Hello,
for a permanent magnet synchronous motor I designed an inverter (based on the TI design
http://www.ti.com/tool/TIDA-00774?jktype=design ), though simplified to a lower power level. It is based upon three halfbridges CSD88584Q5DC (23nC Qdiodereverse) and a smart driver DRV8323 with integrated differential shunt amplifiers.
The idea is, replacing two current sensor by one shunt in the dc link as shown in the attachment. Depending on the switches' state, the measured voltage drop across the shunt is - in theory at least - equal to one of the absolute values of the phase currents. The phase currents change slowly in comparison to the shunt signal, therefore they can be seen as constant during the sampling time.
In my setup though, I have an effect that I cannot explain. For low currents (~1A) my reconstruction method works fine. As the currents are increasing, the reconstructed waveforms become strongly distorted. I guess it's due to the exponential falling / rising of the shunts voltage right after the switching instants. The moments for sampling the shunts voltage is situated almost at the 'end' of each possible time'slot'. If the signal's not yet settled, it results in distortions in my reconstructed phase currents.
I have no idea how to explain that exponential rising/falling of the signal. I cannot reproduce it using LTSpice by adding various parasitic elements to the circuit. Guessing the charge of the exponential area results in several hundred nC - the reverse recovery charge of the halfbridge diodes is 'only' about 20nC.
The recorded waveforms where recorded using DC-coupling at the osci. I get a similiar signal if I use the integrated differential amplifier of the driver (due to the limited slew rate it gets smoothed but the effect is still there).
I'd be glad about hints for solving this puzzling issue. Is it a measuring error?
(similiar version for german readers can be found here
https://www.mikrocontroller.net/topic/486380 - i updated the sketch)
Home
Help
Search
Login
Register
