  first you need 3 to six power amps, depending on if you use stereo or mono amps. You need to adjust the output levels of each amp with each driver, so you need an RTA/pink noise setup to do it right. Then you need a decent XO setup. Anyone know of discrete component class A XOs that have three XO points? And a choice of first, second, or third order slopes? No? Me neither.  There are XOs out there, but most are op-amp based which is unacceptable to me. Finally, if you are using B&W drivers, you have to have impedance and parameter data for the drivers which is unobtanium unless you measure it yourself.

As I stated I considered building my own speakers prior to buying the B&Ws, and was going to go the full active route. But I could never get over the compromises and costs, plus I cannot design a cabinet like B&W does. 

Thank you for sharing your crossover upgrade experiences - I previously owned the 803D and 803 Diamond speakers - great value high tech speakers with timeless industrial design.

You're welcome. They are sounding exceptional right now. That 47 mFd EVO SGO midrange coupling cap is really making a difference.