SpeakerTalk

[rx7rotary]

Hi again,

I am looking to build external passive xover using the jantzen or mundorf type poly capacitors in external enclosure, then tri wire back to drivers.

Any thoughts, ideas or even if you know of anyone that has done this what were the results?

Best Regards
Mark

[audiolabtower]

Many years ago Martin Colloms and Paul Crook did some articles in the Hi Fi press about bi-wiring and suggested taking the crossover to the amp terminals was the equivalent of very expensive speaker leads since the interface from crossover output to drive unit was not as critical in their tests as the normal configuration, and had subjective benefits.

I tri-wired my Spendor BC1s and took the crossovers to the amp terminals and was convinced there was an improvement in detail, bass control and stereo image. My opinion is that it works best on BBC style speakers where excellent pair matching and good imaging and depth with low colouration are there to start with.

Later on I got my 105s and replaced all the caps with parallelled-up polyesters and polyprops to less than 0.25% of the measured originals, and put the crossovers under the 3 power amps tri-wiring the 105s (actually quad-wiring if you count the led circuit). I have no blind tests so can only offer an opinion but again the detail, depth/"blackness' of the acoustic, and subjective lack of distortion or clarity increased.

The Spendors had polyester caps to start with and needed no tweaking. The Kefs need some resistance added to the crossover outputs to sound balanced when taking away the higher dissipation factor of the electrolytics. I suggest around 0.5 ohm for midrange and around 1 ohm for tweeter. I ran them without these for a few years with tone controls at +2db bass and -2 db treble in my room for much the same effect. If you have separate power amps the gains can also be tweaked.

Unfortunately replacing caps without measurement of originals to standard tolerance (even tight tolerance) solids is likely to give much higher response errors than due to DF due to the value differences. The originals were hand measured and computer sorted to balance out tolerances of all components. Reverse the high and low values in a circuit could easily change the total response shaping away from flat.

[iso]

Did you use series resistors between crossover and drive units or in series to each plastic film capacitor?

Lower series resistance of plastic film capacitor has some effect on crossover slopes, especially if steep slopes are used. Hopefully somebody with more experience can tell how much this will be in practice. Otherwise it seems odd if we replace original electrolytics with lower ESR plastic films and add series resistance to increase ESR enough to make crossover calculations correct. ESR is only one characteristic of capacitor, different dielectric might also have some influence on sound quality.

I replaced only HF section electrolytic caps of my Cantatas with plastic films, as Colin recommended. This seemed to work nicely, even I thought that MF LL electrolytics might be problem as there are 50uF in series to signal and 10uF and 7uF parallel to signal.

Best Regards

Kimmo

[audiolabtower]

I found a single resistor between crossover output and drive unit was enough to get the balance I was used to for mid and treble. I did not bother with the bass unit even though I used 12 x 30uFs in parallel plus padding for the series 360uFs - at 40Hz the 360uF would be around 0.5 ohm but I preferred to let the speaker leads provide some resistance, calculated at 0.05 ohm for my leads - possibly not enough? Incidentally the 3 x 120uFs in my original crossover measured 372.5 uF on one channel and 372.6 uF on the other channel which shows the lengths Kef went to in matching - did any other manufacturer go to this trouble?

The DF varies with frequency anyway so an approximate value based on the crossover freqs and an assumption of DF <5% was used as a starting point. I thought that the series caps were more critical generally.

The dielectric "sound" I would say is a second order effect compared with the wrong balance due to freq response being significantly changed by altering crossover values and damping etc.

Best regards,

[speakerguru]

[iso]

If you do think that dielectric sound is second order compared to other issues... did you replace caps only due to the better service life of plastic film capacitors?

I do believe that different capacitors may... or may not... sound different in different places. Results have been quite unpredictable. I tried to upgrade my CD-players Elna Silmic output electrolytics (as far as I remember 2 pcs 50uF caps in series) with plastic films and stock Elnas always sounded better. I even removed output caps and used CD-player DC coupled, as DC offset was very low. Elnas electrolytics sounded better than DC coupled output. In Quad 22 feedback loop tantalums outperformed electrolytics...

Speakerguru said some time ago that peoples usually buy thing by eye. It might be reasonable to add by ear... please read my "should I concentrate on my vintage watch collection" tread in introductions forum. I find this quite natural, as there is so much information available about simple capacitor that audiophile without degree in audio engineering has hard time to consider, how he should value each parameter for given application.

Things get much tougher when you should value amplifier where whole circuity, lay out and maybe 60 resistors, 30 caps and several semiconductors, tubes, mechanical parts will make this thing sing or suck. And somebody tries to explain this with figures like 70W/ch/8ohm, 0,05 THD, DF 50... And when source, pre amp, power amp, speaker matching is done, similar problems will arise.

I do not mean that you should ignore technical aspects of sound, as they are needed when you try to make any high performance product and are basis of sound you hear. I do mean that you should look and hear, before you can be shure if this is the product that you want to live with.

I have said this before, but 105:s have been the best buy audiolabtower has ever made.

Best Regards

Kimmo

[audiolabtower]

My point was that simply replacing a crossover cap that has been selected within 1% to give +/- 2dB flat freq response, with any 5% type which could randomly be 9% away from the original could easily give 1 or 2 db variation in the freq response. E.g. tweeter caps where input cap is deliberately smaller than output cap even though both are marked same uF. A reversal of relative values will boost or cut and lead to brighter or duller results irrespective of dielectric I would guess and therefore be a first order effect. If you replaced caps matched to 0% with same ESR etc for exact same freq response then the differences are second order compared with boosting or cutting freq bands?

By the way I replaced caps in my D/A too

[rx7rotary]

[audiolabtower]

I have no experience of Jantzen caps but would say that I think series caps are the most important, paticularly the tweeter. I highly suspect that removing the crossover from the box is perhaps even more important than the type of cap so perhaps microphony is an important consideration for caps inside the enclosure?

[rx7rotary]

[speakerguru]

Have you ever tapped on a pcb while the gain was turned up? Everything is microphonic. You just have to have enough gain.

Whether you hear it under any particular set of circumstances is another matter.

[rx7rotary]

[iso]

How early you did test microphonic effects of capacitors at Kef.

How important you did find microphony to be in passive crossovers. There is generally no gain in passive crossovers, but the vibration input is highest possible in home audio. Audiolabtowers idea to relocate the crossover might be good choice because this.

PCB:s seems to be cause of much trouble due microphony, especially with tube amplifiers. There is 20 pc 2uF - 30 uF PP caps and 28 pc 0,47uf - 2 uF polystyrene capacitors + maybe 20 smaller polystyrenes in my Conrad Johnson Premier 7B preamplifier. Electrolytics are used only in heater power supply. All bulky plastic caps (2uF polystyrene is quite similar size as 30uF PP) in the main amplifier chassis are glued to PCB most likely due microphony. In power supply chassis 2 pcs cable ties are used to fasten 30uF caps on the PCB and only 1 cable tie is used for smaller valued caps. This is quite lot of work and expense on caps. But... if you look at the PCB design, there is quite long PCB track between grid input pins and grid series resistors. Peoples buy with their eyes... caps are visible and very few will look how PCB has been constructed.

Anyway, it would be interesting to know how much microphony will degrade the signal in passive crossovers? I do have no commercially built speaker which has used any precautions to reduce microphony of caps or parts other than varnishing the chokes. I do know that Tannoy has made claims that they have used damping on Hovland caps on their Prestige series speakers. Glue has also been used to fasten components with point to point wired crossovers.

Best Regards

Kimmo