0
\$\begingroup\$

I have had issues in the past with using too close of an input power supply to the max rating of my main power regulator causing failures. For example, I used a 24VDC supply with a regulator with an absolute max rating of 28VDC. When this PCB was used with a PSU with a long cord, there would apparently be an inductive power spike upon turn on, that would destroy the regulator. The PCB had a 0.1uF ceramic cap on the input line parallel with a 47uF electrolytic cap. The chip also had a 0.1uF ceramic on each input pin. So obviously this inductive spike was a long event, as my capacitance should work for short spikes.

I now am working on another project that I don't have any options to go with a higher rated regulator. I have a 15VDC input supply with a chip with a max rating of 24VDC, recommended 20VDC. This system will be using a long (~6') cord.

What are my options to condition the input supply? A bad solution that I thought of, was to buffer with a linear regulator before this chip, that would only drop a small amount of voltage, but would be rated higher..

Do I just need more bulk capacitance? Something like 100uF+? The switching regulator will be going from 15VDC to 17VDC at up to 15A. Switcher: MP3428AGL-Z

EDIT: Some clarification:

  • Power cord length is from the PSU -> PCB
  • On old PCB, when the PSU was plugged into the PCB and then the PSU was turned on, the regulator would blow. When the PSU was turned on, then plugged into PCB, it would be fine. So the inductive spike was from the initial inrush current.
  • New PCB has different PSU, much higher quality, so it shouldn't be an issue, but I want to plan for the worst case, as this is an important design.
\$\endgroup\$
2
  • \$\begingroup\$ More likely the regulator was not stable enough on startup with load caps and created overshoot.. If it has soft-start by design, then the slew rate can reduce the disturbance to overshoot. OVP is necessary. \$\endgroup\$ May 6, 2019 at 21:46
  • \$\begingroup\$ @SunnyskyguyEE75 So when I used the board with my benchtop PSU it worked perfectly fine. Only with a laptop-style sealed PSU with a long cord there was any issues \$\endgroup\$ May 6, 2019 at 22:56

2 Answers 2

0
\$\begingroup\$

There are lots of possibilities, and it really depends on how much energy you expect to come down that cable, which in turn depends on what is in your environment, and also maybe what is driving it. (If the mains supply is just an unregulated transformer + bridge + reservoir cap there could be spikes coming from the mains itself).

One possibility is a transzorb which can simply clamp spikes coming down the line - it's basically like a fast zener (zeners have too soft a curve and are usually too slow for this kind of function).

But the problem is knowing how much energy is in those spikes and how to absorb it. A small amount of series resistance along with some C/L filtering can also reduce the amount of energy hitting your circuit and damp the ringing. (there are various commercially available filter modules as well.)

Is the issue happening when the circuit is plugged? or are you hot plugging the cable and getting spikes as a result of that?

Without knowing more specifics like the amount of DC current you draw and having some rough idea of the source energy it is not really possible to be more specific. This is EMC design, and there tends to be a fair amount of trial and error in getting it right. But this can give you a start point.

\$\endgroup\$
4
  • \$\begingroup\$ Thanks for that info, see edits about the specific times it would cause the IC to blow. Yeah, I don't have any idea of the energy either. Other than that it is presumably a fast spike since my caps don't help with it. \$\endgroup\$ May 6, 2019 at 20:49
  • \$\begingroup\$ well, if you have a decent DSO can you actually see the spikes coming in, to get a rough idea of amplitude? can you then see if they seem to reduce in amplitude by placing, say, a 1k resistor across the input? in fact, do you know what the source of them is? is there heavy equipment nearby switching a lot of power? \$\endgroup\$
    – danmcb
    May 6, 2019 at 20:53
  • 1
    \$\begingroup\$ it could also be that NEGATIVE spikes are killing the chip. DO you have protection diodes at the input? \$\endgroup\$
    – danmcb
    May 6, 2019 at 20:54
  • \$\begingroup\$ No, I did not. That's a good point. So then, maybe use a bigger cap, in conjunction with a TVS diode that will clamp both high and negative voltages on the input line? \$\endgroup\$ May 6, 2019 at 21:05
0
\$\begingroup\$

I think the best practice is verify the root cause of your failure then correct the cause. Don't guess. Test input start current surge to understand impedance with tower supply using a small sense R or current probe. Low ESR remote load caps with cable inductance can cause startup instability for some regulators on power up.

Usually PTC and Clamps protect for OVP Clamps or NTC inrush limiters for soft cold-start.

If your product is intended to work from these laptop style chargers, test a variety of them for overshoot startup Simulate that load impedance on a variety of long cable laptop supplies for the tendency to overshoot on startup. Some use buck-boost regulators that can reach 45V but regulate to 19.5V.

I used that fact to modify a universal laptop charger into an adjustable V output. It came with different size barrel adapters and a 4 pin header, which I discovered included remote voltage sense to compensate for long cable voltage loss. This meant I could attenuate feedback and raise the voltage. SO I can imagine some impedances that could cause startup overshoot including negative incremental impedances where input current rises if voltage sags during startup.

So the best bet is to verify the potential root cause of failures and then a solution is easy.

  • Test startup steps voltage current response with small R sense.
  • simulate same input capacitance + load
  • then test various external supplies to this remote load to look for overshoot.
    • AC interruption, surge test, DC interruption, AC start-stop cycle test with hot-cold AC start to in case of PTC protection failure
  • don't guess. Simply adding more low ESR Caps may compound the problem or make it better when raising the impedance and clamping may be the best overall.
\$\endgroup\$
11
  • \$\begingroup\$ A PTC does not replace an NTC but might could work with it if the NTC prevents the PTC from tripping and prevents OV on startup. But it is best to verify root cause failures. \$\endgroup\$ May 7, 2019 at 2:00
  • \$\begingroup\$ Thanks for that. I wasn't clear originally I guess. My earlier board which had this problem was fixed by replacing the regulator with one that was rated for a higher voltage. I can't test it anymore, that was from a while back. \$\endgroup\$ May 7, 2019 at 11:06
  • \$\begingroup\$ Now with this new design, since the scenario seems similar to the last one, I am looking for advice on best practices to condition an input supply when it touches a regulator that has an absolute max near the input voltage \$\endgroup\$ May 7, 2019 at 11:07
  • \$\begingroup\$ Since it is dependent on your load it seems, Do as I suggested. Verify power on overshoot. With no load and your new load. If/When you see a difference then perhaps all you need is a some to raise impedance and or OVP. Test/verify \$\endgroup\$ May 7, 2019 at 11:10
  • \$\begingroup\$ So do nothing to address this until the board is fab'd and I can test it? I was thinking of using a tvs diode for ovp and uvp. That should really prevent this from being a problem I assume \$\endgroup\$ May 7, 2019 at 11:45

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.