Some more information: GM, like all automotive manufacturers I am aware of, uses a multi-bus vehicle architecture. Typically, there's one for "infotainment" (including radio/head unit, navi, and sometimes HVAC and other "comfort" associated subsystems), another for the drive train, another for body-control stuff (doors, windows, etc.), and sometimes even another for safety systems (airbags, etc.). GM uses a "single-wire CAN bus" for infotainment which is relatively low speed (maybe 100Kbps); they are alone among major manufacturers using this "flavor" which is completely dependent on solid continuous grounding of all elements on that bus. There is always a gateway as well that allows communication among nodes on the various buses (example: a NAV head unit can get wheel ticks and steering angle from the drive train bus to perform dead-reckoning during loss of GPS signal). It is possible that garbage on the infotainment bus due to a bad ground oculd create havoc on other buses via the gateway, although a PROPER system design would prevent that. Just about every other manufacturer uses the 2-wire physical layer version of CAN instead; since that is MOSTLY differential, it is less dependent on grounding, at least within the common-mode tolerance of the (industry-standard) 2-wire drivers. The MOSTLY is because there is one aspect of signalling that IS ground-dependent even on the 2-wire type. This is the "wake-up" signal which uses a common-mode DC shift as its signal mechanism. BTW, the commonly used physical layer for the "body-control" part is LIN, which is a very low-speed single-wire type. The primary motivations for using single-wire buses at all is very simply the severely inflated cost of copper, and WEIGHT.
I agree Tekochip -- this would be a hard one for the repair shop. This simple problem could have cost an arm and a leg to fix at the shop. In this case, the solution may have bewewn to go to a dealer's shop, since the dealer may have info prompted by numerous occurances. Then again, maybe not.
This is the sort of problem that will pop up when shortcuts are taken in the wiring between different modules in the car. The radio ties to the control module so that it will only function in that vehicle, so a stolen radio is worthless. I learned about that after purchasing a radio from a car in a scrapyard. The radio needs to handshake with the BCM and ECM in order to work.
Now if the connection gets noisy all sorts of bad data will be sent to all of the modules. The reason that the poor ground caused problems is that the harness does not include a dedicated common for the data bus that has a low enough impedance to allow adequate connectivity without the additionsl external grounding connection. It might function correctly with all of the wiring and modules sitting on a display board and no external connection, but in the autobody environment it needs that extra ground connection.
I suspect that the poor ground connection was creating noise on the data bus that was confusing or swamping out the data signals. Since typically ground is common to all car functions, it would not have to be part of that particular data bus to wreak havoc. As far as standard troubleshooting technique, I participate in a BBS dedicated to vintage Porsches. Whenever someone posts that their turn signals are not working correctly, the universal reply is "Check your grounds!" Even on old cars this is not so easily done as there are numerous connections. On newer cars (2003 seems to be a bad year for GM) there are grounds hidden all over the place. Anyway, however unlikely the solution to this problem might have been except by chance, the standard troubleshooting procedure to check the grounds first is not unknown to the experienced mechanic.
The garage tech would most likely tell you that there is no way that your radio could have any effect on your engine and then spend your money replacing all of your sensors before bothering to pull the radio.
Not too long ago I had to replace a passenger door latch assembly in our 2007 GMC Yukon. While doing so, the ignition was on accessory and I was listening to the radio. (Yes, the first rule is always to disconnect the battery, but this seemed pretty benign as far as the likelihood of causing a dangerous short.)
Disassembly of the door required unplugging the window/lock switch module in the door from the wiring harness. I discovered that each time the module was unplugged, the radio would change stations! Factory service manuals later confirmed that the door switch module and the radio were both connected to a common and very simple low-speed (~30kbps if I remember correctly) data bus. Unplugging the module must have introduced noise or power-down signalling or something onto the bus in a way that made the radio think it was being addressed and commanded to change stations. It makes one wonder what else could have been commanded?
It seems plausible that the radio in the Saturn may have shared a data bus with the engine and/or transmission controllers. If the bad ground caused it to make the bus inoperable, it could explain the observed behavior.
I recently watched a diagnostic video online (look for scannerdanner if you are interested) showing a case where a shorted sensor pulled a 5v reference from the engine computer very low and the vehicle would not start -- fuel pump inoperable, no spark, etc. Perhaps a radio without a ground could be driving a data bus high or low to the point that it impact other related circuits, including something related to a knock sensor.
This article demonstrates the problem with modern-day auto technicians. You can teach theory until you're blue in the face but you just can't teach common sense. Every Sherlock Ohms article appearing here should be put into a book and made required reading for all budding technicians. I have yet to see a simple fix that would have been found using 'standard' troubleshooting. There was one where when the truck was parked in a rutted field, it wouldn't start. No problem anywhere else but every time this operator parked in this particular field, the engine wouldn't crank. It turns out the engine ground strap was designed just a little too short and with the left front wheel 8" higher than the other 3 wheels the ground strap would be pulled taught and lose contact. This required the technician (me) to accompany the owner to his work-site and even then I had to use jumper cables to solve the problem which then inferred the battery ground was being lost. Using a jack in the parking lot proved the theory. I lost 4 hours I couldn't bill that day but made a customer for life.
Focus on Fundamentals consists of 45-minute on-line classes that cover a host of technologies. You learn without leaving the comfort of your desk. All classes are taught by subject-matter experts and all are archived. So if you can't attend live, attend at your convenience.