Close

Thermal Evaluation

A project log for TS350 True Sine Inverter

350 Watt inverter that converts 12VDC to 120VAC @ 60Hz.

Brian CornellBrian Cornell 12/01/2018 at 20:030 Comments

After some diversions on other work I'm finally getting around to working on the TS350's redesign again.  I did a poor job on the original thermal model and didn't account for the increase in Rds(on) of the MOSFETs (with temperature) and, more importantly, the coupling of adjacent components.  The result was that the inverter section would overheat (shutdown) at continuous operation much above a 300W output.

One of my enlightenments after some additional book work is the need for thermal imaging to get a true picture of what's going on; my spot IR thermometer & thermocouples don't cut it.  I picked up a FLIR One Pro.  Because it uses a smartphone for display & processing it's half the cost of an entry level professional, self-contained, IR camera.

It's easy to use and with the companion FLIR Tools application enables you to quickly analyze image data.  Here's an image of the TS350 powering a 300W load continuous.  The temperature boxes were added after the image was taken with the Tools appl.

The camera paid for itself on the first use.  Revelations:

My take-away is that SMD heat-sinks are okay when dealing with only a few watts for a single device but for higher powers & multiple devices they are inadequate.  I suspect they can also present a significant near-field capacitive coupling problem with few remediation options other than a shielded enclosure.

So, for the redesign I am creating a linear superposition model that will account for coupling effects.  To do this requires that I know the coupling coefficients and these must be measured from a thermally representative model of the design.  All of this implies that I need to make decisions on the board layout, transistor mounting, heat sinks, etc.

I'm definitely gong with TO220 MOSFETs for the inverter with the goal of a more compact footprint and improved thermal efficiency.  The relay will remain SMD but will probably offload to a separate power board that contains the power connections, user interface, etc., and because of this probably won't need heat sinking.  I haven't decided on the sine section yet.  From an EMI perspective it needs lots of improvement and much of it will revolve around reducing the size of current loops.

Lots to do!

Discussions