Common Mode Chokes

[ricardo]

What carefully-chosen build-out resistors give, inductors taketh away...

How do we tame this ringing? I don't see how a Zobel could fix this.

I try for a Butterworth response, preferably 3rd order and am usually happy with 2nd. These will ring on square waves but are "maximally flat " in the frequency domain.

You never get square waves with fast rise times from microphones unless they are well into overload. On mixdown, digital signals will have severe bandlimiting (and ringing from the anti-aliasing filters on artificial square waves). If they didn't, they would alias and sound bad.

IM very HO (donning flame jacket) the benefit of bombproof EMI immunitiy far offsets any supposed degradation in "transient" or supersonic response.

I've conducted careful Double Blind Listening Tests bla bla . on the effects of severe response limiting.

I'm now off the test the effects of alcohol and X'mas pudding on my performance in Blind Listening Tests ... Purely in the interests of Science of course. Merry X'mas!

[mediatechnology]

Well, without regard to what the build-out resistors do in the time domain, they still reduce current demands and DIM-like distortion. So I guess the inductors do not completely taketh away.
I need to measure the improvement I'm seeing in reduced RF at the line receiver input pins. These chokes do work quite well. And, given a choice, I'll take the common mode choke over the uncoupled inductor pair.

I'm now off the test the effects of alcohol and X'mas pudding on my performance in Blind Listening Tests ... Purely in the interests of Science of course. Merry X'mas!

I'm sure everything is going to sound better. Merry Christmas ricardo.

[emrr]

I know I'm in a minority here but I like band limited signals. Knowing what microphones give out and what speakers and amps do in the ultrasonic range, I'll stick my neck out and claim proper and severe bandlimiting above 20kHz makes things sound BETTER.

Hey, I'm the perv who keeps using transformer coupled tube gear from the 1940's, where it tends to usually be down 3 dB at 20kHz. It sure sounds cleaner, and the brain really knows how to imagine the harmonics in ways better than reality.

But plenty would argue with me. Who cares?

John sent me the Jensen Comtran information he had collected over the years and has also pointed out elsewhere the role of the input transformer as a bandpass/band limiting filter.
The Jensen archive is worthy of it's own thread.

Jensen Comtran? Sounds like something I've not seen. I did recently score an early 1980's Jensen binder with a lot of early parts they no longer make, which needs to be scanned for sure. And yes, many dissections to be had in a Jensen thread. Their approach seems to be the most defined point in the modernization of transformer terminology. It is very difficult to interpret their design standards in the same frame with all that came before.

[mediatechnology]

Hey, I'm the perv who keeps using transformer coupled tube gear from the 1940's, where it tends to usually be down 3 dB at 20kHz. It sure sounds cleaner, and the brain really knows how to imagine the harmonics in ways better than reality.

But plenty would argue with me. Who cares?

Ah, the good 'ole days: viewtopic.php?f=12&t=493

Jensen Comtran? Sounds like something I've not seen. I did recently score an early 1980's Jensen binder with a lot of early parts they no longer make, which needs to be scanned for sure.

John sent me his binder which included the Comtran software literature and cassettes.
Before scanning it I wanted to see how much of the information was already on Jensen's website and check with Bill to see if he was supportive of it being published.

Long chains of bandwidth-limited stages are problematic.

Burdick's Clean Audio Installation Guide talks about system bandwidth: http://www.benchmarkmedia.com/sites/def ... df#page=10

Burdick also has recomendations on common mode chokes and the differences between common mode and differential bandwidth: http://www.benchmarkmedia.com/sites/def ... df#page=16

And his comments on squarewave testing the entire chain: http://www.benchmarkmedia.com/sites/def ... df#page=22

We are also talking about two different things in our discussion of transformers and common mode chokes.

One is limiting RF to prevent rectification using common mode chokes.
The other is the bandwidth limiting transformers provide in addition to RFI protection.

One common theme I see however in the line input/output threads as well as here are the benefits of the fully-balanced or differential interface be it active or passive.

[JR.]

Long chains of bandwidth-limited stages are problematic.

The junior engineer learns early that cascaded frequency response is cumulative. I had this driven home designing 7 pole filters for my early BBD delay efforts. But yes every roll off matters.

RF matters where you touch the outside world and no longer control your environment.

One common theme I see however in the line input/output threads as well as here are the benefits of the fully-balanced or differential interface be it active or passive.

I will add that differential signal consideration is important for almost every single audio stage inside a product. Without differential receivers I could never send signals around inside a 6' long console with any integrity. And there is no such thing as a one-legged signal.. it is always relative to something. I need to stop before I repeat my full "ground is not a voltage" rant...

merry happy all...

JR

[mediatechnology]

Thanks John, Doug and ricardo.

What carefully-chosen build-out resistors give, inductors taketh away...

Fortunately resistors can give most of it back...

A Zobel network was suggested earlier by ricardo.
At the time I was thinking of a differentially-connected network of the type sometimes seen on the secondary of mic transformers.
I don't know if a resistor alone can be called a Zobel network.
But it can be called a snubber....

This is the time domain response with 1 k Ohm resistors in parallel with the 94 uH/leg inductors.


Common Mode Choke, 10 kHz, ~7.5V p-p, 94 uH/leg in parallel with 1 k Ohm, 270 pF shunt.
Top trace is diff amp input, bottom trace cross-coupled THAT1286 recovered output.

Amazing what 2 cents can do.

For whatever reason you don't often see the inductors snubbed. Why is that?

Burdick didn't do it in his prototype and we don't see it in the CRL DynaFex examples given earlier.
Sam Groener does it here and elsewhere: http://www.sg-acoustics.ch/analogue_aud ... oso_r1.pdf
We also saw ricardo do it here: viewtopic.php?f=6&t=464&start=34

Quick tests of HF CMR in an an asymmetrical open protoboard show 40 dB CM rejection from 1-3 MHz.
Tone bursts of square wave modulated RF show no evidence of rectification with inputs as high as 10 V p-p.

This dog may hunt after all.

The role of CM choke and limiting HF differential gain in the mic preamp deserves further exploration.

[JR.]

Sweet....

I am having flash backs to my first technician job in the 60's debugging a pre-historic DC-DC switching power supply.. we had to snub the transformers to prevent the voltage spikes from ringing and killing the crude slow not very robust, circa-60's power transistors with voltage spikes. IIRC the snubbing was the typical RxC load directly across the secondary. I used to have to grind the tops off of the TO-3 transistors to inspect for evidence of voltage punch-though, or melted base and/or emitter wires... interesting stuff,,, you could literally see the failure modes on the devices.

I recall one pretty spectacular bench explosion when a snubber cap released it's internals all over my bench and the immediate vicinity. It looked like the aftermath from an evil pillow fight. Luckily I was on the other side of the room when she blew, but there was no warning. I was just a know-nothing electro-mechanical tech, and the junior engineer I was helping (a MIT graduate student) barely knew much more than me... In fact the engineer was working as a co-op who left after six months to return to classes, and I finished the power supply without an engineer from there... The original design was done by some real engineer at one of the RT 128 companies outside Boston doing government work (maybe Raytheon?). The power supply was for a special Navy vessel, and we just had to make the real version, using real production parts, work. More work than it should have been, but understandable in light of the early days and crude device technology.

JR

[emrr]

This is the time domain response with 1 k Ohm resistors in parallel with the 94 uH/leg inductors.

Very cool, makes sense.

[mediatechnology]

I'm going to be ordering some common mode chokes in the 1-3 mH range from Mouser today and will look at those for mic input applications.

I did run some pulsed RF (LF square wave AM modulation of sine at 100%) tests using a combination of the 94 uH/leg CM choke, 1K dampers, 270 pF/leg with a cross-coupled 1286 line receiver and saw no evidence of rectification up to the generator limit of 5 MHz. I used 10V p-p so it should be a pretty tough test.

I don't know where the attenuation curve starts to rise again with this choke but it would seem that a combination of ferrite beads and a CM choke might provide the best RFI protection.
I wouldn't use these VersaPacs in production as a simple CM choke since they're fairly pricey.