Active clamp / braking resistor circuit design

Circuit design - concept:

Here's my basic concept - it's how most braking resistor circuits work. Threshold comparator with hysteresis, driving a large FET or IGBT (depending on the voltage). It will self switch according to the load current, braking resistor value and hysteresis level.

Of course, the 431 is only rated for 20V, so wouldn't be happy in this circuit, seeing the full 60V or so. So that requires a lower voltage supply which can be provided by a simple zener / transistor follower. The voltage sense potchain obviously needs to come from the input voltage but for the hysteresis I need a voltage that goes high when the threshold is exceeded. Arguably I could pull the reference down with the power FET instead but we are talking small voltages (2.5V ref and getting that way with the FET on voltage).

Here's the simulator - SIMetrix Elements - which is free with "limited" number of nodes but in practice is pretty good for most stuff:

• 140 analog nodes (internal and external)
• 360 digital nodes
• 720 digital ports
• 300 digital components
• 360 digital outputs

Some real component values:

Here's the schematic itself with some practical values. The servo is represented by a current source of 5A (~300W at 60V) and the bulk capacitance has an initial voltage of 50V. These values result in just over 1V peak to peak ripple and a frequency of around 30Hz at something like 40% duty cycle. Doesn't seem silly.

The braking resistor isn't a silly value at 4.7R and the quiescent currents shouldn't cause anything to overheat. Even with 100 duty cycle the "Q3" regulator device should be under 300mW and the other signal compts are seeing pitiful currents.

The hysteresis resistor is a bit high at 2.2MR but I don't want to reduce the impedance of the reference point and it's not worth the ball ache of splitting the feedback just to lower that resistor value. Besides, many people seem happy to use high values like that in their circuits so I'll join the slobs and have an easy life on this one.

The waveforms correspond to the voltage and current probes in the schematic:

The FET isn't going to be suffering from any significant switching losses, as it's only running at 10s of Hz, unless I've got such a shit drive waveform that it's going into a linear mode. Driving it through a 1k resistor and letting it bleed down through a 1.5k isn't a problem at these frequencies - with a device like the IRF530, it's switching in the 10s of microseconds range and it's a purely resistive load.

These look reasonable, so I suppose the next task is to build the damned thing. Obviously I won't be using the 2N2222, as it's only rated at 40V but that part simply came up in the simulator library as the default NPN transistor and the simulator doesn't seem to be upset by the excessive voltage across it - one potential danger of relying on simulations instead of proper circuit design.