I have two XLH1's setup identically, same lens, aperture, shutter speed, white balance, gain, and no picture preset present. I have pointed them both at a grey card and captured a few frames of each via an Atomos Samurai.

When I bring the picture into Premiere, I can see a slight colour difference, and the picture from one of them is slightly darker than the other. It's as if the white balance is off. I have tried them both on Tungsten, Daylight, selected Kelvin temperature and even custom white balanced, and I get the same results consistently.

So one camera is slightly bluer than the other and the picture is slightly darker than the other. Maybe my expectation is off, but I would have thought that if the white balance was set the same (particularly if degrees kelvin was used) then the outputs would be pretty much the same.

Or is this expected and should be corrected by using the presets ?

Fresh from the factory or calibration they should be the same (within say a minor fraction of a stop) given the same lighting and exact same camera position and settings, including zoom. However, components age and could cause a slight drift relative to each other over time. Presets provide a way to compensate for this.

Interesting. They have both been back from a complete service over the last 4 months. The only material difference in terms of what was changed is that the 'correct' camera had new a new CCD Prism / realignment, so perhaps the issue is down to ageing CCD in the 'incorrect' camera. I was hoping that a custom white balance would have worked between the two, but it doesn't. I could send the camera back to CCD prism replacement as well, but if presets are just as good a way of ironing out this difference then perhaps I should go that route?

How far off are they from Matching? Can you read the RBG or YUV values of the charts from the video?

Matching the cameras would be a bit different than adjusting them to be within factory specifications. Specs probably read something like +/- X%. One camcorder could be as much as 2X% different from the other and both would still be within spec. (Looking at some Panasonic ProLine analog VTR alignment specs, +/- 5% appears to the the standard for luma, chroma, and video output levels. If I can remember to do so, I will look at a camcorder alignment spec this evening.)

As a point of reference (with little applicability to video) "matched" microphones (e.g., RODE M5) are typically rated to be within 2 dB of each other; 2 dB represents a roughly 25% voltage difference! 15 and 30mV are with in 2 dB of 20 mV. (A pair of AT AE5100 mics I have (not intentionally matched) were actually within 0.5 dB oveall based on a test using FM interstation noise.)

I would think that using presets, and color grading in the NLE are probably the easiest way to do it at the XLH1 price points. However, if matching at factory defaults is important contact Canon service, or a local competent repair shop to see if they can match them for you.

I attach two screen shots of the color rite chart, and a grey wall and skin tones. Both are zoomed in post.
The shots on the left are raw from the two cameras. The top slice is what I am calling the 'problem' camera and the lower shots are from the 'correct' camera. The difference is clear. I can correct in post, hence the shots on the right where the upper one has been adjusted to match (mostly!!).

I've had to do this kind of adjustment with different camera models, but this is the first time I've had two exact same model cameras, It does bring into question why have a kelvin white balance setting if the end picture is going to have a colour cast anyway? Perhaps this is a learning point for me, i.e. that no camera truly matches and that white balancing doesn't guarantee the end picture in terms of colour cast.

The RGB values for the white square are:

'problem' camera
R 216
G 244
B 234

'correct' camera
R 237
G 245
B 227

Interestingly, both cameras white balance was out in this shot so have both got a colour cast. What I have labelled as the 'correct' camera is more close to what I expected. I have a number of picture presets which I was hoping I could just simply load int the cameras that will give me the look I want for various shooting conditions, so I guess I will need to take into account the differences and create custom versions for each camera.