Impedance~Ignorance

Preface: I'm an idiot.

How on earth do you calculate in/out impedance? If there are caps in the signal path at the input of a micamp (DC blocking) and i'm sure i've read that the impedance of a cap is dependant on frequency, is a stated input impedance referenced at a certain frequency?

....and, per the datasheet p.3 of the ina217, the input impedance is 60M||2pF. (in series with?) Why o why didn't i take the blue pill?
Based on the circuit recomendation in the DS, p.7: the two 47uF dc blockers, the two 6k8 phantom Rs, and the two 2k2 input Rs, oh and throw in the four in4148 Ds, how does this effect the input impedance? If someone could give me a straight answer as to how to calculate the impedances (in and out) i would forever be in your debt.

oh, i see no stated output impedance in the datasheet for this IC. Is this an "assumed" value? I'm lost. How is 60M||2pF useful?
Thanks, Ben.

Post script: I'm sure as with most things i've found a way to over complicate this.

> I'm an idiot.

Not proven.

You do seem to be ignorant.

Nobody is born knowing this stuff.

Ignorance is sometimes curable.

> a way to over complicate this.

Yes.

> Based on the circuit recomendation in the DS

To dodge the bigger question: assume that this chip's makers are fairly smart, and that they intend you to use it like any other mike preamp. What is the usual rule of modern interfacing? "Load >10 times bigger than source." Source is 200 ohm mike. We ass-ume that the chip in suggested configuration will show over 2,000 ohms to the mike. Do we need to know more exactly? Usually not.

And for its output: it is rated to drive loads to 2K or lower. We again ass-ume the output is under 200 ohms. There's good reason to think it may be closer to 1 ohm, but do we care what the exact value is?

bachevelle52 said:

How on earth do you calculate in/out impedance? If there are caps in the signal path at the input of a micamp (DC blocking) and i'm sure i've read that the impedance of a cap is dependant on frequency, is a stated input impedance referenced at a certain frequency?

Click to expand...

It should; most of the times it is assumed to be 1kHz. Over the 20Hz-20kHz range, the impedance may vary significantly, although it should not have any significant consequence.

....and, per the datasheet p.3 of the ina217, the input impedance is 60M||2pF. (in series with?)

Click to expand...

|| means in parallels.

Based on the circuit recomendation in the DS, p.7: the two 47uF dc blockers, the two 6k8 phantom Rs, and the two 2k2 input Rs, oh and throw in the four in4148 Ds, how does this effect the input impedance? If someone could give me a straight answer as to how to calculate the impedances (in and out) i would forever be in your debt.

Click to expand...

The straight answer is you have to take an EE course. It's not difficult, (even I managed to grab it), but it takes time... You have to learn Ohm, Kirchoff, Thevenin, ...

abbey road d enfer said:

You have to learn Ohm, Kirchoff, Thevenin, ...

Click to expand...

... And an insatiable appetite for alcohol.

> 60M||2pF. (in series with?)

In parallel with. (mnemonic: parallel lines)

> over complicate this

This plan is over-complicated. Everything is duplicated. For a reason, but it confuses analysis. When faced with a symmetric mess, cut it down the middle, deal with the half-section, then work out what both halves together come to.

Here's the half-section of this input:


> i'm sure i've read that the impedance of a cap is dependant on frequency

You need to KNOW this, not just read it. You gotta learn to look at parts the way a stone-layer looks at rocks. You got your flat rocks and you got your slanty rocks. One is good for getting the same height all the way along a wall, the other is good for stuff which is different heights at different places. We use caps when we want different heights at different places: different response at different frequencies.

All caps are wide-open at infinitely low frequency (DC) and dead-short at infinitely high frequency.

So what happens if all caps are assumed "open"? We see only the 6K8 resistors.

What happens if all caps are "shorted"? The path through 47uFd and 2pFd is a dead-short, so everything is a short, the resistors are unimportant.

BUT what happens in between?

The caps are an in-between short and open value depending on the frequency and their value. You MUST be able to know/estimate cap impedance -quickly-. I like a Reactance Chart. Some people use calculators, though this is slow and mistakes may be missed. Out in the woods, it is useful to know a few "common values" and the way to transcribe to other values.

Ignore the 2pFd for a moment, it is quite small, which means it won't matter until some very high frequency.

I "know" that 100uFd against 100 ohms is about 20Hz (16Hz). Therefore 50uFd is 200 ohms at ~20Hz. What is 200 ohms in series with that 2.2K? Pretty-near 2.2K. The cap has very-small effect until it rises to near 2000 ohms, If 50uFd is 200 ohms at 20Hz, it is 2,000 ohms at 2Hz. At some point near 1Hz it goes through about 5K or so, which starts to swamp the 6.8K.

So we have 6.8K for frequency much-much below 1Hz. Above 2Hz the 47uFd is a very low value compared to 2.2K. So total impedance is 6.8K||2.2K which is (have you mastered parallel resistances?) 1.7K.

Ah, that pesky 2pFd. Where does it come down near 1.7K and start to short-out everything else? Well, I recall that 0.01uFd against 1K is around 17KHz, and 0.01uFd is 10,000pFd. So a 2pFd cap has similar impedance at 10000/2= 5,000 times higher frequency. That's up around 85MHz, which is so far above the audio band that I won't bother re-figuring for 1.7K to get 50Mhz.

So half-section input impedance is 6.8K below 1Hz, 1.7K above 2Hz, then falling above 50MHz.

Corecting for the full duplication, input impedance is 13.6K below 1Hz, 3.4K above 2Hz, then falling above 50MHz.

> is a stated input impedance referenced at a certain frequency?

If it matters. Here we have nearly exactly 3.4K from below 20Hz to over 20KHz. The lower bend is 2Hz so we might approximate the 20hz impedance to be less than 10% high. The upper bend appears to be out of sight from an audio viewpoint. Assuming "3.4K all across the audio band" will not make any trouble.

BTW: that "2pFd" is the chip alone. There is capacitance everywhere. 30pFd per foot of cable. Several pFd at any length of wiring. So if you go out to the mike on the other end of 100 feet of cable, there's 3,000pFd. This will begin to droop below 1.7K ohms by 33KHz. Someone designed it this way: the common mike impedances are not much troubled by reasonably long cables.

> Why o why didn't i take the blue pill?

There's a joke there but it could be rude.

Oh. I fergot the 60Megs.

I "knew at a glance" that 60Meg||2.2K "is" just 2.2K for any practical purpose. Doing (yuk) math, I find that the total is 2.1999 ohms, and I never worry about 0.003% errors in a world of 20% mike impedances and 5% resistors.

Hi PRR,


  after all that fabulous explanation, I really don't think anyone would begrudge you a bit of rude! - Have one on me . . .


      "You got your flat rocks and you got your slanty rocks."!!!!!! fantastic! . . . I am mopping up the beer I that sprayed through me nose . . . .



      Kindest regards,



          ANdyP

...didn't mean viagra...

Holy crap my brain actually works! Thanks to all who responded, and Paul, the freaking oracle. ;D

...so, anyone care to explain the interaction between R and C in an RC filter. I "get" that one orientation is a LP, and the other is a HP, but i don't "understand" exactly why. What is it about the placement of the R in conjunction with C makes a HP or LP respectively.

bachevelle52 said:

...so, anyone care to explain the interaction between R and C in an RC filter. I "get" that one orientation is a LP, and the other is a HP, but i don't "understand" exactly why. What is it about the placement of the R in conjunction with C makes a HP or LP respectively.

Click to expand...

Do you understand how a voltage divider works? If so, consider the RC filter as being the same, except replace one of those resistors with a cap. Now watch how the (AC) voltage drop across either component changes as the reactance of the capacitor changes - it acts a bit like a resistor who'd resistance is frequency-dependent.

Now, as to whether it's going to be a LP or HP, that depends on where your take your output from - either across the resistor, or across the capacitor.

There's probably a nice page on the net with drawings showing how this works. Try Wikipedia. It's relatively simple.

Do you understand how a voltage divider works? If so, consider the RC filter as being the same, except replace one of those resistors with a cap. Now watch how the (AC) voltage drop across either component changes as the reactance of the capacitor changes - it acts a bit like a resistor who'd resistance is frequency-dependent.

Click to expand...

Yow, can we get more explanations like that but for compressors and preamps?  that is nice and easy to understand!    I've been hounding the guys in #electronics about circuit analysis and 'do i need to know how to create transfer functions for a circuit' and whatnot, and this is nice and simple and to point!    more more more!!!!

> more explanations like that but for compressors and preamps?

EVERYTHING is a Voltage Divider (or may be understood as a voltage divider).

LA2a _IS_ a voltage divider: 68K fixed and variable LDR. Everything else is a frill. Because LDR-lamp power generally exceeds LDR happy-power (or available signal power), you often want a sidechain amp. Because the whole point of the LA-2a is to "raise loudness" within an existing system, you often want a booster-amp on the main output. But just frills.

Preamps are just small amps. Amps are usually rows of tubes/transistors fighting (voltage-dividing) resistors. The tube/transistor is a variable resistor. You put in boring B+ DC, voltage-divide that according to small interesting signal, and get large interesting signal.

The US Marines "builds Men". They start with Basic Training: a zillion push-ups. If I had a basic-training to build audio designers, it would start with a zillion voltage-divider (and similar) exercises. However audio design is not as romantic as leather-necking, and nobody will pay me to be their divider-drill sargent.

> more more more!!!!

You can't learn to play guitar without putting finger to fret many-many times (which is why I don't play). You MUST be able to play a Gmin cord without thinking. And you must be able to recognize and rough-answer a voltage-divider situation with negligible brain-strain.

Which is kind of why I urge people to learn Ohm's law. Not just know it, feel it!

Ok, I'm drunk, that might be going a bit far, but you catch my drift.

Roddy,


  wahay! Naughty . . .


      ANdyP

In my mind, resistance is a fixed DC resistance, while impedance is a measurement of AC resistance, that is FREQUENCY-DEPENDENT.

1. So how can we say what the impedance of anything is, without mentioning what frequency we're testing under?

2. Assuming what I just said is right, then how does one look at an unknown transformer, for example, and determine it's input impedance?
Putting a meter over the primary would tell us the DC resistance of the primary. But how do we measure the impedance? I'm going to guess that we pick some standard measurement frequency, like 1kHz, pump that in as a sine wave at some standard level, and then measure (somehow)???

3. Case in point. I have 32 Sescom 600K:600K isolation transformers. DC resistances of all primaries and secondaries measure about 320R unloaded. So far, so good. But a whole bunch pass audio at real low levels, while others are fine. Obviously, the DC resistance of the coils is not enough to tell a "good" tranny. I'd bet an impedance measurement would uncover some issues...

You guys rock. Still  l e a r n i n g...

One more though on the topic. Look at car speakers. Nominal impedance 4 ohms. But look at the impedance curve and you'll see 4 ohms only around 200Hz or so, then it varies greatly over frequency.

riggler said:

One more though on the topic. Look at car speakers. Nominal impedance 4 ohms. But look at the impedance curve and you'll see 4 ohms only around 200Hz or so, then it varies greatly over frequency.

Click to expand...

There's a regulation there, by the AES, that acknowledges this fact and says that nominal impedance is 1.25x minimum impedance.

riggler said:

In my mind, resistance is a fixed DC resistance, while impedance is a measurement of AC resistance, that is FREQUENCY-DEPENDENT.

1. So how can we say what the impedance of anything is, without mentioning what frequency we're testing under?

Click to expand...

Very often, it is not very frequency-dependent, and (almost) everybody knows that with the bridging notion (Zin> 10xZout) it is not very important.

2. Assuming what I just said is right, then how does one look at an unknown transformer, for example, and determine it's input impedance?

Click to expand...

The notion of a specific impedabnce for a xformer is an almost completely different subject. The input impedance a xfmr shows to the world is dependant on what is connected to the other side of said xfmr. development would be too long, but basically when you see a xfmr rated at 600:600, the manufacturer guarantees that it is optimised for such operation.

Putting a meter over the primary would tell us the DC resistance of the primary. But how do we measure the impedance? I'm going to guess that we pick some standard measurement frequency, like 1kHz, pump that in as a sine wave at some standard level, and then measure (somehow)???

Click to expand...

Inject an AC current of known value and measure voltage across it, then use Z=v/i. See attached

3. Case in point. I have 32 Sescom 600K:600K isolation transformers.

Click to expand...

I assume it's 600R:600R

DC resistances of all primaries and secondaries measure about 320R unloaded.

Click to expand...

That's quite high...

So far, so good. But a whole bunch pass audio at real low levels, while others are fine. Obviously, the DC resistance of the coils is not enough to tell a "good" tranny. I'd bet an impedance measurement would uncover some issues...

Click to expand...

Probably...Hell, something's very wrong. Most of the times defective xfmr's measure open-circuit (because the tiny wires act like a fuse), but sometimes they just fry, the thin insulation burns and a few turns become shorted. This generally does not show when measuring DCR, but the Z curve is way different.
[/quote]

I like the 'tone' of this discussion!
and especially the slant PRR gave to it!

Let's talk cuisine and cooking....

You want to make pasta....
The recipe says: 3 litres of water, 1 tablespoon of salt, 2 tablespoon of vegetable oil.... bring to a boil, put 400grams of pasta and let boil for 7 minutes turning twice.

Yeah it will work probably...

Here is my Mom in law recipe, she is from central Italia (the Abruzzes)
Take a large pan, put a lot of water and bring to a boil... add salt and oil (how much nonna?) I don't know... put some!
then add the pasta...

For how long nonna?

I don't know... until it taste good!!

Electronic is quite the same for some aspects....

You gotta feel it!
you gotta taste it (test it)
everybody has his recipe!
If done with respect with the traditions and rules at hand, it will be good, if different!!

I'm hungry!
Luc ;D