Next, we are using the 8874 a bit off-book. We are not directly controlling a motor in most cases, we are using it to provide power to things on a bus and sending a signal as well.
Specifically, this is for the model railroading industry where a "DCC" Signal is generated by flipping the direction pin at around 8kHz so that on each output, one side of the H-Bridge is positive relative to the other for half a cycle and then they switch. The PWM pins are held at source voltages, which is about 12-18V. So we generate a bipolar pulse train that would look like a square wave on a scope. This signal goes to the track that locomotives and accessories connected to.
Locomotives ride on the track. A rectifier takes what it sees as a square wave and converts it to DC that is used to power the electronics and another H-Bridge that sends the PWM that controls the motor. The varying pulses of 58uS and 116uS are stripped off and read as data. Accessories like "boosters" and other controllers connect to the track also. So a simple way to describe this would just be that we are making a 4-8kHz square wave at the motor outputs and connecting several opto-isolators and full bridge rectifiers to it. Some of these devices have capacitance that adds up along the track. The inrush current is a big problem for us in certain situations. The 8874 may only need to supply 2 Amps, but it may be going to 5A or a bit more for just a few cycles. We don't have an easy test case other than connecting a bridge rectifier to the track and charging a 1000uF capacitor through it. We connect it to a 5A or 8A switching power supply brick.
What seems to be happening is that we get an inrush of current, in the 1000uF test case, we could calculate it, but don't have an easy way to measure it. So some overcurrent occurs and the 8874 gives a fault condition in only 3 to 6 microseconds. Then 2ms later it tries to reset, and the inrush happens again and so on. There isn't enough time for the capacitance to charge with our pulses being 58uS and the 8864 cutting out.
Any help would be appreciated. Is there anything you can do with configuration of the chip? Another chip we can use? A new feature in the next version for us ;) Some cheap external solution? For users with the issue, we have had success with a toroid at the output to slow the inrush current, but that isn't ideal. Though we could use help determining what inductance we need. We also created a high tech solution where we we use our microcontroller to sense the fault, halt what we are doing and send PWM at a certain frequency and duty cycle to the brake pin to charge the capacitance, then switch back to normal mode. Thank you
Fred
