Formula for Input Coupling Capacitor

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
EmRR said:
Jives with my mental benchmark, 100K/0.1.  I do the math in my head relative to that reference.  Usually in tube world here.
actually 15.9Hz -3dB

Back in the old slide rule days we had to estimate ball park RC filter pole calculations in our head like those, and then use the slip stick for resolving the significant digits.

JR
 
Everything inside the brackets gets done first , that bit I did remember from my maths class .
Equations I didnt mind so much , trig didnt sink in well with me at all .

I was just wondering about the role of output impedance , is it that with modern op amps its so low it has a negligable effect ?
In the calculator for  the tube coupling cap I posted  the output Z is factored into  the equation.
Might have made sense to include the actual formula each calc uses for educational  purposes .
 
ruffrecords said:
This would be extremely unusual in my experience. Can you post an example schematic?

Unfortunately not.. the line/RIAA/mic pre opamp kits/circuits I built had 1..10uF electrolytic input coupling whereas the FET/BJT circuitry I built for a mic input used a lower value, like 100..220nF cap. I guess it's similar to suspension in a car, a tradeoff/compromise between stability vs. frequency response and permitting possible low-frequency signals entering the circuit.

Afaik the higher value cap you use the more "suspension" your circuitry has against voltage spikes, rumble etc but it also high-passes the signal (making it sound thin) since the low-frequency content is blocked by the capacitor. By using a smaller cap it's like a race car with very stiff suspension but it's also prone to the aforementioned nuisances (in audio you'd generally high-pass at 30Hz, as anything below is wasting power since low frequencies require larger currents)

Another way to think it is water analogy; you'd have a water reserve prior to the supply pipes but sizing it too small drains it empty very quickly so you need to pump water into it faster to meet the demand, and making it too large the water goes bad as it sits there unused or requires a larger pump and an equivelant outflow (most water supply systems are automated though via use of float switches, a familiar phenomenon is loose piping rattling around caused by sudden changes in the system pressure)
 
efinque said:
Unfortunately not.. the line/RIAA/mic pre opamp kits/circuits I built had 1..10uF electrolytic input coupling whereas the FET/BJT circuitry I built for a mic input used a lower value, like 100..220nF cap. I guess it's similar to suspension in a car, a tradeoff/compromise between frequency response and permitting possible low-frequency signals entering the circuit.
It is all about impedance. The BJT circuits you built have a relatively low input impedance and so need a relatively high value capaitcor to ensure low frequencies are passed.  FETs are very high input impedance devices so they need a lower value capacitor to achieve the same low frequency response as in the BJT case..

Cheers

Ian
 
Tubetec said:
Everything inside the brackets gets done first , that bit I did remember from my maths class .
Equations I didnt mind so much , trig didnt sink in well with me at all .

I was just wondering about the role of output impedance , is it that with modern op amps its so low it has a negligable effect ?
In the calculator for  the tube coupling cap I posted  the output Z is factored into  the equation.
Might have made sense to include the actual formula each calc uses for educational  purposes .
are you sure you really want to know?  ;D

OK for the specific case of an input DC blocking capacitor the source impedance the signal is coming from is added to the R to ground (or virtual ground).  For typical audio interfaces we ASSume bridging terminations so load impedance is 10x source impedance or more, so generally inconsequential shift to HPF pole.

In extreme cases of high source impedance (like some consumer gear) the higher source impedance can shift pole tuning "and" will create a flat signal loss like a simple resistive pad.

For the example given of 20 Hz pole (10K and 0.8uF). A source impedance of 1k (not that rare in consumer gear) would result in an 18 Hz -3dB skirt and a flat wide band 0.8 dB attenuation. 

Is that TMI yet?  ::)

JR 
 
> Jives with my mental benchmark, 100K/0.1.  I do the math in my head relative to that

Mine is 0.01u for 1Meg; augmented by 100u for 1k.

This is for -3dB at <20Hz, as I wrote. Somebody here wrote "15.9Hz" which is true but fruitless. Assume caps and resistors can be 30% off their markings. Bad day at the factory OR 20 years of heat.

And you want only about ONE "20Hz" cut in a complete system. Six 20Hz cuts cascaded is -3dB at 80Hz and even a guitarist might notice (and like it). It is not uncommon to find ten bass-cuts in one box. As has been said here, most of your couplings should be aimed 2Hz or below. If you have "20Hz" memorized (or written on the restroom wall), transposing to 2Hz "should" be easy. If not, maybe take up origami?

And I have always measured frequency response with a meter. And a proper signal generator (Sine). While hiss-testing has use in acoustics, it's just wrong for electronics. And always needs more equipment than a plain HP 200AB and 400E ACVM.

But the title of this thread is "Input Coupling Capacitor". It may be utterly reasonable to have the FIRST blocker inside the box up near 20Hz, so subsonics do not muddy-up the input stage. And that's the one where we *might* use a calculator. All those others, we would be figuring "2Hz here means > 1.2345uFd", look in the catalog, 10uFd is same-price as 1.5uFd, we use 10uFd and probably a lot of places. (Some consoles use 220uFd in "high" impedance, ~~10k, links; with maybe one input and one output coupling cut a little closer.)
 
The reason I asked is that I have the the input impedance of my Input Amplifier set at 20K. That makes the input impedance either 20k or 10K depending if the two Input Amps are set to the same source.

I'm getting a level shift of about 0.2dB from some sources with the two impedances. I want to raise the input impedance to the point that any level shift is less than 0.1dB. I'm not worried about Johnson noise.

I'm going to try 50K first. I ordered some 4K7 and 10uF C0G (X7R) ceramics. It looks like they will fit from the data sheet. Probably an improvement over an Electrolytic.
 
I'll scarf the info down one chunk at a time and hopefully it sticks  :-[
maybe  lack of usage of that part of my brain has led to a dulling down over time ,
use it or loose it ,
Many thanks once again gentlemen.
 
Gold said:
I'm going to try 50K first. I ordered some 4K7 and 10uF C0G (X7R) ceramics. It looks like they will fit from the data sheet. Probably an improvement over an Electrolytic.

I feel obliged to warn that X7R dielectric will likely give measureable distortion due to voltage coefficient ( and thermal - but you can't avoid the voltage). To avoid this you need C0G/NP0 ceramic caps ( not all equal and likely not practicable at values required)  or look at film caps (BIG). Oversized electrolytics may be best option as often used in pro-audio kit,
OTOH you may not notice or even like the X7R contribution...
 
Newmarket said:
I feel obliged to warn that X7R dielectric will likely give measureable distortion due to voltage coefficient ( and thermal - but you can't avoid the voltage). To avoid this you need G0G/NP0 ceramic caps ( not all equal and likely not practicable at values required)  or look at film caps (BIG).

I thought C0G,MLCC and X7R are all different names for the same thing? Now that I read the data sheet more carefully it seems both C0G and X7R are types of MLCC but are different dielectrics. The PP caps I have won't fit. I thought I was getting C0G's but I guess not. It looks like 1uF is the largest the C0G's come in. Oh well an order down the drain. I have good quality electrolytics which is what I have been using.

 
Gold said:
I thought C0G,MLCC and X7R are all different names for the same thing? Now that I read the data sheet more carefully it seems both C0G and X7R are types of MLCC but are different dielectrics. The PP caps I have won't fit. I thought I was getting C0G's but I guess not. It looks like 1uF is the largest the C0G's come in. Oh well an order down the drain. I have good quality electrolytics which is what I have been using.

COG and NPO are the same thing (good).... X7R is not very linear, but ok (smaller) for PS decoupling.

JR
 
I like to use this online calculator:

http://sim.okawa-denshi.jp/en/CRhikeisan.htm

Go to higher levels in the page navigation at the top of the page and you get calculators for different RLC permutations, simple active filters, and more.
 
There are lots of often used equations.  What I've found useful is to put them in a spreadsheet , add to it as you go.  Then it's just a click away for any relevant info.
 
efinque said:
No offense but these types of questions are very theoretical with a lot of modifiers and the best (if not only) option is to try.
What is your basis for questioning theory?

Very often series input caps in audio circuits are small ceramic ones, in the range of 10-100pF, maybe 10nF max.
This is simply WRONG!

This is to keep the frequency cutoff within a reasonable range as you don't want the signal attenuated above 30-40Hz but it depends on the input signal strength, ie. whether it's a low-level mic input or a line level signal.
Formulae do not depend on signal amplitude. Electrons do not care if they represent music, digital data, or RF garbage.

As for the inverting configuration an opamp usually has an internal resistor across the Vin+ and - input terminals so the non-inverting (+) input is directly fed into the amplifier circuitry and the inverting (-) one has a series resistor.
Where have you seen that?

I think it introduces a phase shift as well,
Please don't use concepts you obviously don't understand.

their use as comparators is based on this feature and many audio applications simply ground the inverting input.
NOT TRUE!

Like in filter theory there quite likely is a formulae for calculating the value but it's application-specific I think as it's very different to use it for audio than for example driving a LED
There's no filtering involved in LED driving, and if there was, the formula would be the same

IN BRIEF: Please avoid interfering in a conversation to which you only bring confusion and misinformation.

 
> impedance either 20k or 10K ... getting a level shift of about 0.2dB from some sources with the two impedances.

If source impedance were "ZERO" there would be no shift.

I bet you have a 220 Ohm source. You could just buffer this to nil. A 19 cent opamp will do <10r over the audio band.

I suspect you do not want this point to define your lower bass limit. So you want this <<20Hz, even tenth-Hz. For 20k load:

20Hz @ 10k is 1uFd
20Hz at 20k is 0.5uFd
0.2Hz at 20k is 50uFd
If this must be an Electrolytic, go 10X bigger to reduce distortion.
So either a BIG film cap or a 470uFd e-cap.
 
PRR said:
I bet you have a 220 Ohm source. You could just buffer this to nil. A 19 cent opamp will do <10r over the audio band.

I was getting a level shift with both the Studer A80’s and the DA converters. The input is a four resistor around an op amp quasi balanced arrangement. I’m using 0.1% resistors.  I like better than a trimmer.


I thought I remembered that oversizing this capacitor could cause problems. I guess I misremembered.  I have plenty of electrolytic caps.
 

Latest posts

Back
Top