Hi Matt,
Looks like my QA402 shipped today and I want to be crystal clear on the input protection. Is one of the following statements true and if not can you define the input action.
Can handle the Max value on any range and just clips it internally.
It can handle the Max input on any range and it auto ranges to protect itself.
It does not auto-range and can’t handle Max input at lower ranges.
Hi @ggorman,
This is a very good question. It’s a combination of 1) and 2). When you hear the input attenuator click it means the input is larger than the range you have picked. The activation of the attenuator is not a problem at all–as long as the input level is less than the max input level of +32 dBV, the system will be able to self protect regardless of input level you have chosen.
If you are interested in more detail, that is below.
When you apply a value that exceeds the full scale, the ADC indicates an overflow internally, and that is used to engage the input attenuator for one second. After 1 second, the attenuator releases and the process starts over.
It’s a non-destructive process. You can see how it works when you get your QA402 via the following:
Now, first let’s look at +18 dBV out with +42 dBV in (drag a box with the mouse to zoom). Below you can see a burst measurement with the QA402.
Zooming in more, note the arrow is roughly where we cross the 1.41V level after a few cycles of ramp. On the 0 dBV input range, this will be full scale.
Now switch to 0 dBV input and keep the +18 dBV output and run a single cycle again. Below you can clearly see where the ADC limits (at about +/-1.48Vpk). From the graph, you can see the first limit occurs around 3.1 mS and you can see the attenuator kicks in at 5.3 mS. This is about 2 mS, and it’s a combination of a bit of hysteresis + the relay pull in time.
Note also the RANGE indicator in the upper left. Whenever you see this, it indicates at some point during the acquisition the front-end attenuator was engaged, and thus the measurement isn’t reliable (if you are doing remote measurements via the web interface, you should see a non-200 HTTP return code with a message signifying range issues).
Now, during this 2 mS before the atten has kicked in, the input protection is actively steering the excess voltage inside the unit through a special pulse-withstanding resistor and some steering diodes to a pair of TVS devices. The pulse-withstanding resistor can take 10W for 10 mS, and the steering diodes can withstand 100 mA (non-repetitive peak). And at the maximum input +/-56V peak, the current through the diodes is 100 mA.
So, in short, having the relays “protect” an over-voltage event is not a big deal at all. Now, if you are going to be running in a factory and routine overvoltage is going to happen, then there are two considerations to keep in mind: First, the relays probably have an upper life of 800K cycles or so (virtually zero current flows through the relay normally–manufacturer life curve below). So, if you are switching the relays every 20 seconds (180 times per hour) in automated testing around the clock 24x7 in a factory, the relays will probably wear out in two years. For normal lab use, you’ll likely never encounter this limitation.
Second, in the input capacitor is an aluminum bipolar cap. These caps have a finite life as the electrolyte will eventually dry out. BUT, when treated nicely (as they are in the QA402) they can last for decades. But if you are going to drive the inputs with very low frequency square waves of +/-56V, then it might accelerate the demise of the caps. it would likely manifest as a shift in capacitance, and you’d see the low-end corner of the QA402 change.
In short, for normal use doing normal things on a bench, and encountering a “RANGE” event (overload protection relay activation) is no big deal (as long as the input is < the max input of 32 dBV). There’s no need to “act fast” or panic. The QA402 will take care of itself.
PS. If you want to see the full cycle of protection, just pick an FFT that is longer than one second and you can see the full protection cycle:
