This is insignificant in a crossover and can be ignored. interesting post on Audiokarma 12 years ago: Last night, inspired by this thread, I ran a bunch of capacitor curves. I tested six different types from my parts bin, ranging from the cheesiest old lytics to fancy film, and including garden variety audio NPE. All the caps were 4.7uF, within a couple of percent. I hooked each one in series, directly to a well-known 1" dome tweeter, then swept the voltage plot across the speaker terminals.
At 2 or 3 KHz, there was about a 1.5dB total spread between the units, increasing to almost 5dB at 20KHz. Interestingly, (or coincidentally?), the curves clustered into two fairly distinct groups. Also, interestingly, the two data clusters each contained both lytics and films. I will post curves and details when I have time, and will run impedance, ESR and DF curves on all the caps.
This was intended to be a quick, but very demanding test: the tweeter is 4 Ohms, and no other components were used at all. (EG- no pad resistor.) Eventually, I will test a more typical circuit topology.
The lessons, re-learned:
- changing cap types is perilous, and, will probably change the frequency response.
- Modern NPE’s can hold their own with film caps; there is no readily obvious, consistent advantage to either type, though there are various differences. ken kantor, Mar 23, 2009
And what I am describing has nothing to do with capacitive reactance. It is called series resistance for a reason. |
@pesky_wabbit, interesting article thanks. An increase in resistance in series with a DUT will attenuate the signal I agree but the ESR delta between good film caps will have no noticeable effect. Any slight effect will be completely swamped by the room's response which is why I termed it a non-issue.
Certainly I would never add an extra small resistor to the XO. When I design a speaker I try my best to closely match, if possible, the sensitivities of the drivers so that no attenuating resistance is needed resulting in a simpler and cheaper XO. You may be underestimating the harm extra components do.
Vance Dickerson, Joe D'appalito and Sigfried Linkwitz's work is mostly what I studied regarding speaker design. None of these gentlemen even mention ESR when creating an XO from first principles. I think I will take their teachings before
ken kantor.
|
Between good good film caps perhaps, but between old design electrolytics and film caps, possibly quite significant. Way back when it was being discussed on this thread in some depth, various speaker manufacturers‘ philosophies were examined WRT ESR and were found to differ. You have found a couple who concur with your own. There are well respected manufacturers who do not. |
The point I am trying to make is that Ken’s data confirms that you can’t just. throw in a modern film cap in place of an old style electrolytic and expect your frequency response to remain totally in tact. You may get lucky, but I wouldn’t bet on it.
Measured deviations of up to 5db at 20kHz between samples show considerable scope for audible changes.
If you are swapping a modern film with a modern film, your chances of success are far greater.
|
Hey pesky_wabbit stop chasing carrots and educate yourself.
To get an understanding about XO design try reading articles from the experts I mentioned. You are not arguing with me, you are arguing with Sigfied Linkwitz. Yes it's that Linkwitz of Linkwitz/Riley fame. His filters are probably more widely used than Bessel, Butterworth or Chebychev in crossovers.
You state: (ESR) "altering a fundamental characteristic of the crossover which affects frequency response"
For simplicity lets look at a first-order XO. A series resistance will NOT alter this basic function, the position of the XO knee or -3dB point is determined by the VALUE of capacitance, (not the value of capacitance plus ESR) The XO response has now been established and any resistance in series will merely attenuate the level to match the other drivers.
Now what about speakers with level controls for the tweeter? Are you claiming that when you adjust the level setting, which just switches different value resistors in circuit and here we are talking 10 or 20 ohms and not milliohms, that it is changing the XO point?
I consider neither your nor K Kantor's unsubstantiated bloviating worth hijacking this thread further.
|
