Welcome in the forum, Chris.
Your description smells a bit as if the PSoC doesn't come out of reset fast enough. Are your voltages each bypassed with caps as suggested?
I am running virtual machines on macintosh, and have had some of the same issues.
The first run of our boards, we missed the VDD/VCC lines that *should NOT* be tied to the 5V or 3.3V power supply. Those boards were DOA. If you are running at a lower voltage, your boards might not be DOA, but horribly crippled.
In regards to the 68pin parts, we often have at least one pin that doesn't solder right during oven heating. It varies, but is usually a pin not necessary to functionality and we reroute in the code.
I would suspect the same for you. A hot air gun and some flux might fix this. We have a 50/50 success on prototype boards fixing this.
I heartily recommend using the TQFP 100 package where space permits. It's gull wing design allows for fixing such stuff for small run boards.
Thank you Bob for the tip.
The reset pin (XRES, pin 10 on QFN68) is tied to 22k pullup to Vddd (~3.1V). The only other connection is to the Miniprog3 via 3" wire.
Thank you WSM.
Rechecked the supply voltages and I have all of them correct. However, I have read somewhere that I still need to turn off the internal 1.8V regulator - as its still creating 1.8V - from my 1.8V supply - somehow.
Definatly will use the TQFP next time. It took more than an hr to put the QFN on, but all the pins are soldered. (I checked again under the microscope.) The other trick for QFN is to make the footprint much larger so that the iron can touch the pad. But I've also ordered a hot air pencil to help with this board.
Re bypass caps, all power pins have a 0.1uF around 3mm from the pin. (The ground return path via the GND plane is ~5mm). For each digital supply 3.x and 1.8, there is a single 1uF cap 'nearby'. The 3.x V supply doesn't have any other caps (probably need one for the LDO), but the 1.8V switcher has 10uF ~26mm from PSOC center.
So I think the answer to the question is to update the software.
My PSoC Programmer was version 3.22.0 ? (build 2034?) - I've updated to 3.22.3 (build 2043?) and it started working. At least I think that is what seems to have fixed it.