You need to turn your brain around completely from the familiarity of voltage sources (like batteries or the mains, or even the a.c. adapter!), when dealing with current sources, which is what a pure current transformer behaves as.
The c.t. sources a current. That secondary current is proportional to the primary current. The c.t. also, like every other transformer, has a VA rating. Unfortunately, few manufacturers and distributors specify it directly.
Most ADCs are looking for a voltage. The way to convert a current into a voltage is to use a resistance, specific to a c.t, it’s called the burden. (Historically, the burden would be an electro-mechanical instrument such as an ammeter or a wattmeter current coil, or a watt-hour meter current coil, or a combination of these.) Therefore, the value of the burden directly determines the voltage that appears across it.
The value of the burden also determines the power that the c.t. needs to deliver. That’s where the VA rating comes in. When the voltage becomes too great, the c.t. is unable to deliver the power, and the voltage you expect no longer appears. You can see that in the droop of the graphs on the data sheet.
So — the value of the burden has absolutely no direct effect on the rating of the c.t. But the voltage that the ADC can withstand, reflected back as determining the maximum secondary current and hence the maximum primary current, does affect the maximum that the combination of c.t. and instrument can measure. And that is precisely the case with input 4 of the emonTx, where the higher burden reduces the rating of that input to 18.33 A when using the 100 A c.t. [It doesn’t change the c.t’s rating.]
No, that wasn’t the error. The error is the headline ratio: “25:1” which defines a 25 A rating and a 1 A secondary current, and that’s in conflict with the data sheet that gives the rated current as 25 A but the nominal turns ratio as 500 : 1. ‘Very high output voltage’ defines a high VA rating, it’s not an open-circuit warning. ‘small size’ is a consequence of the ring-core construction and (probably) a distributed secondary winding.
The simple fact that the burden value is mentioned means to me that it isn’t integral. I’ve never seen it mentioned by any other manufacturer in their data for the voltage-output types. The value is the maximum recommended value - clearly good to give 1.6 V rms at the rated current, therefore good for a 5 V span ADC, and as I’ve so often written here, any lower value to suit the lower voltage that you might need - including a short circuit (when of course the c.t. is delivering zero power into its burden, but still dissipating the same power internally - I²R using the rated secondary current and the winding resistance).
And then what? Use a voltage divider to reduce the burden voltage to the 1.1 V rms that the 3.3 V ADC of the emonTx / emonPi can handle?
No, I would regard 33 Ω as the maximum recommended value. That, with the 25 A c.t, would give you an input rated at 16.66 A for the emonTx/emonPi, but it would be much more accurate at low currents than the 100 A YHDC with the 120 Ω burden.
Maybe we can persuade @glyn.hudson or @TrystanLea to obtain a sample of each so that we can verify the performance and measure the phase error etc.