What performance factors matter for an electronic load?

[jbb]

So I am thinking about developing a high performance electronic load, but I am not sure what performance factors really matter.

My experience with purchase electronic loads (admittedly they weren't high quality ones) includes a few pitfalls:

• Often they are oversized (well, that's our fault for buying an oversized one :-))
• Their minimum operating voltage (a.k.a. compliance voltage) can be higher than I'd like
• Their ramp up from 0 mA to some mA can have big ugly transients
• Limited dynamic range. A lot of modern electronics can have uA sleep currents and >1A peak currents to run radios etc.
• If the input voltage is zero and the load is therefore lower than the setting, the internal control loops wind up. When the input voltage comes back, the electronic load then sucks down a huge inrush of power
• Tendency to oscillate and/or limited bandwidth
• Tendency to have current spikes and dips when input voltage changes suddenly[\li]
  

In addition to this, a lot of hobbyist designs seem to have pitfalls:

• No reverse polarity protection
• No overload protection
• Choice of parts which have widely varying offsets etc.
• Inability to go down to zero current in all conditions (e.g. depends on your luck with opamp offset voltage)

I want to focus on making a really good core current sink by addressing these problems (and not spending huge piles of money either). But these may not be the only problems - what have I missed?

------

I'm thinking about a small load to reflect what I would like: say 0-2A, 0-15V, 10W maximum.  As an extra credit undertaking, I'm also thinking about ways to get a dual range in there, e.g. 0-20mA and 0-2A, with seamless switchover.

[larsdenmark]

If you haven't already watched Scullcom's electronic load project now is the right time:

The series is long, but covers most of the things one would consider relevant.

More things to consider on top of your list:

• Constant current, constant resistance and constant power should be supported
• Should external control be supported for setting the parameters?
• For testing batteries a logging feature is necessary. People seems to want this on an SD-card.
• What happens if you connect 100 V to the unit when it is not powered?
• How large/fast changes do you expect the source will vary? How fast must your load respond to that?
• What accuracy do you need?

[nctnico]

Sounds more like a source measuring unit with current sinking abilities.