Recently, I found a MN3010 chip, in a pretty fancy and unconventional ceramic/gold/metal package. Turns out, it is a DAC; but not a normal DAC – rather a 2 digit BCD DAC.
Quickly hooked it up to an 8 bit port, counting in cycles, from 0 to 255… This is the resulting output:
Bottom line is about -13 volt, top is 0 V. 2 V per vertical division.its
Notice the strange steps – this is a clear indication of the “BCD” nature of this chip – 4 upper bits encoding the output in 1 volt steps, and the lower 4 bits encoding the 0.1 V digit.
Now, to put it to a test – re-programmed the counter to actually count in BCD digits, from 0 V, to -10.9 V – which is a bit above the specification – the device is supposed to output 0 to -10 V.
Next, let this thing run for about 18 hours recording the voltage at each of the allowed input codes, using a HP 34401A 6.5 digit volt meter, using a little program and GPIB cable.
These are the output voltages, as a function of time, for certain fixed codes:
… I would say, the DAC is rather stable: it consumes about 0.5 Watts, and is running warm, but it appears to be reasonably insensitive to temperature, far better than 0.1 LSB (more like 0.01 LSB!)
After recording all the numbers, some number crunching using well-known Excel…. multi parameter linear regression.
Regression analysis yields the voltage values associated with the bits, they are close, but the 0.2 and 0.4 V have some deviation. It’s a pitty I didn’t test this device about 30 years ago, then we would know if this deviation is due to age, or just the result of normal manufacturing tolerances (certainly all within specified limits).
Here, as a last graph- the total deviation (which contains all error sources, including any offset errors), accordingly, the output tends to be 0.005 to 0.02 V (0.05 to 0.2 LSB) low, digital code vs. actual output voltage. Pretty good (limit is +-0.25 LSB), still.
Now, let’s keep this device another 30 years, and repeat the test…… I will keep you posted.