• Disclaimer: This Vendor's Forum is a paid-for commercial area. Unlike the rest of diyAudio, the Vendor has complete control of what may or may not be posted in this forum. If you wish to discuss technical matters outside the bounds of what is permitted by the Vendor, please use the non-commercial areas of diyAudio to do so.

Crossover Curiosity

DavidLS

Member
2007-03-22 4:17 am
As I have written in thread below I have rebuilt my Ellis 1801 crossovers using platinum bypass capacitors (.047 uf), new resistors, and new wire. After about 100 hours of listening I decided to put the crossovers in the boxes I purchased to move the crossovers out of tight space in the speaker. In moving the crossover I broke one of the old tweeter capacitor connections off flush to the face of the cap. No way of re-connecting. Tweeter is dead on right channel.

So I have ordered some mundorf supremes to replace the original capacitor - 8.2 uf - mostly to try something different.

Waiting for them to arrive - today I got the bright idea of using a salvaged soniccraft capacitor from the woofer side (value is 5.1 uf) to get the tweeter going. The capacitor is bypassed by new platinum capacitors and voila everything works great. In fact sound is great.

So that is the curiosity - is there not much difference between 5.1 and 8.2 uf? Is the bypass capacitor playing a key role here (it didn't work on its own). New territory for me so just wanting to learn about what is going on.
 

Jay

Banned
2003-02-11 9:02 am
Jakarta
So that is the curiosity - is there not much difference between 5.1 and 8.2 uf? Is the bypass capacitor playing a key role here (it didn't work on its own). New territory for me so just wanting to learn about what is going on.

I honestly doubt you can hear 47n of anything bypassing 8.2uF MKP. Remember that a cap labeled with 8.2uF may be measured 7.4uF if 10% tolerance is in effect (which is very rare to be honest).

If you hear the 5.1uF sound better than 8.2uF, that's normal, but assuming that you have good ears, it is a sign that the crossover is not perfectly designed. Why?

To make a good speaker crossover for 2 drivers you need to do these 2 things:

(1) The drivers must blend perfectly, they are said to cross each other at certain frequency, lets say 2.5kHz (where for example you need 8.2uF). Biggest issue is to get the phase right between the 2 drivers.

(2) Each drivers must operate in their comfort zones. Most tweeters will feel comfortable crossed as high as possible, away from their resonance frequency. So when you hear the tweeter individually, most probably the sound is less stressed if the tweeter is crossed above 2.5kHz (in your case 5.1uF will do this).

For a normal crossover, in my experience case #1 (driver a little bit stressed) is better than case #2 (drivers don't blend).
 

DavidLS

Member
2007-03-22 4:17 am
Jay - thanks for the reply - good information. Not sure the sound is better - but also not noticing anything off and speakers definately sound better with two functioning tweeters. The speakers are mis-matched the way they are currently set up so the 8.2 mundorfs will get installed this week and will be interesting to see if I can hear any changes to the new caps and balancing the speakers with the same crossovers.

Find this whole crossover construction / components quite interesting.
 
I honestly doubt you can hear 47n of anything bypassing 8.2uF MKP. Remember that a cap labeled with 8.2uF may be measured 7.4uF if 10% tolerance is in effect (which is very rare to be honest).
.

I can hear 10nF bypassing 10uF. While your argument regarding tolerance is correct, I find that two caps paralleled that equal X value do not sound the same as a single cap that equals X value. It is not the subtle difference in value that I hear. It is the completely different cap that I have added to the stack. Even the very small values.

If you hear the 5.1uF sound better than 8.2uF, that's normal, but assuming that you have good ears, it is a sign that the crossover is not perfectly designed. Why?

To make a good speaker crossover for 2 drivers you need to do these 2 things:

(1) The drivers must blend perfectly, they are said to cross each other at certain frequency, lets say 2.5kHz (where for example you need 8.2uF). Biggest issue is to get the phase right between the 2 drivers.

I agree. If that much value change does not hurt the speaker, you either have some serious issues with the setup, room, etc..., or the resulting phase rotation does not hurt the design because the design does not track phase very well.

I believe that David is just starting to bring the speaker around, so it would not surprise me if was not really all that "close" to these speakers anyway. Meaning, he is just know getting to know them.
 

Jay

Banned
2003-02-11 9:02 am
Jakarta
.

I can hear 10nF bypassing 10uF. While. your argument regarding tolerance is correct, I find that two caps paralleled that equal X value do not sound the same as a single cap that equals X value. It is not the subtle difference in value that I hear. It is the completely different cap that I have added to the stack
You are correct. You have very good ears. What i was saying is that such good ears are rare. I always bypass big cap with series cap in top speaker. Also there is burning in time until we can hear the real sound after chsnging caps.
 
Maybe a Summary from Dave

I realize this is a very old post, but others may have the inclination to read and discover herein. So I will take the time to comment. I didn't think this would become a long response, but my response seems be getting very long. Simple issues are not always super simple.

I suppose my comment will be a partial summary and a partial opinion based on experience. I do agree with most of what is written above, but perhaps future readers might want to know my opinion on the matter.

First, I did match capacitors using non-calibrated capacitance meter. While this method would not meet scientific rigor objectively, I was able to PAIR capacitors having similar values to the .01uf level. A batch of 20 capacitors might have values ranging from perhaps 7.97 to 8.31. I wrote my measured values on the capacitors, then matched them with respect to values. I would then match capacitor 7.97uf with capacitor 8.02uf - for example. I did this because it was easy and I am slightly OCD in many regards.

Second, I never performed an a/b experiment listening to a speaker with roughly matched capacitors, then comparing this with a speaker with unmatched capacitors. I suspect the impact would be very minimal, with perhaps some imaging impact. But, I haven't performed this experiment, so I really don't know.

Third, I couldn't measure the amplitude effect of adding a .1uf bypass capacitor, but it was VERY audible and not always good.

Fourth, capacitors DO sound different (with and without bypass), and these differences are often accommodated via amplitude to accommodate the balance that most people prefer in their living room... oh my the acoustics. The very best a designer can do with their "perfect" crossover design using the very best drivers and crossover components is to target a the human preferential "average" then declare their work complete. In this realm, a guy once told me that loudspeaker "sins of admission are worse than sins of omission." I think he was correct.

Fifth, specifically, swapping from a 8.2uf capacitor to a 5.1uf capacitor WILL have an impact measurably and audibly. The speaker will work and might still seem pleasant. There IS a chance the impact might sound better in the acoustic environment. If a customer likes it better - great ! But, everything that departed my workshop was measured and tested. A 3uf change will result in amplitude ripple and phase smear.

Last, on my soap box, the general quality control level present in most loudspeakers is... bad. There are a only handful of people that measure components, measure loudspeakers, and listen to their product before it's sold. This effort takes considerable time, and costs $. THE former Washington D.C. audio enforcement and consensus guy (to remain nameless) conveyed that he spent most of his time working on food and drug issues because nobody cared about audio. I think his sentiment is true. Most customers FEEL better saving perhaps 20%. I would rather pay 20% more for a product that has been tested.

On quality control in loudspeaker woofers and diapers.... The raw chemical in the barrel affects what emerges from the mold.

After measuring several hundred high quality woofers over a 5 -7 year span, I became very irritated with changes in the response from the woofers. Batch-to-Batch the woofers were not the same and these changes were so dramatic that I had to redesign the crossover circuit. I probably could have just allowed the amplitude dip 2-3db at 2500 hz, but I just wasn't willing to do that. I was comforted to learn that my lament was not isolated.

I recently visited with a retired 3M chemical engineer at the gym. This gentlemen was intimately involved in the development of hook and loop materials from 3M that arrived in many places - mostly diapers. He conveyed that he would test ALL of the raw material that arrived in barrels that would eventually find its way into the hook and loop fabric. The effective production of this material necessitated extensive quality control in the raw materials used - just like loudspeakers.

So, know that correct control of the bottom-end (woofers and diapers) is rooted in the uncommon practice of tedious quality control.