I have come across many references to the use of bypass caps. Can someone explain in simple language to a novice why they are used and what do they do? Can all crossovers benefit? How does one calculae a suitable value? etc etc etc
Bypass and decoupling are often intermingled terms, could you be more specific in how the caps are going to be used...
If you mean a smaller cap bypassing a bigger one the general consensus is it can cause more problems than it solves.
Bypasing device pins is another ball game all torgether, though I suspect the first if you are thinking of crossovers...
Some reading... sort of audio based, bypassing a large electro or similar with a smaller cap is generally only done in the audio world... lead or trace inductance usually makes the smaller cap useless.
Long thread with some interesting links.
http://www.eevblog.com/forum/projects/bypassing-large-electro's-with-smaller-caps/
If you mean a smaller cap bypassing a bigger one the general consensus is it can cause more problems than it solves.
Bypasing device pins is another ball game all torgether, though I suspect the first if you are thinking of crossovers...
Some reading... sort of audio based, bypassing a large electro or similar with a smaller cap is generally only done in the audio world... lead or trace inductance usually makes the smaller cap useless.
Long thread with some interesting links.
http://www.eevblog.com/forum/projects/bypassing-large-electro's-with-smaller-caps/
Last edited:
I disagree with that consensus. 🙂
No, not all crossovers, depends on the basic cap.
Usually around 1% to 4% of the original. So for a 10uF cap, bypass with around 0.1uF or so.
The only way I know of to tell is to simulate AND listen. I use simulation to ensure my changes (uF and ESR) are minimal to the tonal balance AND impedance curve. Then I listen to see if there's any point.
Especially in the 20th century, bypass caps were used because film caps were way too expensive. Electrolytics were used instead, but because of their construction, they get inductive. That is, impedance changes, rises, as F goes up. A small value bypass cap was a cost effective way of making the capacitor in the crossover behave more "ideally." The electrolytics also had horribly high ESR, around 1-2 Ohms, vs. 0.3 Ohms for films.
Now, electrolytic caps are better, but film caps are cheaper, so almost everyone uses at least an inexpensive film cap in crossovers now.
The idea among boutique snake-oil sellers (like me) is that you can get a cap to work more ideally by either using a bypass cap which is much better, or using multiple smaller caps to build up your larger cap (5uF+5uF= 10uF).
Not that I like these brands, but one thing maybe done is to use Mundorf Supreme's as the main, and then buy Mundorf SIO EVO as bypass caps, to try to approach the sound of the latter for the price of the former.
Many disagree that there's any benefit to this approach at all, or that electrically, with vastly different ESR's you could be doing more harm than good, or what you are hearing is introducing non-linearity.
You may still see this in commercial speakers but usually for caps in the woofer, where you need a very large uF (> 20uF, sometimes even > 100uF) value, and a film cap would be super expensive. You'll see a nice bi-polar electrolytic for 100uF with a small film cap 1uF as bypass, or something. Ideally, this type of matching should be done paying careful attention to the ESR across the audible spectrum. It really helps to have something like the Dayton DATS to measure this instantly.
So, I would say, experiment for yourself, inexpensively. 🙂
Best,
E
No, not all crossovers, depends on the basic cap.
Usually around 1% to 4% of the original. So for a 10uF cap, bypass with around 0.1uF or so.
The only way I know of to tell is to simulate AND listen. I use simulation to ensure my changes (uF and ESR) are minimal to the tonal balance AND impedance curve. Then I listen to see if there's any point.
Especially in the 20th century, bypass caps were used because film caps were way too expensive. Electrolytics were used instead, but because of their construction, they get inductive. That is, impedance changes, rises, as F goes up. A small value bypass cap was a cost effective way of making the capacitor in the crossover behave more "ideally." The electrolytics also had horribly high ESR, around 1-2 Ohms, vs. 0.3 Ohms for films.
Now, electrolytic caps are better, but film caps are cheaper, so almost everyone uses at least an inexpensive film cap in crossovers now.
The idea among boutique snake-oil sellers (like me) is that you can get a cap to work more ideally by either using a bypass cap which is much better, or using multiple smaller caps to build up your larger cap (5uF+5uF= 10uF).
Not that I like these brands, but one thing maybe done is to use Mundorf Supreme's as the main, and then buy Mundorf SIO EVO as bypass caps, to try to approach the sound of the latter for the price of the former.
Many disagree that there's any benefit to this approach at all, or that electrically, with vastly different ESR's you could be doing more harm than good, or what you are hearing is introducing non-linearity.
You may still see this in commercial speakers but usually for caps in the woofer, where you need a very large uF (> 20uF, sometimes even > 100uF) value, and a film cap would be super expensive. You'll see a nice bi-polar electrolytic for 100uF with a small film cap 1uF as bypass, or something. Ideally, this type of matching should be done paying careful attention to the ESR across the audible spectrum. It really helps to have something like the Dayton DATS to measure this instantly.
So, I would say, experiment for yourself, inexpensively. 🙂
Best,
E
Last edited:
That neatly encapsulates both ends of the sensible spectrum. As long as you measure the results you'll be fine.
Thanks guys. I dont have any plans for a new project right now, was just wondering about it since i keep reading posts about it with no explanation. Maybe something worth considering for a future project. Cheers!
- Status
- Not open for further replies.