Choosing capacitors for AD725 encoder

Question

This question generally applies to many situations, but in this case I am trying to figure out the best choice (no limit to the perfection).

I am stuck at both input coupling capacitor and output coupling capacitor.

1. Input coupling.

   Datasheet states that 0.1 uF capacitor is recommended. It shows non-polarized capacitor (but that does not 100% guarantee it may not be polarized). People just copy the reference circuit from the datasheet putting caps they have or are easily obtainable - X7R of 0.1 uF.

   However reference design uses 1000 pF caps, with this unconfirmed brief info on the AD forum that AD must have used X7R caps. I am skeptical about it, and given tons of documents and opinions read last days I suspect that AD actually uses NP0/G0G 1000pF capacitors (no BOM for the device is available, and I suspect asking AD will not give any outcome). This nominal exists and is available in 0805 package. I did not find direct mentions that X7R are bad to use on the video path, but there're numerous mentions that it is not so good for audio paths, however minding voltage rating of cap (bigger is better).

   Am I correct in my conclusions, and it is much better using NP0/C0G as input coupling caps? Or even electrolytics?

2. Output coupling.

   I bought EEU-EB1E221SB, which is 220uF/25V rating. Now I am not that sure that I have chosen correct cap. Forums and publications mention that lower ESR is better, but I can not find ESR informaton for EB family of caps. This information is only available for F* families. Looking through forums found out that F* must be "audio-grade" caps. Panasonic does not state the target usage for the families at all, it would be much convenient if they list typical applications for the families.

   Do "audio-grade" electrolytic caps apply to the video transmission bandwidth and minimally impede video signal transmission? How to choose this cap type for the video application?

---

I want to hear your experiences, supported by the proven theory. Please refrain from statements that something will work just OK in generic cases. Please note I do not ask about nominals of the caps, I am asking about technologies of the caps.

Answer 1

Input coupling caps are there primarily so that the internal DC clamps can find the black level and as such they can be X7R (I have used those for this purpose) although 1000pF C0G would probably be appropriate.

The real advantage of C0G is the temperature coefficient (and Analog Devices video parts can be very susceptible to signal path variations due to thermal issues - I do not know of this is the case here).

The datasheet states:

The inputs should be held at the input signal’s black level during the horizontal blanking interval. The internal dc clamps will clamp this level during color burst to a reference that is used internally as the black level. Any noise present on the RIN,GIN, BIN or AGND pins during this interval will be sampled onto the input capacitors. This can result in varying dc levels from line to line in all outputs, or if imbalanced, subcarrier feedthrough in the COMP and CRMA outputs.For increased noise rejection, larger input capacitors are desired. A capacitor of 0.1 μF is usually adequate.

So the larger capacitor (and 100nF is very common) keeps noise levels low.

For the output capacitor, keep in mind that the output is designed to drive 150\$ \Omega\$ so it is wise to keep the ESR reasonably low. Low ESR tantalums might not be appropriate as their capacitance rolls off quite quickly above a few 100 kHz although the large capacitance is for the low frequency components. As I have a lot of different reasons for not using tantalum devices in general, I have not used them in this particular scenario.

AD does not have the best reputation for support from (at least the European) video support group incidentally.

I have used OS-CON devices for the output capacitors and they worked very well. These have ESRs on the order of < 40 milliohms.

I don't really think 'audio grade' devices are necessary for clean video; they are optimised for performance in the audio spectrum which is only a small part of the bandwidth of a composite video signal.