I have to agree that it's not likely that the manufacturer is at fault here. However, I'd mention that, in the 35 years or so that I've been fooling around with this sort of thing, I've had exactly one component with built-in crystal oscillator fail to oscillate. That was in the last 5 years, and I'll invite you to guess which manufacturer produced it. If the ATMEL folks had rotated the die 90 degrees counterclockwise in the lead frame, the pinout would have been much more conducive to a good oscillator layout, however. The crystal connections would have been on pins 9 and 10, so the crystal could have been mounted at the pin 10 and 11 end of the IC, and GND from pin 15 would easily route to the two cap's attached to the crystal. Marketing probably devised the DIP pinout.


You're right, in that 10 cm is entirely too long. That places the crystal and/or cap's 2 inches away from the MCU's pins. However, it happened often enough and, when it did, it seems to have worked out OK. You're right, too, in that 10 cm at over 10 MHz would radiate quite a bit, hence wouldn't pass current EMC standards, but we didn't have them in the mid-'70's.


So far, one of the things on which we certainly agree is that solderless breadboards are a poor choice of prototyping environments for MCU's and for circuits operating at frequencies far above the audio range.


True enough, but now there's an additional question, namely what the difference between the two "breadboard" circuits is. Since a change in the wind direction, seismic activity, or phase of the moon can alter the characteristics of a "solderless breadboard" circuit, I doubt much can be learned from this.

I basically disagree with the notion that "solderless breadboards" are unsuitable for digital circuits, but I freely admit that their behavior is very much affected by the "solderless breadboard". Even with properly terminated low-impedance signals using ECL, you can observe the interaction of adjacent signals at frequencies above 10 MHz. With high-impedance circuits, e.g. ACMOS at 40 MHz, I've observed state changes repeatedly occurring when I passed my hand between the circuit under examination and an adjacent computer display. At frequencies well within the audio range, ~10 KHz, one can observe the interaction between adjacent signal paths within the "solderless breadboard" without even increasing the amplification of the signal on the 'scope display. After all, the contact-contact capacitance, at least on my "solderless breadboards" from Continental Specialties, is purported to be typically 5 pF and the capacitance of the probe is not much more than that (~8 pf).