This call started with the engineer insisting that the INA126 instrumentation amplifier was loosing gain as temperature increased. This occurred with over 50 percent of the devices he had received. I was a bit skeptical but willing to listen if he would send me a copy of his schematic. The critical section is shown here.
Figure 1. AC coupled stages can mask DC flaws in operation.
The circuit was operating on a single 3.3V supply with the VB equal to 1.2V. On my request he sent in a handful of parts, which I tested and found no flaws. He next sent in six parts that were known to be bad. These were the worst of the worst. Again, no problems appeared.
At this point both of us were a bit frustrated so I asked for a complete assembly so I could look for unspoken interactions. Testing this unit began to unravel the problem. The signal from the sensor was a pulse of a very low magnitude and about ten microseconds in duration.
Once I saw the nature of the signal I had a suspicion that I knew the problem, but first I needed some scope time. On some units, as the amplifier heated up, the pulse into the A/D converter grew shorter. If I moved the scope probe back to the output of the instrumentation amplifier it became obvious why. The DC output was rising until the pulse bumped into the output voltage limit. On the other side of the capacitor it looked like the pulse was simply growing shorter. In the eye of my caller the instrumentation amplifier gain was decreasing as temperature increased. On some units the pulse was reduced to zero.
There are several guidelines for gain distribution in a signal chain. Where frequency response is of prime importance, the rule of thumb would be to equalize the gain across the stages. Therefore, in a two stage system, as we have here; set each stage at the square root of the total gain needed. This results in a maximum bandwidth channel
Another rule of thumb has been to place as much gain as possible in the first stage. This is based on the idea that a high precision front-end, running at high gain minimizes the signal corruption contribution by the subsequent stages and the subsequent stages do not degrade accuracy. Then the second stage operational amplifier can then be a device of much less precision and therefore a lower cost device. But enough is enough.
The solution here was to drop the front-end gain to 750 and increase the second stage gain to eight. This gain distribution made the most of the precision instrumentation amplifier and allowed a low cost operational amplifier to be used as the second stage.
Thanks for reading…………Comments invited, Please. firstname.lastname@example.org
And be sure to copy, what's-his-name, email@example.com…