Here's what I think is happening though I'm not sure exactly why.
Starting with everything off, you switch CH on. The CH-ON voltage from the controller (coming via the room stat) drives the valve to its mid-position (HW and CH both on) but the boiler doesn't start yet. Inside the motor head, two switches operate. One opens to reduce the motor power to a low level, just enough to hold the valve against the return spring. The other switches closes.
If, and only if, you also have a HW-OFF voltage coming from either the controller or the tank stat, that second switch connects it through to the motor which now drives all the way over to the CH only position. At this point, a third switch closes and connects the CH-ON voltage to the boiler, which finally lights.
So far so good, the valve has done exactly what it's supposed to do.
If you listen to the motor during this time, you'll hear it turning as it moves the valve but the boiler will only come on when it stops. As long as you have HW-OFF, the valve will be held in the CH only position and the presence or absence of CH-ON will control the boiler.
Your problems begin when HW-OFF is replaced by HW-ON. At this point, HW-ON supplies the boiler, just as it does when you want HW only. It has a direct connection. Meanwhile, with HW-OFF gone, the spring pushes the valve back to it's mid-position. (You should be able to hear this.) Everything appears to be fine in that you have both CH and HW - and yet something in the motor head isn't right!
My best guess is that the valve has run back just a little too far.
Remember those two switches that operated when CH-ON originally drove the valve to mid-position against the spring? One of them closed, allowing HW-OFF to drive the valve all the way over to CH only. I think that one has opened - though it shouldn't have. Meanwhile, the other one, which dropped the motor power coming from CH-ON to a low level, is still open. The valve is sitting in limbo, held just short of mid-position. You will have CH for as long as HW-ON powers the boiler but HW-OFF cannot drive it over to the CH only position - not until you give that lever a very slight nudge.
If you can confirm all this by careful observation of - and listening to - the valve as it goes through the sequence CH only -> CH+HW -> CH only -> nudge lever (while also watching what the boiler is doing), it's a pretty fair bet that the new motor head is faulty.
I just thought of this last night: what if the HW-OFF and CH-ON connections were crossed over at the valve?
Thinking about it, a lot of things would still work as expected.
1) HW and CH both off: The valve sits in mid-position, driven there by HW-OFF, but the boiler doesn't light because that requires either HW-ON or the valve right over to CH position.
2) HW only: HW-OFF disappears, allowing the valve to run back to the HW only position. (You can listen for this.) HW-ON lights the boiler.
3) CH only: HW-OFF drives the valve to mid-position and CH-ON takes it on to the CH only position. That's the wrong way round but so what? As far as the valve is concerned, it's just two wires with volts on them. It's now HW-OFF that powers the boiler but this doesn't matter because, when the room stat removes CH-ON, the valve drops back to mid-position and the boiler goes off.
4) CH first, then HW: CH-ON is holding the valve in the CH only position. HW-OFF, which was powering the boiler, has gone but HW-ON takes its place. You will only get HW by leakage past the valve - until the room stat cuts off CH-ON, which leads us to ---
5) HW first, then CH: This is your problem condition. The valve is in the HW only position so CH-ON can't get past that switch in the motor head; not until somebody pushes that lever.