Archived Forum

It's unlikely to be a problem. All you need is a relatively low (but not too low else it's possible to overload the protection diodes in the event of a fault) impedance to bypass the divider chain. (I assume you're looking at the explanation in Building Blocks and NOT the production schematic of the emonTx V3.)

I believe a ceramic cap marked 10 would be a 10 picofarad cap - much too small as RW pointed out. The largest ceramic cap values are typically in the 0.1 microfarad range.

Correct me if Im wrong, Robert, but shouldn't that cap value be a minimum of ~1 microfarad, give or take a bit?

Regards,

Bill

Thanks for the replies.

P.S. 10uF @ 16V it is.

http://www.digikey.ca/product-detail/en/FK26X7R1C106M/445-8550-ND/2815480

I bought a bunch before I realized the EmonTx was using electrolytics.

Bill: As I said, there are two conflicting requirements. The first and most obvious is to tie the centre point of the divider to ground for alternating currents in order to provide a solid reference for the voltage generated by the current flowing in the burden resistor (or for that matter for the divided-down voltage from ac adapter) , and the second is there needs to be a bit of impedance present in the input circuit so as to limit the current into the protection diodes on the ADC input pin in the event of a fault. Incidentally, that's the reason for the 1 kΩ resistor included in the emonTx V3 input circuit. A 10µF capacitor in steady-state at 60 Hz is marginal, and whilst there hasn't been a reported case of damage, it would be better not to go above 10µF. In the other direction, and considering what happens when the multiplexer switches, that capacitor has to source enough charge to fully charge the sample & hold capacitor, and the source impedance has to be low enough to allow that to happen in a timely manner. The recommended maximum source impedance is 10 kΩ. Therefore if the capacitor is too small, the voltage will change unacceptably as the multiplexer switches. The S&H capacitor is 14 pF, so if we want less than 1 count error, the capacitor needs to be at least 1024 times this on that criterion alone. Accepted wisdom is that these capacitors should be as large as possible, a nominal 10 µF will have an impedance of about 350 Ω (worst case low) at 50 Hz to 244 Ω (worst case high) at 60 Hz, assuming 10% component and 1 % frequency tolerance. A 1 µF will clearly be 10 times these values. I would feel happier with 3.3 or 4.7 µF than I would with 1 µF. It would take a long time to scour the data sheet to get the data to construct an exact model, but why worry when space is not at a premium and there's not a significant cost advantage?

Right, the main point I was aiming for was that a ceramic cap marked 10 was likely to be a 10 picofarad cap which would be much too small. Understood regarding ADC input impedance vs AC return path for the burden resistor.

My misjudgement regarding his cap was thinking it was a disc ceramic as opposed to the MLCC type.

Which brings to mind a question. Would a low ESR type electrolytic be a better choice then?

The linked catalogue page says it is 10 µF, and the "106" in the part number is a hint that that it is indeed 10 x 10⁶ units large, so I think it will be 10 µF.

Saw that when I followed the link. Still stuck in the electronics "stone age." Old habits die hard!