Subjectively, I can hear ferrite beads on interconnect cables as having a negative impact on sound. While it's hard to explain in words, to me personally it sounds like extra overall hash has been introduced and maybe even some comb filter effect.
If RF is not a problem in 99.9% of the cases, then why introduce something that some careful listeners object to?
If RF is not a problem in 99.9% of the cases, then why introduce something that some careful listeners object to?
Thanks Daniel.
I visited AD’s education section
Search Education | Education | Analog Devices
One of the many subjects there that fit nicely in this thread I guess is this
http://www.analog.com/media/en/trai...-op-amp-articles/EEE-Ray-Stata-Speaks-Out.pdf
(I am in doubt as to which emoticon to choose)
George
Ditto.
Something is just wrong with ferrite beads on a lot of things. But they're a big benefit in my CDP on certain ribbon cables.
But it is a problem almost everywhere in audio. You do not need to hear radio stations or cell phone ring from speakers to warn you have issues with RFI. It is much more subtle. In case you hear differences after changing your link cables etc., you may be sure that EMI/RFI is the case.
Hi ridikas,
So I guess you must really enjoy the sound of a system suffering from some level of RFI and any reduction in that sounds odd to you. No problem, you're not the only one who gets use to a sub-optimal situation. After all, sounds good to you doesn't mean the sound is accurate.
I used to get CD players that wouldn't operate yet functioned properly at the shop. Turns out that radio stations in the area were strong. Tinfoil really did control the problem well enough to listen to music on a CD for them. The CD players? Philips and Philips transport based units (they had a plastic case with the only metal as the top cover).
-Chris
Must be an extremely high value of inductance, really extreme and probably special order.
No, that figure is incorrect. If you have a cell phone, you do have an RFI problem. The same holds true for wifi. I'd say that 99.9% of people and even higher have an RFI problem. RFI never does good things for any system as individual components have enough bandwidth to respond even if the system cut-off is a couple hundred kilohertz.
So I guess you must really enjoy the sound of a system suffering from some level of RFI and any reduction in that sounds odd to you. No problem, you're not the only one who gets use to a sub-optimal situation. After all, sounds good to you doesn't mean the sound is accurate.
I used to get CD players that wouldn't operate yet functioned properly at the shop. Turns out that radio stations in the area were strong. Tinfoil really did control the problem well enough to listen to music on a CD for them. The CD players? Philips and Philips transport based units (they had a plastic case with the only metal as the top cover).
-Chris
It's just a bead from Newark, no special order. Not everyone hears the same and that's perfectly okay. Those who can hear these differences must have better hearing, or just know what to listen for.
But these are small differences. Would be very hard to tell in a double blind test listening to an unfamiliar system. But at home, I can hear it no problem.
Just use good MIT cables, no RF!
But these are small differences. Would be very hard to tell in a double blind test listening to an unfamiliar system. But at home, I can hear it no problem.
Just use good MIT cables, no RF!
Hi ridikas,
Even I have been working steadily to improve sound quality for more than 40 years, and I'm sure as hell not the most experienced by a long shot. Never would I presume to know more about sound quality than most people on this site. You did notice the name of this web site, didn't you?
-Chris
I am fairly certain that most people here can define "better sound" easily. Your assumptions are pretty arrogant considering the company we have here.
Even I have been working steadily to improve sound quality for more than 40 years, and I'm sure as hell not the most experienced by a long shot. Never would I presume to know more about sound quality than most people on this site. You did notice the name of this web site, didn't you?
-Chris
If you read some posts above, I was not the one questioning the definition of better sound.
I've been attending live symphony orchestra and jazz 2-3 times a month in Chicago for the past 20 years. So I'm able to do subjective comparisons quite well.
Most of the time better overall measuring components do sound better, but some of the selective strickness around here has no correlation IMO.
I've been attending live symphony orchestra and jazz 2-3 times a month in Chicago for the past 20 years. So I'm able to do subjective comparisons quite well.
Most of the time better overall measuring components do sound better, but some of the selective strickness around here has no correlation IMO.
This is sort of a fallback when you absolutely must have the chassis at PE because your supply isn't designed class II. But it relays the problem to the other components in your system and when you have two devices with audio GND at PE (even if remote via the typicall R//D//C GND "isolators") and unbalanced interconnects you're asking for trouble.
Best practice both for EMC and audio signal integrity is making the common-mode supply impedance as big as possible (class II supply with a transformer with low interwinding capacitance plus generous common-mode supply filters, preferably all in a seperate shielded sub-chassis) while making the audio GND interlink between devices as low impedance as we can (lowest possible shield impedance), so all the equipment audio GNDs are bonded tightly together. Getting low mains coupling with SMPS is tough as most textbook designs have an ~1nF EMC capacitor right from (rectified) input to output. Only overspec'd medical units may apply for better when off-the-shelf supplies are used.
Even better yet, use transformers with shields (but still having lowest coupling capacitance to it) and provide an extra grounding post for an additional low-Z path -- thick ground wire -- between devices so that this path is again much lower Z than the interconnects so they don't carry the bulk of balancing current that is left. This path also may go directly to a massive GND plane or grid, known as ZSRP (Zero Signal Reference Plane).
Having the bulk supply in an extra box which may be PE'd and connecting it to the main unit via umbilical cable is a good idea (distance is our friend when it comes to magnetic fields), at least in a no compromise design scenario, where we also might go into the trouble to use mu-metal shield (carefully designed, since mu-metal is saturated easily).
Also make sure the balancing current (both from the supply as well as between inputs and outputs) doesn't flow *accross* the PCB. This means that the GND "star point", including that of the supply, is where all the connector backshells are bonded together onto the chassis (360° bonds, that is, only non-isolated RCA sockets). This also means all connectors should be tightly crowded together (except mains inlet, where some distance is required).
If the design must cope with unshielded cables (which are in use for audio interconnects, sadly) then there is no way around differential mode filtering on inputs and outputs. We can avoid (bigger) ferrites by using proper air-core inductors paralleled by resistors but such a L//R will be capacitive at some point and hence a path the RF will take, so better add at least a small ferrite bead right at the connector center pin. Shunt caps are needed anyway. Of course some strategic low capacitance shielding may be required to really keep the RF out.
Guessing values and layout seldom works unless for the extremely experienced with strong RF background, you need to test and measure everything rigorously.
Finally, bundle all signal cables (and the mentioned additional GND link) between equipment to reduce loop area prone to pick up magnetic fields.
All these things are well known by the instrumentation guys and outlined in every book about signal integrity and solid EMC design, like Henry W. Ott's "Electromagnetic Compatibility Engineering".
And for EMC/ESD testing ideas details, websites like D.C.Smith's emcesd.com are a treasure.
Once we have these things settled we can go ahead and start reasoning about the actual audio circuitry.
Last edited:
Hi ridikas,
I quoted you exactly. You said it, my comments pertain to your comments with regard to the question that was posed. If you happen to be talking off-line to someone else without disclosing that fact, you can take the heat for what came from your mind.
You said it. Own it.
-Chris
I quoted you exactly. You said it, my comments pertain to your comments with regard to the question that was posed. If you happen to be talking off-line to someone else without disclosing that fact, you can take the heat for what came from your mind.
You said it. Own it.
-Chris
Thanks for the info and suggested reading. The above is what I have done, amps about 1 meter away in an ungrounded (no PE) chassis. A number of useful suggestions and links came my way from this thread Cable shield as a Faraday cage
This is not what one would get from reading the article today.
(the subject isn’t time related)
George
The issues aren't related only to op-amps though. Basically any active device rectifies RF from a cell phone. I have seen this in compliance testing for various products.
This is not what one would get from reading the article today.
(the subject isn’t time related)
George
Give it a few modernizations, such as "test-in specific application" instead of breadboard (as needed for more modern critical circuits), and he/she. Also my lunchtime read of the article's takeaway is that "nothing is quite so clean as we'd like" and "mind the particulars as the details matter".
Which is always a great reminder. This morning we had a pow-wow to figure out exactly what experiment we actually tried versus the one that we wrote up after getting some head scratching results.
Also, great stuff, KSTR.
- Status
- Not open for further replies.