In the circuit design of a peak-detecting receiver for shaped tone burst testing by Siegfried Linkwitz, there is a 47 uF polarized tantalum capacitor between preceding amplifier stages and a full-wave rectifier. See Figure 12 of on p. 259 of his AES paper that I'm including a link to. As far as I can tell, the output terminal of the preceding op amp connected to the anode of this capacitor doesn't have any significant DC offset with respect to ground. Thus I don't see the purpose of AC coupling to the full-wave rectifier. Is there any good reason for including this capacitor in the circuit? The capacitor that I'm referring to is at the top right-hand side of Fig. 12 and connected to the inverting input terminal of the op amp functioning as a FW rectifier by a 10 k Ohm resistor.

I'm constructing a modified version of this receiver for myself and would not include the 47 uF capacitor in my receiver if it isn't really needed.

https://www.linkwitzlab.com/JAES/jaes_papers80.htm

 

the preceding op amp connected to the anode of this capacitor doesn't have any significant DC offset with respect to ground.

But it may be significant.
The intrinsic op amp differential input offset may be small (perhaps a few mV) but it is amplified by the DC gain of the circuits, so the capacitor is likely needed to block any amplified offset from the preceding stages so it won't be amplified by the following stages.
I would leave it in.

Incidentally, I consider Siegfried Linkwitz as one of the premier audio speaker designers ever, with his speaker designs likely giving the most transparent sound of any available, with the speakers (audibly) disappearing from the room.

 

There can be an inherent unwanted dc offset from each of the op-amps, which is then amplified by the following op-amps. This will not normally be an issue where the output is ac coupled to remove the offset; as in this circuit the output is to a rectifier circuit, therefore any dc offset will appear at the rectified output, hence this will be removed by the 47µF capacitor.

You might be lucky and your circuit have minimal dc offset at the output of 10/20dB op-amp stage – but I would include the possibility of including the capacitor in the circuit (or replacing it with a link). I would think that you could get away with a 1µF capacitor having an effective impedance of ~1.6kΩ at 100Hz.

 

Thank you @crutschow and @Hymie for your responses. One other question that that I have is this. Even if there is a significant quiescent DC voltage at the output terminal of the op amp connected to the anode of the 47 uF capacitor, there is no guarantee that it will be of positive polarity, is there? If it isn't positive, then the polarized 47 uF capacitor would have a reverse polarity voltage across it.

 

Op-amp offset voltages are known and specified as to tolerance and polarity in data sheets. And they certainly do matter in a lot of circuits.
In addition to all that, offset voltages are frequently temperature dependent, which can cause problems . So that series capacitor is quite important in real world circuits.

 

If it isn't positive, then the polarized 47 uF capacitor would have a reverse polarity voltage across it.

That is true.
So either use back-to-back electrolytics, or perhaps a smaller value non-polarized capacitor.

 

In an electrolytic capacitor, usually the problem with reversed voltage is increased leakage. In an audio application that may be an issue and it may add slightly to distortion, as a result. THAT is the motivation behind the direct coupled audio amplifiers. Unfortunately, it also provides a quick and easy path for massive failures if some component drifts a bit..
A bit of reversed polarity seldom leads to instant failure in most capacitor applications. So on many audio amplifier circuits there is no problem, mostly.

 

Further to what MisterBill2 says, if the tantalum capacitor has a 35V rating, with a +6V op-amp supply voltage the maximum reverse bias is likely to be of the order of 4V which is very unlikely to damage the capacitor.

 

Voltagegain i/p of NE5534 -> output 2nd TL072, 20dB-60dB (10 - 1000), however chain can be adjusted to zero offset by balance pot on first stage
Input offset NE5534 0.5mV typ, 5mV max x 60dB = 0.5v typ, 5v max but adjusted to zero
input offset 1st TL072 1mV typ, 5mV max x 30dB = 31mV typ, 155mV max
input offset 2nd TL072 1mV typ, 5mV max x 20dB =10mV typ, 50mV max

output of 2nd TL072 = 1v audio on max 205mV DC assuming NE5534 offset nulled out.

 

Actually I think that I'm barking up the wrong tree in raising the issue of output offset of of the op amp preceding the FW rectifier. My guess is that the way that the rectifier is configured results in its rest or quiescent voltage negative with respect to ground. Looking at Fig. 12 of his schematic of the receiver, you can see that Linkwitz has added circuit components to what would otherwise be a conventional rectifier. In his schematic, the resistor with the value 10 // 10 is two 10k Ohm resistors connected in parallel.

Edit: I take it back, the full wave rectifier is conventional, only drawn in a different way than what I have previously seen.

 

The fact is that I HAVE suffered from DC offset issues in an op-amp design. If the application requires DC amplification, then do not include a series capacitor, but you will need to adjust out the DC offset in each stage. That is a tedious process, at least I found it tedious. But if you are not in need of DC amplification then the series capacitor is a handy tool to use.