Phantom Power Capacitors

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Status
Not open for further replies.
Indeed it's interesting, but I have never seen a distinct correlation between DF and distortion.
DF is a very important factor in reliability, though.
DF is loss and loss is not fidelity because something's missing. Whether the missing info is perceptible is another issue. Sort of proving a negative. . .

There was a study on caps like John says in the 80s by Walt Jung and Rich Marsh. Of course it is 45 years old. I'm just posting a link to it but not getting into the fray over it.

https://reliablecapacitors.com/information/picking-capacitors.html
 
Ein. Equivalent Input Noise.

If you want to measure resistor noise, as opposed to the circuits self noise, be my guest.

Actually noise factor would be a better measure but that also requires a realistic assumption about source impedance rather than the marketing fantasy of zero ohms.

Semiconductor mic pres with 3dB or better noise figures have been around for many years. Any improvement over that, real or imagined, is pointless in the real world.

Cheers

Ian
 
I remember when electric cars had nobody convinced "that there was any merchantable (sellable) benefit".
The electric car was invented over 100 years ago. I'm old but I can't remember back that far.

If they were such a good idea it wouldn't take both the government's thumbs on the scale and over a century for them to catch on.
By the same logic a McD burger is superior to real food, which has few if any "merchantable (sellable) benefit.".
Can't say that I follow your logic but I vaguely recall grabbing lunches from Micky D's back in the 60s.
Fun thing is, getting rid of those caps actually makes commercial sense.

Thor
Is that common core math?

JR
 
DF is loss and loss is not fidelity because something's missing.
DF is a catch-all that includes various losses, some linear and other non-linear.
Among them, the most significant is ESR, which is a combination of both linear and non-linear phenomena.
The purely reactive, non-variant part of ESR does not generate distortion.
So the correlation between DF and distortion is vague.
There was a study on caps like John says in the 80s by Walt Jung and Rich Marsh. Of course it is 45 years old. I'm just posting a link to it but not getting into the fray over it.

https://reliablecapacitors.com/information/picking-capacitors.html
I have a problem with this article. See.
DF1 resized.jpg
The problem is they don't consider the capacitor in the context of it's surrounding impedances, they assume very wrongly that the signal passing through the high-pass circuit resulting from the connection will exhibit the same frequency and phase response than the capacitor alone.
Measuring the impedance of a capacitor actually operates it with a zero-ohm load.

DF2 resized.jpg


In addition to the typical audiophool language, there's a gross over estimation of the consequences of the impedance variations on the overall response of the circuit. Once you consider these variations in respect to the rest of the circuit, it becomes clear that the effects they predict are largely exaggerated.
 
On I see, a marketing spec not a real world one.

Cheers

Ian
Yes, of course. 10 Ohm would produce an EIN of -142,7 dBu :):):)
Most MicAmps are more likely in the 100 Ohm region, -132,7 dBu unweighted, that is completely sufficient...
 
When someone proposes a "capacitor whose value changes with frequency" I think of "a linear time-invariant system" and the link to distortion is, I'm afraid, quite hard to imagine. Likewise any models which consist purely of R's, C's and L's.
 
I'm always amazed at how your posts are provocative

Of course some are. There is too much BS around that needs challenging, the that about capacitors with enough DC voltage across them distorting less.

and counter educative.

Really? Did you look up Cohen and Birt? High side current sensing?

And then take into account my parts list?

Commonly I'm quite forthcoming with schematics. In this special castle, forgive me if I'm not.

What I noticed however that most people went out of their way to assert that there is no need for such a thing anyway.

Not one person said: "If you make a kit that sells at 100 USD each, I take eight!"

Thor
 
I'm always amazed at how your posts are provocative and counter educative.
This thread has turned to s..t.
Word. It's one thing not to share a circuit, but quite another to write a dozen long posts begging people to ask about it.
 
Last edited:
Actually noise factor would be a better measure

Actually, Ein with input short is best, IMNSHO. Here is why.

The noise of a system is always that of all noise sources.

By lumping multiple separate noise sources into one, it becomes hard to determine what is what.

but that also requires a realistic assumption about source impedance rather than the marketing fantasy of zero ohms.

We have two separate tests.

Self noise and system noise.

A system noise where electronics do not increase noise would have a noise factor of 0dB and an Ein (0R source) of -oo dBu.

Semiconductor mic pres with 3dB or better noise figures have been around for many years. Any improvement over that, real or imagined, is pointless in the real world.

3dB noise factor means the noise of the electronics is the same as the source.

If an improvement beyond that is meaningful or if we may even release our spec to -120dBu Ein (0R source) will depend on the noise of the actual microphone.

Ideally we want to have a noise that significantly ( > 6dB) lower than microphone noise. That is what really matters in the real world.

Let's say I have STC 4104A ribbon Mic, my Z is 30R. Suddenly a Noise factor of 3dB at 150 Ohm seems not so great.

Hence I do not consider noise figure or Ein stated at a specific impedance all that useful.

Two preamp's with similar Ein at 150R can have hugely different noise with our low output, low impedance Ribbon Mic.

Thor
 
The noise of a system is always that of all noise sources.
By lumping multiple separate noise sources into one, it becomes hard to determine what is what.
+1. A shorted input is an unambiguous universal standard that isolates the issue of noise to the DUT alone, which is what we as designers generally want to know.* EIN voltage (and current if necessary) then tell the whole story; it's not marketing at all. 'Noise Figure' is a concept better suited to RF design where source impedance is generally well defined and standardised, and flicker noise rarely contributes.

*For the same reason I am irritated that Jensen persuaded the IEC to change their CMRR test standard to include a deliberate mismatch in the output impedance of the source device, so it doesn't test the CMRR of the DUT alone!! /rant
 
Last edited:
Actually, Ein with input short is best, IMNSHO. Here is why.

The noise of a system is always that of all noise sources.

By lumping multiple separate noise sources into one, it becomes hard to determine what is what.
except presenting a dead short at the input completely ignores the input noise "current" a significant noise contribution with bipolar transistor input devices.

JR
We have two separate tests.

Self noise and system noise.

A system noise where electronics do not increase noise would have a noise factor of 0dB and an Ein (0R source) of -oo dBu.



3dB noise factor means the noise of the electronics is the same as the source.

If an improvement beyond that is meaningful or if we may even release our spec to -120dBu Ein (0R source) will depend on the noise of the actual microphone.

Ideally we want to have a noise that significantly ( > 6dB) lower than microphone noise. That is what really matters in the real world.

Let's say I have STC 4104A ribbon Mic, my Z is 30R. Suddenly a Noise factor of 3dB at 150 Ohm seems not so great.

Hence I do not consider noise figure or Ein stated at a specific impedance all that useful.

Two preamp's with similar Ein at 150R can have hugely different noise with our low output, low impedance Ribbon Mic.

Thor
 
except presenting a dead short at the input completely ignores the input noise "current" a significant noise contribution with bipolar transistor input devices.
That's why you publish the noise current too, if it is significant.
 
Using NF includes all these sundry sources into one consolidated measurement.
Exactly, which is confusing since you can't extrapolate it to any other source impedance, you're stuck with whatever the publisher chose. Fine for RF where everything is always 50 ohms or whatever.
 
Exactly, which is confusing since you can't extrapolate it to any other source impedance, you're stuck with whatever the publisher chose. Fine for RF where everything is always 50 ohms or whatever.
Low z audio mics are nominally 150-200 Ohm source impedance so NF wrt 150-200 ohm should be representative and easier to understand*** than ein noise voltage in dBu without bandwidth or other supporting details. Sadly the consumers will take dBu numbers and try to compare different companies SKUs without the rest of the story needed to make those comparison apples to apples.

JR

PS: I didn't have much success trying to promote NF for specifying audio preamps to customers last century.
 
The correct type and vlue of phantom power blocking capacitor is whatever sounds best and has no longterm problems. You can use your own ears to determine what sounds best but there's not enough time to test for long term problems but in theory and practice film caps will have fewer if any long term problems. Electrolytic caps from major manufacturers (but not made in PRC but ok if from ROC) are all going to be similar as all the manufacturers know what each other knows and is doing. Japanese and German parts are probably the best due to cultural influences fostering integrity. BUT.... a mic pre without input and output transformers is like a vegan meal....it won't satisfy and can lead to sarcopenia.
 
Status
Not open for further replies.
Back
Top