It's a pity those PS regulator designs are "closed", commercial types instead of DIY. Even eBay power supplies allow some playing with the discrete parts. That would allow trying discrete parts, particularly capacitors, before and after the regulators, tuning them to compare results. That would allow sharing results among readers here, instead of measurements that are hard to repeat without rather expensive instruments.
Not a critic, just wishful thinking.
Last edited by a moderator:
Power supply regulators. Lots of places on this thread where PS was used as an abbreviation for power supply.
Same thing for a lot of power supply articles around.
Does image of products in digikey are always accurate?
This cap look like old ones
MAL211815102E3 Vishay BC Components | Capacitors | DigiKey
This cap look like old ones
MAL211815102E3 Vishay BC Components | Capacitors | DigiKey
That's interesting, I will have a look later, thanks for the tip...
In the case of these bipolar xover caps, which have same lenght legs, I presume the left ringed side is the "positive"??
Last edited by a moderator:
In the case of these bipolar xover caps, which have same lenght legs, I presume the left ringed side is the "positive"??
Yes. The outer aluminum foil ('NEG') is the one connected to the can.
I found my copy of part 6 of Bateman’s paper published in Electronics World, 1-2003, but thanks for the offer!
I think the primary difference between our measurements is due to the drive level used on the caps, the value of the caps, and the way we quote the results. Bateman is driving a 100µF cap at 100Hz through a 10Ω resistor at a level such that 0.1V is developed across the cap. A 100µF capacitor has an impedance of 15.9Ω at 100Hz, so developing 0.1 V across that cap will require a current of 159mA. My drive levels with a 470µF cap at 2kHz are around 47mA, with about 0.01V (-41.2dBV) developed across the cap, so I would expect my results to show lower distortion, both because my drive current levels are smaller, and also because my cap is larger, so it will develop less voltage across it at a given drive current.
Rather than use dBV, which is my primary measurement data, I need to convert the voltages I measured to the relative ratios that Bateman uses in order to be able to compare the numbers directly. In this particular test, my data says that the 470µF Silmic II cap has -41.2dBV of 2kHz across it, and a 2nd harmonic level of -162.1dBV, and that means that the relative distortion figure of the cap is -120.9dB. His value of -94.4dB is not that far away, considering the 20dB drive voltage difference and the resulting ~12dB drive current difference between our two different test methods.
So, I have no reason to doubt Bateman’s results, but I will say that I am designing no circuits that will put 160mA of signal current into their bypass caps during normal operation. I’m also pretty sure that the Audio Precision APx555 that I use can’t produce 160mA of clean output into any load, and an external amplifier will probably have a higher distortion residual than the APx555, further complicating the issue. Therefore, I cannot reproduce his results directly. Still, the actual numbers from Bateman’s tests, as well as my tests, are useful for relative comparisons of different capacitor types, since they are both reliable measurements of distortion relative to the test equipment residual, and they can provide insight into the relative merit of one capacitor type compared to another.
However, it’s not clear how these high drive measurements will relate to the typically much smaller drive applications in typical audio circuits. To be fair, given the relatively high distortion residual of Bateman’s DIY analyzer, it is important for him to increase the drive level to the caps in order to get the capacitor’s distortion significantly above his analyzer residual, in order to produce reliable data, and he’s managed to do that. But again, it’s not clear how these levels relate to a real circuit, since it depends on how the capacitor is being used. An amplifier’s power supply bypass cap could pass a ripple current equal to the amplifier’s output current, but in a typical home audio circuit at typical signal levels, these currents are usually on the order of 1mA or often much less, sometimes 10-100µA. This is vastly different from a 160mA drive level, so it might not be warranted to be alarmed at easily measurable capacitor distortion figures with these gargantuan drive levels.
To sum it up, I think I’ve shown how Bateman’s numbers and those I’ve obtained are not that far apart, but the issue of “whether it matters or not” is up for debate, depending on one’s application. I measured these capacitors to determine how they would work in a specific circuit I’m designing, with a specific drive level in mind, and by the results obtained, pretty much all of the caps I’ve tested are good enough with regards to distortion - other parameters are more important to me, such as expected lifetime. This is because my circuit uses regulators after the caps that add another 60-70dB of attenuation to these distortion products, and the amplifier stages that the regulators drive have very high (100dB+) power supply rejection on their own, so they are insensitive to this problem by design. Still, I needed to verify the magnitude of this potential issue, and I’m happy with the results.
However, these caps can be applied as power supply rail bypasses for discrete amplifier circuits that have low or zero power supply rejection ratio, so the errors of these caps in those circuits can be directly coupled to the output of such a circuit, making these distortion levels possibly matter a whole lot. Maybe -120dB or -94dB distortion levels aren’t good enough for such circuits, so closer scrutiny of these caps is warranted.
It all depends on how your circuit works, and where you’re applying the caps. Blanket pronouncements that vilify or sanctify caps without any context as to how the caps are used are counterproductive statements.
Last edited by a moderator:
Found the article about snubbers / rectifiers (maybe not all of you have seen it)
http://www.hagtech.com/pdf/snubber.pdf
http://www.hagtech.com/pdf/snubber.pdf
Come on guys. Go and read the Quasimodo thread and buy one of the boards. No math needed and you can determine the snubbers values with a scope. Just turn the pot until the pulse is damped optimally then measure the required resistance. And the snubber is not across the diode it is prior to the rectifier as it knocks out the oscillation caused by the diode switching interacting with the inductance of the transformer winding.