LTspice datasheet.

Should I Believe the Data Sheet?

I asked the question, “What is the true value of LTspice®?” The straightforward answer was, “Insight,” and I shared an example where LTspice provided insight into the behavior of a circuit. This was in my previous post.

Here, I ask a different question, “Should I believe the datasheet (when it seems to diverge from what I see on the bench)?” When there is a discrepancy between the datasheet and bench results, can we resolve the difference? In the following real-life example, the insight afforded by an LTspice simulation allowed us to do just that.

Background

In this case, the customer was using the LTC7001 fast 150 V NMOS static switch driver to drive two MOSFETs in a back-to-back arrangement - to protect their load from voltage drops on VIN. As they had a capacitive load, they were also concerned about inrush current. In the lab, this peaked at around 6 A, beyond the capability of their main power rail, so they needed to limit the inrush current to protect the upstream supply.

Crunching the Numbers

Using the equations in the data sheet, they calculated the correct circuit values for a 2 A current limit. Since they could not immediately test on hardware, they simulated the circuit using the calculated component values in LTspice and noticed some strange behavior in the charge pump circuit: If the supply voltage (VCC) was lower than 7 V, the voltage level on the bootstrap capacitor (VBST – VTS) was lower than the nominal 12 V used in the datasheet equations and consequently the time for the switch to turn on was much longer.

They had planned to use a 5.1 V supply, so they reworked the calculations, replacing 12 V in the equations with 7.1 V, the lower bootstrap voltage. Everything seemed to be in order. However, when they made the changes suggested by the equations in the data sheet, they found that the inrush current was limited to 1.2 A instead of the calculated 2 A. They confirmed this on the bench. At this point, they threw up their hands in despair.

In this case, they needed their calculated predictions and simulation to match real-world performance, showing that they understood the circuit, as designed, in their review process. They even got to the point of considering whether they should increase VCC to >7 V just so the datasheet equations would hold true…

Lies, Damned Lies, and Data Sheets

I assured them that such drastic action was not necessary and that we could reconcile the data sheet equation results and the bench measurements, using LTspice to help.

Equations in data sheets have the look of authority about them, akin to the Ten Commandments, set in stone, and immutable. Look closely though, and you will see the "≈" sign in place of "=" in many of them; less carved in stone and more written in pencil on the back of an envelope. Furthermore, the equations in the datasheet are based on assumptions and are valid only if those assumptions are true. If not, then the equations will not match real-world performance.

As the customer found by measurement, there are some variations in the behavior of the charge pump circuit that provides the bootstrap voltage due to VCC. This is indicated by some of the typical performance curves in the LTC7001 datasheet:

LTC7001 Typical Performance Curves

The equations assume that VBST – VTS = 12 V and that IBST = 30 µA. Looking at the curves, for VCC = 5 V, VBST – VTS ≈ 7 V, which matches our simulated value. This means that IBST is going to be under 5µA, based on VTS = 5 V on the second graph assuming it’s slightly lower than the 6 V (dashed blue) line.

LTspice Insight

Measuring <5 µA in real life is a challenge, so we can use LTspice to check that our reading of the data sheet has been correct.  The LTspice file I used is below:

LTC7001.asc

When we run the transient simulation, we can check the value of VBST – VTS is 7.1 V by entering the expression ‘V(bst)-V(ts)’ in the Expression Editor in the plot window and the cursor to measure it:

LTspice Expression Editor

Vbst - Vts plot from LTspice

We can also check IBST, holding down the control key and clicking its label to integrate it to get its RMS value:

Current flowing out of BST pin

The simulation shows approximately 3 µA RMS matching the datasheet curves. Recognizing the value of IBST is an order of magnitude lower than expected, the customer adjusted the component values to suit.

Conclusion

It turned out that all the information was in the datasheet, but not necessarily all in one place. LTspice helped the customer bring the required equations and typical performance curves together to reconcile the equations with the measured results. Find the next blog in this series here.