The problem is usually the wiring/trace inductance from the the bigger caps (the two 100uF 'lytics in the MyRef FE) forming a tank circuit with the smaller cap (the 10nF rail-to-rail). The inductance from the wiring of the 10nF is quite unimportant unless it is excessivly large.
The lower the series impedance of the 100uF (or the shunt supply or whatever) and the smaller bypasse, the worse it gets. And the farther apart the capacitors are in values the worse it gets, too.
Example : effective resistance at the 100uF's+supply of 0.1R, some 50nH inductance from wiring, 10nF with 0.1R, too ==> impedance peak of 20Ohm at 6.5Mhz, drops back to 0.1R at 15Mhz, then inductive rise above that, again.... easy to observe with LTspice (just inject a 1 amp AC current source at the output and plot node voltage which scales 1:1 to impedance).
Those are critical frequencies for this (and most any) audio opamp circuit.
Simple way to test it if you have a pulse/square generator and 'scope. Inject a current pulse/step via a good capacitor (big silver mica) and clean wiring into the rail and observe the pulse/step on the scope for ringing....
The lower the series impedance of the 100uF (or the shunt supply or whatever) and the smaller bypasse, the worse it gets. And the farther apart the capacitors are in values the worse it gets, too.
Example : effective resistance at the 100uF's+supply of 0.1R, some 50nH inductance from wiring, 10nF with 0.1R, too ==> impedance peak of 20Ohm at 6.5Mhz, drops back to 0.1R at 15Mhz, then inductive rise above that, again.... easy to observe with LTspice (just inject a 1 amp AC current source at the output and plot node voltage which scales 1:1 to impedance).
Those are critical frequencies for this (and most any) audio opamp circuit.
Simple way to test it if you have a pulse/square generator and 'scope. Inject a current pulse/step via a good capacitor (big silver mica) and clean wiring into the rail and observe the pulse/step on the scope for ringing....
Last edited:
I may have asked this question recently but I'll do it again.
Is anyone aware of a scope tool that is PC based and is stable and powerful enough to give accurate and useful results? This is of course a cost consideration on my part. I'm aware of some of the deals on eBay for real hardware units if I have to go that way.
Is anyone aware of a scope tool that is PC based and is stable and powerful enough to give accurate and useful results? This is of course a cost consideration on my part. I'm aware of some of the deals on eBay for real hardware units if I have to go that way.what would a typical ringing/overshoot sound like? wikipedia states that it is hard to describe in words http://en.wikipedia.org/wiki/Ringing_(signal)
http://en.wikipedia.org/wiki/Ringing_(signal)
Bob,
Try to get a decent fully tested used 2ch. analog oscilloscope with 1:1/10:1 probes from a used test equipment seller, local surplus store etc (ebay deals can be problematic at times). 20Mhz bandwidth is not ample but OK for the most part (50Mhz++ is better). Then get some reading on 'scope usage and measurement and play with it. It will be an eye opener, literally.
Soundcard-based 'scopes are OK too for some first insight for zero cost (check out that Visual Analyser, a current thread here atm -- a very complex tool alas not perfectly suited to beginner's needs) but are VERY restricted (only a few kHz) in bandwith and typically not that easy and convenient to operate.
PC-based signal generators are OK too, but again restricted in bandwidth. No way to make square waves or pulses with steep edges.
Try to get a decent fully tested used 2ch. analog oscilloscope with 1:1/10:1 probes from a used test equipment seller, local surplus store etc (ebay deals can be problematic at times). 20Mhz bandwidth is not ample but OK for the most part (50Mhz++ is better). Then get some reading on 'scope usage and measurement and play with it. It will be an eye opener, literally.
Soundcard-based 'scopes are OK too for some first insight for zero cost (check out that Visual Analyser, a current thread here atm -- a very complex tool alas not perfectly suited to beginner's needs) but are VERY restricted (only a few kHz) in bandwith and typically not that easy and convenient to operate.
PC-based signal generators are OK too, but again restricted in bandwidth. No way to make square waves or pulses with steep edges.
The ringing we talk about here is the ringing of the supply at MHz frequencies, way way above the audible range.
It depends so much on the specific opamp(s) and circuit/layout details if/how this tendency of ringing "will sound", that is, affect music signals. Often it just may sound harsh and noisy or even fully distorted when the circuit is going havoc (or even already frying), but the effects may range from none, to subtle ocasional "loss of precision" (in lack for a better term and "precision" being quite multi-faceted) to fully and clearly audible severe degradation...
Wow - VA looks very sophisticated and useful. Thanks for the tip.
Do you see much variation between a new cap versus after 100hrs or more run in?
A new cap is effectively one that has been freshly reformed. This applies irrespective of whether it was manufactured yesterday, or last week, or last month, or a year ago. If it has been on the shelf and/or in transit for longer than 24hrs, then it must be reformed immediately before testing/measuring.
Once you compare a freshly reformed "in spec" cap to a 100hrs in operating mode also "in spec" you should find that there is no difference.
If you pick an "off the shelf" cap that is leaking "out of spec" because the insulation has degraded, then it almost certainly will be different. It is "out of spec" and by definition it is different.
Once you compare a freshly reformed "in spec" cap to a 100hrs in operating mode also "in spec" you should find that there is no difference.
If you pick an "off the shelf" cap that is leaking "out of spec" because the insulation has degraded, then it almost certainly will be different. It is "out of spec" and by definition it is different.
Last edited:
No, as the series inductance (ESL) of any cap is only dependent on its geometry, caps from the same maker which have the same dimension (can size and lead spacing) have identical ESL. ESR is mostly dependant on temperature, though.
An old thread about X and Y caps.
http://www.diyaudio.com/forums/power-supplies/178240-x2-ac-line-capacitors.html
Thanks Kevin. I notice change in sound after 100hrs of operation, but the change is really small. This caps are often used to prevent noise from coupling into the mains power line and making an antenna out of those lines. Interesting to see people reporting change in value to the tolerance limits in as little as 1000hrs.
X & Y capacitors are self healing. That is one of the basic characteristics that allow them to be used on the Primary side.
As a result of self healing any transient that "burns out" a section of the capacitor must become a non capacitor area.
Each overloading transient will reduce the area of the capacitor's plates.
It is a natural consequence of overloading a self healing X or Y capacitor that the capacitance will reduce.
The worse and/or more frequent the overloading the quicker that the X or Y cap will lose capacitance.
If you read the X, Y cap info you will find all that information either directly or by inference.
Learn to read.
As a result of self healing any transient that "burns out" a section of the capacitor must become a non capacitor area.
Each overloading transient will reduce the area of the capacitor's plates.
It is a natural consequence of overloading a self healing X or Y capacitor that the capacitance will reduce.
The worse and/or more frequent the overloading the quicker that the X or Y cap will lose capacitance.
If you read the X, Y cap info you will find all that information either directly or by inference.
Learn to read.