I think the idea is that Phd programs are paying for you and they want evidence that you understand research and want to do it and the only way to show that is having done it. Plus they want evidence that you are potentially effective as a researcher so they want the letters of recommendation for that. I think you can look to do a research based MS program to prove yourself. And at some that MS can morph into the Phd program. At UW-Madison, they admit MS and PhD as a single pool, so that means that you do not have to apply again to do the phD, plus they try to fund MS students to the extent they can, I think somewhere around 85pct funded. So to move to the PhD, I’m not sure what you need aside from passing the quals, faculty or department support I suppose, but it is more seamless (this is true for CS I am not sure it works exactly the same for engineering.)