Coupling Cap polarity..?

[PRR]

The direct (before cap) output of the preamp is "nearly zero DC". But it may be a few tenths Volt of DC, which should not be allowed to flow to the next input.

If all inputs were high impedance, and there were no buzz in the world, we could use a 0.5uFd film-cap to block this DC.

But 0.5uFd into 10K load is -3dB at 34Hz, -1dB for kick-drum, disappointing.

And we would like to have the line driven with low-low impedance at 50/60Hz, to absorb buzz that sneaks into cables.

And 0.5uFd film is not cheap. (Even at the lowest voltage they can make film.)

So we want 10, 20, even 100uFd caps, to block DC while still passing bass and absorbing buzz.

These "have" to be electrolytic (or VERY big).

We do know that the preamp tends to cancel DC voltage at its output. But the cancellation leaves the difference of small errors. So while the output DC is "small", it may be either polarity.

Ideally an electrolytic has a constant DC voltage and polarity to keep the oxide formed. However modern electrolytics stay formed with proper voltage, or no voltage, or up to 1V reverse voltage.

Therefore in _this_ case, the output caps may be installed EITHER way. Whatever looks better on the drawing or to the builder.

If the output DC was large and we knew what polarity (such as a single-ended amplifier; or the Phantom voltage at the input of this preamp), we could use a standard electrolytic carefully installed the right way around.

If we were not sure the output DC was "small", we would have to use Bipolar Electrolytic, which can block large DC either way. This is always "safe", but more cost and size.

 

[saint gillis]

Ricardo told me in some other thread that  an ne5534/5532 because of his npn inputs had a slightly negative dc current at its inputs (maybe I didn't understood correctly), does it makes sense?

 

[JohnRoberts]

Ricardo told me in some other thread that  an ne5534/5532 because of his npn inputs had a slightly negative dc current at its inputs (maybe I didn't understood correctly), does it makes sense?

Revived after Six years....Oh well..

If you review the 5534 data sheet they spec a typical input bias current of 500 nA and max 2000 nA.

The input LTP is NPN so this bias current will flow into both the + and - inputs.  In addition to the bias current there is an error (offset) current +/- 20 nA and max 400nA.  So worst case bias current is 2000nA +/- 400 nA.

You need to factor this input current times the DC resistance of input and feedback resistors, and add that to the op amps input offset voltage  +/- 0.5mV  but max +/-5mV.

Or you could measure the DC voltage  at the input but it will probably be negative.

JR

 

[abbey road d enfer]

Ricardo told me in some other thread that  an ne5534/5532 because of his npn inputs had a slightly negative dc current at its inputs (maybe I didn't understood correctly), does it makes sense?

The polarity of current is a relative subject.
Just like the direction of current of a river depends on what bank you're looking from.
NPN transistors require current flowing into their base, meaning it has to be delivered by positive voltage, and again positive is relative. In this case, the reference point is the emitter. So current is flowing from the bias source (quite often 0v, which is more positive than the V- rail) into the base. The base voltage is thus negative in reference to the bias source.
For offset optimization, it is recommended that the DC resistance presented to the inverting and non-inverting inputs are equal, but that is not always possible and in particular with variable gain mic pres, not actually feasible. In fact, it is not the best option when dealing with audio circuits (that don't pass DC and thus need removing).
Depending on chosen architecture, one of the inputs can be chosen as having a higher DC resistance, which will (may?) alter the balance in a way that the output offset will always have the same polarity.
For example, if the mic pre uses a xfmr, the secondary may be connected directly to the non-inverting input and the NFB network AC-coupled, with a DC path of about 100k; the resulting DC voltage will be almost zero on the non-inverting input and about minus 50-200 mV on the inverting, yielding about the same positive output offset. Then , connecting the + side of the cap to the output would be correct.
OTOH one may chose a different structure, with the secondary AC-coupled, with a 100k input resistor and the NFB being low resistance (typically would vary between 0 and 10k); then, the balance would shift in the other direction, and the output offset would be negative, necessitating reversing the direction of the coupling cap.
Indeed, the analysis of offset is one consideration that should influence the choice of topology of the circuit.

 

[saint gillis]

Thank you JR & ARd'E for these great explanations!