Close

Design Decisions Philosophy

A project log for Axiom: 100+kW Motor Controller

High Power, High Performance 400V 300A 100+kW Motor Controller fully compatible with VESCĀ®

sonny-lloydSonny Lloyd 08/17/2019 at 03:412 Comments

You can have it good, fast, or cheap. Pick any two. Good and fast, it will be reliable but expensive to design. Fast and cheap, probably going to end up with something that causes fires rather than drives an EV. Cheap and Good is possible but it’s going to take some time to deliver. Obviously there are many design decisions made along the way of producing a final schematic and bill of materials for a high performance full up motor controller. But how are those decisions made? That is the focus of this log post.

And it turns out.. it's not that hard. During the initial project meeting EV Power Designs asks the client what are their top three priorities. Cost, weight, volume, performance, reliability, protection from obsolescence, manufacturability, time to market and so on. OK, I lied. It's hard; because after all don’t we want it all? Pick only three? This simple question tends to bring out a lot of discussion about the client’s expectations, desires, goals and general philosophy. The client gets to hear about how each design objective will influence the actual product so there is an appreciation both ways. Client and consultant get an opportunity to better understand the project but more importantly, each other. From this conversation, the three .. yes only three.. top priorities will eventually be ascertained.

Here is an example of how narrowing down the product development philosophy to three key drivers can help with decision making.  Let's say the three most important goals are Weight / Volume / Cost (typical for new power electronics design where time to market is not a key factor). How does this impact the decision for torque control, arguably the most important function of a motor drive? Depending on the power level there are a lot of different design paths that will lead to lower weight, volume and cost but at all power levels it's possible to reduce the phase current sensor count from three to two. The motor load is still 3-phase, but the current in only two of three phases is measured for control purposes.  The current sensor required for high performance motor drive is relatively expensive because it must be high bandwidth and low error so removing one out of the system saves a lot of cost. And the thing is rather large and heavy because of the internal ferrite core, so weight and volume come down as well. It's a win… win and win. but. As my old professor used to say, you never get a free lunch in engineering. There is a small price to pay to have higher performing op-amps that are now required for adequate design purposes, and with only two sensors the system will never be aware of circulating currents and thus never have the ability to eek out some more performance by nulling out the circulating current. Or in other words… with only two sensors, the system just became lower performance.  Thankfully, in this example, performance wasn’t one of the top three design goals.

We applied this same philosophy to the Axiom control board design process. What three top priority goals should we have? Performance, reliability and versatility. 

Discussions

Gravis wrote 08/17/2019 at 16:41 point

I've found that maximizing modularity to be a worthwhile investment for working on prototypes.  Sometimes specs change mid-project and redeveloping individual modules is easier than a tightly integrated system.  When you are happy with the performance then you should move on integrating and optimizing. As always, your mileage may vary.

  Are you sure? yes | no

Marcos wrote 08/17/2019 at 23:59 point

That depends on who's in charge of the project. If you are the contractor and your specs changed, thats likely an order change that should be billed.

If you're in charge, you can use an Axiom that afaik is only vesc with a decoupled powerstage that lets you cover a 50kw to 200kw power range only changing a few part#. A wider range and voltages can be covered using an adapter board for different modules, which we supply the kicad template for that :)

  Are you sure? yes | no