Traco Input Capacitor Requirements

[paulstoffregen]

TL;DR = Please consider adding a suggestion to the Traco product pages (1065 & 1066) recommending a 10uF (or larger) capacitor be used at their input.

Long version...

Many months ago, a customer built an impressive project using an Arduino Due and three Teensy 3.0 boards. It was powered from a 9 volt power supply, which directly fed the Due. To power the Teensy3s, he purchased this Traco step-down converter:

http://www.adafruit.com/products/1066

He experienced strange, very difficult to diagnose power-up problems. Here is the conversation:

http://forum.pjrc.com/threads/23647-Powering-Teensy-3

After much frustration, he purchased a lab bench power supply. The Teensy3s were able to boot up reliably when the Traco received power from the lab bench supply. But when used with the wall wart, the would often not start up.

He sent a Traco to me for testing. I did some initial tests. Recently, I investigated in more detail. It turns out the Traco can enter a state where it becomes an oscillator if the input supply is weak and no capacitor is used at its input. Here is detailed info with waveforms:

http://forum.pjrc.com/threads/23647-Pow ... #post37519

The solution is very simple, just add a 10 uF capacitor at the input.

However, ordinary customers have no way to know they should add this capacitor. The Traco datasheet says a 22 uF capacitor is needed only if the input voltage is higher than 32V. That implies no capacitor would be needed for lower voltage. But indeed at only 9V, the Traco becomes an oscillator if the source has higher output impedance.

So I'd highly recommend adding a mention on those 2 product pages, suggesting a 10uF capacitor should be used at the Traco's input. The capacitor might not be needed in all cases, but when it is needed, the oscillating problem is extremely confusing.

Hopefully this will spare anyone else from so much project power-up frustration.

[adafruit_support_rick]

Wow - whodathunkit? Thanks for the research, Paul!

[adafruit]

Wow, that is annoying! product descriptions updated!

[paulstoffregen]

TL;DR = Please update the text for INPUT, not OUTPUT.

Just took a quick peek. Looks like this text was added:

"We suggest adding an additional 10uF+ capacitance on the output for additional stability"

However, the 10uF capacitor is needed between the input pin and ground, to avoid this problem.

Adding more capacitance at the output, but not the input, is likely to only aggravate this problem. Fundamentally, what I believe is happening involves the Traco draws a pretty substantial amount of current at startup, mainly to charge the capacitance at the output. In the input source is weak, the voltage at the input can drop to the point where the Traco shuts down, then is tries again, causing oscillation you see in those waveforms. With a 7805 linear regulator, the regulator would momentarily operate in its "dropout" region, where the output isn't regulated, but it still delivers as much current as it can, which causes the output capacitance to charge up and reach the correct voltage. But with the Traco, the behavior is not nice and linear like the linear regulator. It temporarily shuts down if the input sags too much.

With an oscilloscope and plenty of time to try things, it's possible to see. But for ordinary hobbyists, these types of problems are nearly impossible to properly understand and diagnose.

[adafruit_support_bill]

Thanks Paul. I'll see that this is corrected.