Question about capacitor ripple current rating on input capacitors

Thread Starter

crs831

Joined Sep 18, 2013
7
Hello,

I am trying to understand the effectiveness of filter capacitors on a 125A, 50V brushless DC motor controller circuit. Are they even doing anything?

The circuit contains two 470UF/63V electrolytic capacitors at the power input connector. The incoming power is 58.8V DC. For clarity, let's call these nodes BATT+ and GND. The three n-channel bridges are connected up to the same BATT+ and GND. There is about 3V pk-pk ripple on BAT+ at the motor switching frequency while the motors are turning. It's the capacitor's job to hold the 58.8V node steady during the sag, but it's obviously powerless against the size of the load. Therefore are the capacitors necessary? (see picture attached for glimpse at the schematic)

The problem is these capacitors are blowing up but the reason is unknown. Could it be the ripple current is exceeded or because the ambient temperature is high (70-80C ++?)

Any help understanding intuitively what these capacitors are good for would be appreciated.

Thanks, Roy
 

Attachments

Thread Starter

crs831

Joined Sep 18, 2013
7
Thank Bertus.

I'm trying to understand the purpose of these capacitors a little better. Are they doing any filtering? Or is the load so strong that these caps are essentially worthless?

Thanks,
Roy
 

ebp

Joined Feb 8, 2018
2,332
470 µF 63 V capacitors will have reasonable but not especially high ripple current rating.. Any decent caps will be well specified. 80°C is no longer a difficult requirement to meet, but lifetime must be considered. Most good quality caps will be rated for operation at 105°C, though there are some available rated at higher temperature. Lifetime and temperature ratings are a bit of an exercise in specsmanship since they trade off against each other. The larger the capacitor the higher ripple current rating and the longer the life, within a series. At 470 µF and 63 V it shouldn't be too hard to find a rating of 8 to 10 thousand hours at 105°C at full rated ripple current.

Ripple current rating is more or less inversely proportional to equivalent series resistance (ESR). Ripple voltage due to ESR will appear as step change in voltage with switching of the drive FETs, whereas ripple due to capacitance will be sloped.

The very fact the caps are blowing up says that they are necessary and insufficient. Depending on the wiring to the supply, the capacitors may be supplying a very substantial amount of each current pulse, with the resistance and inductance of the wiring acting to impede current from the supply. Without the capacitors the ripple voltage at the controller terminals might well be tens of volts of fast spikes and several volts of slower change. Without the caps the current through the connecting wires would be chopped and thus generating very large amounts of EMI and RFI.

Having said the caps are necessary, it may be that the system will function without them, however it must be remembered that the voltage drop across the supply wires is voltage that can't be applied to the motors. Ripple could be measured without the caps in place, but this is somewhat perilous because of voltage spikes due to inductance, as previously mentioned. Power loss in the cables to the supply is proportional to the RMS current. Unvarying 125 A is 125 A average and 125 A RMS. 250 A at 50% duty cycle is 125 A average but 177 A RMS, so the power loss due to cable resistance would be twice as high as for 125 A RMS.

Film capacitors may be adequate/better, but even then care is required to select appropriate types. Because film capacitors generally perform quite well at high frequency there is some risk of creating a resonant circuit with the connecting wire inductance and producing really huge amplitude ringing. "Loss" must be introduced into the circuit to damp the ringing and that isn't especially easy at high current.

Without doing the calc's, I would probably try at least ten times the capacitance currently in use. It is often easier to get good performance from multiple capacitors in parallel than a single large capacitor.
 

Thread Starter

crs831

Joined Sep 18, 2013
7
epb, thank you very much for the reply. I am trying to understand conceptually what's going on, so am I correct to think the following:

"ESC capacitors are used to hold up the voltage rail during switching, else, the voltage drop during switching, caused by leads and connectors, is voltage that can't be delivered to the motors for that brief moment in time? In other words, if we simplify this down to V=IR we have less voltage drop across the coil windings and therefore less current that can be delivered (not to mention the excess EMI and RFI)"

If so, does the same concept apply to filter bypass caps on ICs? I thought bypass caps were only used to keep HF from entering the IC. I never thought of them as needing to hold up the voltage rail during fast voltage dips and spikes (caused by a variety of analog things).
 

MisterBill2

Joined Jan 23, 2018
6,019
EBP was certainly right in that last post. Capacitor blow up is due to heat, and the heat comes from charge-discharge current flowing through the equivalent series resistance. If there is still that 3 volts of ripple then either the supply impedance needs to be reduced or the capacitors need to be able to handle the current better. An interesting possibility for effectively reducing the supply impedance is a series choke with a diode back to battery negative so that the energy stored in the magnetic field can be delivered to the load during those high current pulses. This is just like a switching power supply, and so at that current the inductors will not be small or light. And the diode must be high current and very fast.
 

MisterBill2

Joined Jan 23, 2018
6,019
Quite possibly, unless you had something on hand that could serve as a suitable inductor, such as the low voltage secondary of a large transformer. If the design is for a production item then it may not be a good choice, but for a one-off application it can be an option. And a choke inductor failure is not as messy as a capacitor explosion, so that is another consideration. The diode portion would not be huge because the diode will only be conducting during that brief spike interval.
The inductor may indeed cost more than capacitors, but I have not calculated the value of inductance needed to store enough energy to prevent the voltage drop that you are seeing. The size and cost will depend on the wire size and the needed energy to be stored. It may be that a combination, with capacitors as well, is the solution.

It may also turn out that some mechanism of control to prevent the current spike at the motor switching frequency. It may be that you are experiencing a momentary condition of both top and bottom of one ohase being on at the same time. That is a real and known hazard. The solution is to have a short time delay after switch off before the next phase switches on, often referred to as a deadband. It only needs to be long enough to allow the power devices to completely switch off.
Sorry that I didn't think of this sooner, but I was thinking of a brush motor.
 
Top