Except that trying to twist the two wires on a big capacitor will result in bigger length of cable from the capacitor ends to the twist starting point, than what you would have if you just place the caps on top of the board and secured them with some other means like a horizontal wood plank above the pcb or something.
I posted that your idea was spot on -- very thin insulation means smaller loop size, and more subtractive inductance.
The formulas from the article I posted shows that the wires can become too close together and the inductance can become negative.
Twisting the wires also decreases the loop area by making smaller loops. Ethernet/UTP seems to use both ideas -- If I were to hazard a guess, I would say that the thickness of the insulation in UTP matters a great deal.
The formulas from the article I posted shows that the wires can become too close together and the inductance can become negative.
Twisting the wires also decreases the loop area by making smaller loops. Ethernet/UTP seems to use both ideas -- If I were to hazard a guess, I would say that the thickness of the insulation in UTP matters a great deal.
Hi Andrew
I thought up to now we had only been talking about twisting the wires - please supply a link to where you discussed another method/solution.
In looking at the board, I don't see any difference between using an on-board and an off-board 0.22 uF capacitor, provided that:
1) the wiring to the capacitor is *very*tightly coupled,
2) possibly twisted, and
3) the wiring is connected to the solder pads that are closest together.
While the math indicates that twisting is irrelevant (point 2) if the wiring is tightly coupled and insulation is the correct thickness, twisting reduces the starting inductance and therefore the errors in selection of insulation and it's thickness.
Re point(3): The math also indicates that properly selected insulation and thickness could null *all* inductance in this part of the circuit, including the ESL of the capacitor leads and PCB traces.
AndrewT, you've had a really good suggestion there!
I need to order that LCR meter after payday.
1) the wiring to the capacitor is *very*tightly coupled,
2) possibly twisted, and
3) the wiring is connected to the solder pads that are closest together.
While the math indicates that twisting is irrelevant (point 2) if the wiring is tightly coupled and insulation is the correct thickness, twisting reduces the starting inductance and therefore the errors in selection of insulation and it's thickness.
Re point(3): The math also indicates that properly selected insulation and thickness could null *all* inductance in this part of the circuit, including the ESL of the capacitor leads and PCB traces.
AndrewT, you've had a really good suggestion there!
I need to order that LCR meter after payday.
So, I'm trying the 0.1 uF bypass cap. I recorded a few seconds of a track via audio diff maker first, as a base line to create a difference track tomorrow, then without the caps, and created a difference track. Pretty huge difference -- only 24 db down.
During the plugging and un-plugging, I noticed a problem.... a Furutech 903G female RCA had deposited it's center plug onto a Furutech maleRCA jack. I was unaware of that, but wondered why that particular rca was so hard to fit, and there were issues with getting the right channel to work. Turned out, I had poked the signal connection of the right channel input right through the back of the Eichmann female RCA!
It would barely make contact, but it "worked".
So, the benchmark without the bypass cap is blown. This will now be the new benchmark. Not sure about the sonics with the bypass right now. (MKP1837)
During the plugging and un-plugging, I noticed a problem.... a Furutech 903G female RCA had deposited it's center plug onto a Furutech maleRCA jack. I was unaware of that, but wondered why that particular rca was so hard to fit, and there were issues with getting the right channel to work. Turned out, I had poked the signal connection of the right channel input right through the back of the Eichmann female RCA!
It would barely make contact, but it "worked".
So, the benchmark without the bypass cap is blown. This will now be the new benchmark. Not sure about the sonics with the bypass right now. (MKP1837)
Last edited:
I think it was in one of the My Ref Threads.
They were installing a big capacitor where there was space for a small cap.
The pair of wires that form the inductive loop are the Flow of Signal and the Return of Signal.
If we assume that the capacitor is in the Flow of Signal wire then the Return wire must lie parallel to the flow route through the capacitor and the gap from the return wire to the flow route through the capacitor must be minimised.
A big capacitor limits our ability to minimise the gap.
That's why I suggested introducing more than return wire.
Placing them diametrically opposite each other, with the capacitor in between. With respect to a distant field one loop has +ve phase and the opposite has -ve phase. These inductances thus substantially cancel leaving a minimised inductance of the flow and return pair even though the "gap" remains big.
Last edited:
OK thanks, I think I found it here:http://www.diyaudio.com/forums/chip-amps/234032-my_ref-fremen-edition-build-thread-tutorial-6.html#post3523898
The article I referenced on inductance assumes that the distance between the wires, d, is much larger than the radius of the wire.
This is not so when they are closely coupled.
When very very closely couple, the math is somewhat different. It shows that the inductance at High Frequencies, which is where we are concerned about oscillation, is very very close to zero.
The equation for inductance L is of the form L = k*l*ln(x). where x = (d/2r + sqrt((d/2r)^2 -1) ), d is distance between wire centers and r is the radius of the wire. k includes other constants, and l is the length of the wire run.
As d -> 2r, we have x -> 1 and ln(x)-> 0 which means L -> 0.
Bravo AndrewT!
This is not so when they are closely coupled.
When very very closely couple, the math is somewhat different. It shows that the inductance at High Frequencies, which is where we are concerned about oscillation, is very very close to zero.
The equation for inductance L is of the form L = k*l*ln(x). where x = (d/2r + sqrt((d/2r)^2 -1) ), d is distance between wire centers and r is the radius of the wire. k includes other constants, and l is the length of the wire run.
As d -> 2r, we have x -> 1 and ln(x)-> 0 which means L -> 0.
Bravo AndrewT!
Last edited:
Hard core theory:
Includes formulas for two rectangular bars... or rectangular wire, which have less self inductance than round wire.....
http://www.g3ynh.info/zdocs/refs/NBS/Rosa1908.pdf
Includes formulas for two rectangular bars... or rectangular wire, which have less self inductance than round wire.....
http://www.g3ynh.info/zdocs/refs/NBS/Rosa1908.pdf
I tried just the 0.1 bypass cap, but the bass suffered.
With the bypass cap, and FT-3 bank,the bass returned, but the treble was still a bit forward. The tonal balance is more pleasant to my ear without the bypass. The bypass is a bit "cleaner" sounding at the expense of detail. The FT-3 is a nice sounding cap. The meter will be ordered shortly and numbers will be put to the harness.
Would these changes be more evident @ 600ma or @200ma levels ?
With the bypass cap, and FT-3 bank,the bass returned, but the treble was still a bit forward. The tonal balance is more pleasant to my ear without the bypass. The bypass is a bit "cleaner" sounding at the expense of detail. The FT-3 is a nice sounding cap. The meter will be ordered shortly and numbers will be put to the harness.
Would these changes be more evident @ 600ma or @200ma levels ?