OK, let's continue with the RC BLDC inverter offtopic. For the possible motor inductance discharge, the problem with a TVS diode is that it's so imprecise that it imposes an additional maybe 40-60% Vds overrating spec, hence an active crowbar circuit may make sense over it for a tight, economical design, if you can afford someone's time designing it. (Electrolytics almost always impose the same kind of additional overrating requirement since you can't store much energy in them unless you allow some dV. You can calculate it!)
(By additional requirement, I mean, for a 25.6V battery pack, you end up with actual TVS clamping at around 35-40V, and at that point, you haven't added any voltage robustness to the system yet; you still need to overrate this like you would normally do for very local layout parasitics, safety margins, etc.
For example, I'm just now designing in a SMC5K28A TVS in a 6s-li-ion driven 2kW BLDC system - it's not allowable to let it leak during normal operation, so this is the closest part at 28V working voltage. A 26V part could maybe possible, but as these working voltages are defined at room temp only, I did say no to it.
Now, it breaks down at max 34.4V, but only at 1mA - actual max clamping voltage is as high as 45.4V at 110A surge. In reality, if the motor current is less than this 110A, the clamping voltage is going to be something between these two numbers, let's say 40V. So now we are 59% over our nominal maximum DC bus voltage, and still need to account for normal switch node ringing, whatever derating we do; well let's say 30% if you know how to do layouts. Of course, such an overvoltage event is not commonplace, so we may be able to depend on additional avalancing in the FETs.
Now, let's see how the elcap-only solution would work. Let's estimate the total motor inductance at 200uH, magnetized by 40A. The energy stored is 160mJ, then. Let's assume some not-over-the-top electrolytic capacitance, like 3000uF (still small fraction of the BOM cost, surface area and volume of the total product; I hope. This also satisfies Celectrolytic > 3...4*Cceramic for good damping). Finally, we can calculate the voltage rise from demagnetizing the motor as:
V = sqrt(E / (0.5C)) = 10.3V
So the discharge peak would be limited to nominal max Vdd + 10.3V = 35.5V. So, this does a bit better than the TVS example (around 40V), but not by much.
Now, the numbers are somewhat pulled from thin air. I didn't go back to modify them to prove any point, and happened to end at a boring "they are almost the same" result, but I think this isn't too boring after all because it shows the interesting nature of the design: "it depends". Had I chosen much higher current, or limited the size of capacitance due to space constraints, using a TVS or an active crowbar would become mandatory.
I agree that a TVS probably results in a smaller and cheaper solution overall, especially at very high motor currents; then it would double as the input hot plug inrush limiter as well, although for that purpose, it's not as effective as a large enough elcap bank, but it doesn't matter when the components are dimensioned to survive the clamping voltage anyway.
But, in any case, it helps to understand all the complexities of designing such a circuit - this may explain why, in some cases, someone has seen a real benefit adding a retrofit elcap on an cheap ESC. Maybe it had stopped them blowing up, for instance, during inrush or during motor inductance demagnetization. Maybe a TVS would have been an even better choice, or maybe not. But, if you go and state that people who say that elcaps are used to "suppress transients" are wrong and stupid, it's you who are wrong, because that's exactly what they can be - and are actually used for!
So, to be actually helpful, you'd do better if you could show that maybe the value of electrolytic used cannot provide enough transient suppression, or there is something better. Or, you would do best if you showed how complex the matter is; it's something even an experienced designer would need to carefully calculate, simulate and test, and not something that should be left for the end-user of the RC modules to fix as an afterthought. Finally, the use cases of the ESCs vary. Not all are always wired up using very short wires, and the battery isn't always just centimeters away - sometimes there can be significant inductance which will require damping.
Sorry if I appeared condescending. I think we both know what we are talking about.
Home
Help
Search
Login
Register
