Building a 80V power supply surge generator

[ivonenand]

Hi Guys,
I'm a student at a faculty of electrical engineering and recently I got fascinated with avionics hardware. One of the standards involved is the DO-160G which gives the mechanical and electrical requirements. The longest section is about power supply and I got really interested in that. For a 28 Vdc system, the electronics needs to handle 80V spikes. I have already build a power supply input stage which is capable of handling such surges (basically a mosfet voltage limiter), but I have no way of testing if it actually works. Here's where I need your input.

So the standard specifies the power supply waveform in the attachment. The device needs to handle 80V for 100ms, then 48V for 1sec and then back to 28Vdc normal (see attachment). The rise time needs to be < 1ms and fall time < 5ms. The suggested test circuit is in the attachment. The standard also says that pull down resistors might be added to meet the 5ms fall time requirement. The thing is, that this surge generator is quite boring and not very adjustable. You need three power supplies and more or less you need three adjustable power supplies. For instance, in some categories you only need to test up to 60V surges.

I was thinking that it would be a better idea to build like a power B-class type amplifier, but for DC. That way I could generate the waveform with an arbitrary waveform generator and have the amplifier amplify the amplitude by 10x. I did a few quick simulations in LT-spice and the problem I'm getting is that the B class amplifier overshoots by a lot when a capacitive load is added to the output. Of course, the DUT will always have some capacitance on it's power supply input. Does anyone have any experiences in this field, or maybe a different suggestion on how to build such a surge generator? Any input is more then welcome.

Regards
Ivo

[Niklas]

I have used a circuit similar to the one shown in picture 2 for load dump testing (200V, 1 ohm) for an automotive sensor. A relay was used to switch the surge voltage but I have seen a test circuit from Linear (DC1950A) that used MOSFETs for waveshaping.

Have you checked the standard for series impedance of the surge voltage? All the automotive test pulses are given with voltage and series resistance. That impedance might dampen the ringing at the amplifier output.

[T3sl4co1l]

Regarding the class B amplifier: overshoot is a sign of poor compensation.  If it's the typical discrete amplifier circuit, then for the Miller compensation capacitor, try a resistor in series with the capacitor.  Adjust both values until the step response is as desired for both resistive and capacitive load conditions.

Tim

[ivonenand]

Hi Guys,
Sorry I'm late. Thanks for the input. Unfortunately there is no output impedance specified; the voltage in the graph must be measured at the DUT inputs. Anyways, at the moment I'm thinking of two ways to implement the generator:

1. B-Class amplifier, and of course get the compensation right
2. A Buck Converter

I'm thinking that a Buck converter might also be used. I have done digital power supplies before (with STM32F334 and its high-res PWM timers). The advantage of a step down converter is, that it can be used both ways, to supply current to the load, or take the current and dump it back to the power supply (in this mode it acts as a boost converter). Would this be a good approach? The problem I see is that the converter would have to be able to handle high transient currents (to achieve the 1ms rise time) and that the capacitor at the buck converter's output slows down the voltage slew rate. Any thoughts on this idea?

Regards,
Ivo