Transcript
Testing the Timing Accuracy of the BOLT
Today, we’re going to be testing the timing accuracy of the BOLT. Time is, of course, important when you’re collecting data and the BOLT has two mechanisms for recording time. It has what’s known as a real-time clock that allows it to keep track of the time of day and accurately timestamp information to the second. And the heart of that is the real-time clock that produces a one pulse per second output that is then fed into the rest of the BOLT’s circuitry.
The other aspect we’ll be testing is the BOLT’s ability to phase-lock to the power line frequency, because some of the timing that the BOLT performs is in terms of counting 60 hertz cycles. So, its power line tracking is another time aspect that we’ll be testing here at room temperature and over temperature.
Test Equipment Setup
To test the BOLT’s accuracy, we have some equipment from the PMI test lab. We have a three-phrase power supply, of course, supplying stable voltage to the BOLT, and then we have a rubidium atomic clock. This provides a very accurate time reference for the frequency counter.
This atomic clock is based on rubidium atomic oscillations, and that produces a 10 megahertz reference clock that is accurate to 10 to the minus nine or better accuracy. So, this feeds into the frequency counter. The counter is accurate as a standalone device, but it’s even more accurate when it is tied to the frequency standard. So, we have an accurate time base here to measure another clock like the BOLT.
To do that, we’ve put the BOLT in an environmental chamber, and we are looking at internal signals in the BOLT. Their real-time clock pulse per second is not a signal that’s brought out to the outside, so we have signals that are going to the digital circuitry in the BOLT, and we’re connecting to these signals with an oscilloscope so we can look at them on the scope and also put them into the frequency counter to measure at the circuit level our one pulse per second output. We’re going to be doing this at room temperature and then use the chamber to go hot and cold to make sure it’s accurate at the extremes of temperature.
Room Temperature Measurement
So, the BOLT is inside the chamber. We have the oscilloscope that’s looking at the pulse per second, and we’re also running it into the frequency counter. On the scope you can see this is a negative going pulse, and this comes out from the real-time clock inside the BOLT once per second. With the scope, we’re gonna look at drift over temperature. We’ll do that after we’ve made the room temperature measurement.
We have the pulse per second going into channel one on the scope and teed off to go into channel one of the counter, and the frequency counter is measuring the period of this one pulse per second. And here, it’s reading 1.000001397 seconds. So, let’s put that into our sheet. That gives us an error of 1.3 parts per million. That’s a very small amount of error. That’s a few seconds per month, so that’s a very accurate real-time clock. A typical real-time clock is on the order of plus or minus 50 parts per million over its entire temperature range.
We can set the scope to show us that drift over time. Here, we’re turning on the negative going edge of the second. So, you can see the negative going edge of the pulse. This is the pulse one second after the trigger. So, the first pulse was way over here. Now, we are looking at the one-second-later pulse. So, now if there’s any drift at all over temperature, we’ll see this slowly move back and forth.
Line Frequency Phase Lock at Room Temperature
So, we’ve seen that the BOLT is very accurate at room temperature, but let’s check to see if there’s any drift at hot or cold temperatures. The second test that we’re going to do at all the temperatures is measuring its line frequency phase lockability. Here, we’ve set the AC power supply to exactly 60 hertz, and the BOLT is synchronizing to that, and that measurement output is accessible through PQ Canvas.
Here, I have a live connection through wifi. The BOLT is joined to wifi right now. I’m connected to PQ Canvas, and I’m looking at its real-time frequency display. And we can see that it’s 60.003 hertz. Let’s take a look at the frequency counter, and we’re measuring 60.0028 hertz. That gives us about 2.3 parts per million difference, extremely accurate timing at room temperature for the 60 hertz phase lock.
Cold Temperature Testing
So, now let’s go cold in temperature and see if there’s any drift in the BOLT in either of its time measurements. We’ll set the chamber down to minus 40 Celsius, and we’ll give that a good time to soak and reach its final temperature, and then we’ll make these measurements again.
Okay, now we’re at minus 40 degrees Celsius. We’re at the lower end of the BOLT’s operating range, and we can see on the scope that we see a little bit of drift. We’ll turn on the cursors. That’s about eight microseconds or so. We look on the counter, we’re at 99.9618 hertz or millihertz. Let’s type that into the sheet. That’s about 38 parts per million difference, which is still well within a plus or minus 50 ppm of a good real-time clock.
And if you look at the frequency, the output of the supply is unchanged. We’re at 6.02, 0027293, and that still matches PQ Canvas of 60.0003, so that’s really unchanged. So, that’s still two to three parts per million measurement accuracy on the line frequency.
High Temperature Testing
Okay, now we’ve let the BOLT soak at high temperature. Let’s take a look at the real-time clock accuracy. We can see that on the frequency meter, we’re reading 0.99962 seconds. That gives us a difference of 38 parts per million, so roughly the same drift as we had in cold. Still well within good real-time clock performance.
And if we take a look at the scope, we can verify the same measurement. We have about 37 microseconds difference, so that corroborates the frequency measurement here with the frequency meter. We have the AC power supply still putting out that same frequency. So, we’re still at 60.003 roughly, well under a two parts per million difference.
So, the BOLT is very accurate and shows absolutely no drift on power line frequency phase-locking, and the real-time clock is well within its spec. It should provide good accuracy over time.
