Can Switched Mode power supply be a problem when powering MCU?

Thread Starter

dandy1

Joined Sep 30, 2017
178
I have built a pcb that uses a (D-Sun) 3 Amp SMPS. Little cheap things found everywhere on ebay etc.
I've used many in the past without incident and they tend to be very good for up to 1.5 Amps. Low quiescent current and very cheap.
without the SMPS connected I am able to upload to the mcu (atmega328pb) via the program header and everything works fine and done
blink tests on random pins.

There is 12V (lead battery) to power the SMPS which goes through a PMOS -ME12P04 (reverse polarity protection) and is shared with a HS switch, and there is no problem with that.

All grounds are shared.

The SMPS (@4.1v) is used to power the MCU and goes through another HS switch for sensors etc.

When i connect the SMPS it appears to short the MCU and gets very hot and the output voltage bounces 400mv
first chip fused with 160 Ohm short second with 2.5.

Connecting the SMPS to a small motor and runs without a problem - steady output on 50mA, even with 7 ohm resistor
only 10-20mV drop.

The only thing different to what i have done in the past is use a different PMOS, but that looks all in spec.

Could it be a problem between the PMOS and SMPS switching speeds? or low current? or ground issues?
I got a few of these SMPS but they all the same batch So i'm asking here if anyone has had similar experience or Knowledge before I start cooking again!

Thanks for looking.
 

dl324

Joined Mar 30, 2015
16,845
Could it be a problem between the PMOS and SMPS switching speeds? or low current? or ground issues?
I got a few of these SMPS but they all the same batch So i'm asking here if anyone has had similar experience or Knowledge before I start cooking again!
Most inexpensive power supplies are switching these days because you don't need cost or weight of a transformer. Because they're transformerless, they might not provide isolation from AC mains.
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
Most inexpensive power supplies are switching these days because you don't need cost or weight of a transformer. Because they're transformerless, they might not provide isolation from AC mains.
There is NO AC - I'm bad enough with DC
 

Papabravo

Joined Feb 24, 2006
21,159
Which chip got hot and burned, the MCU or a chip on your power supply? What bounced 400 mV?? What fused with 160 Ohm short??? What second with 2.5???? This one sentence reads like a stream of consciousness:

You wrote:
When i connect the SMPS it appears to short the MCU and gets very hot and the output voltage bounces 400mv
first chip fused with 160 Ohm short second with 2.5.


It is odd to think that the power supply is shorting the MCU. A schematic would be a big help; it is not your fault that verbal descriptions are not up to the task of communicating the nature of your problem.
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
hi Papabravo,

APOLOGIES; DC-DC converter NOT SMPS


The MCU suffered on both attempts, fusing with 160 and 2.5 ohms respectively between VCC and GND. On the first attempt (which i did not initially notice) and actually managed to set fuses and upload 20K program via ISCP method. But then Touched the MCU and noticed it was very hot. I removed converter and tried power via a VCC pin, but felt it get warm instantly. Checked for shorts and noticed 160 Ohms between VCC and GND.
So I tested converter as noted above and found nothing untoward.
I decided to start again from scratch this time only adding minimal components as required. Added MCU with 2 bypass caps to power rails and a bypass on the Aref Pin. A 10K pull on reset. Connected to ISCP and this time burned a bootloader - everything working as expected. This tells me there is no problem Between VCC and GND.
So next added the reverse polarity FET and the High side switch circuit for 12V input but left off the converter - done a few switches on the high side FET - again no problem.
Added the converter and powered from 12 v battery and now problems back again, Very Hot MCU after a few seconds. Checked between VCC and GND and 2.5 ohms recorded.
Now i removed MCU and this removed the short, so it looks like the converter done the damage,
1600298811792.png
converter.

I have included the schematic but it wont make much sense because im new to this stuff so i dont ask you to comment on that.
 

Attachments

Last edited:

Papabravo

Joined Feb 24, 2006
21,159
It sounds like there are one or more faults on the MCU board.
I still don't know what you mean by fusing. If you mean measuring the resistance between Vcc and GND on a PC board before or after it has been powered I can tell you that it is a completely meaningless number.
 

kaindub

Joined Oct 28, 2019
125
You're schematic is very hard(impossible ) to follow. What are all the Xs? You may know where they connect to but we dont. We're only getting half the story
To guess a your problem. Does your hardware work at all? Have you tested the interfaces separately (without running any program) and seeing if you can get inputs and outputs to work? Just needs a variable voltage supply and a multimeter.
Are you breadboarding this?
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
You're schematic is very hard(impossible ) to follow. What are all the Xs? You may know where they connect to but we dont. We're only getting half the story
To guess a your problem. Does your hardware work at all? Have you tested the interfaces separately (without running any program) and seeing if you can get inputs and outputs to work? Just needs a variable voltage supply and a multimeter.
Are you breadboarding this?
Im sorry i knew it would be difficult, im working on the schematic to tidy things up and label everything

The Xs are just terminals, usually power pins.

Yes I did upload to the board without the converter connected, and tried a few digital writes on random pins to see if things were correct, this was successful and the PMOS high side switches worked (no load) without a problem. I did not bother to connect any other hardware at this point. I then connected the converter and this is where i get the problem.

I will post the layout maybe that will make some sense.

Do you think i should try to connect serial adapter to the board (with the converter removed) Im afraid that it might damage the serial adapter after my earlier cooking lesson.
 

Attachments

Thread Starter

dandy1

Joined Sep 30, 2017
178
It sounds like there are one or more faults on the MCU board.
I still don't know what you mean by fusing. If you mean measuring the resistance between Vcc and GND on a PC board before or after it has been powered I can tell you that it is a completely meaningless number.
OK thank you Papabravo, so if i remove the MCU from the board, can i check it some how with a multimeter? Ive done some continuity checks on a bare board and cant seem to find a fault.
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
Did you adjusted the trimmer on the smps module to the correct output voltage?
Yeah i make sure that is set correctly to 4.1v as it is also used to power a modem. Not sure how these things work on power up, i mean is it possible there is some kind of transient?
 

Papabravo

Joined Feb 24, 2006
21,159
Using a multimeter to check continuity between Vcc and GND on a board with components may or may not provide useful information if it something else besides a dead short, If you suspect the board has a problem you can try using a bench supply with a current limit feature to see if you can get your board to work. To test the SMPS you want to make some dummy loads using resistors with a suitable power rating. 4.1 Volts at 50 ma. is 200 mw. so a 1/4 watt resistor of 82 Ω should work. If both of these experiments work then you have a more insidious problem.
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
Using a multimeter to check continuity between Vcc and GND on a board with components may or may not provide useful information if it something else besides a dead short, If you suspect the board has a problem you can try using a bench supply with a current limit feature to see if you can get your board to work. To test the SMPS you want to make some dummy loads using resistors with a suitable power rating. 4.1 Volts at 50 ma. is 200 mw. so a 1/4 watt resistor of 82 Ω should work. If both of these experiments work then you have a more insidious problem.
I had previously tested the smps and found no fault. Put a 7 ohm load at 4.1v and got a 0.57 amps, with little voltage drop (20 - 30mV). also it draws 200uA with no load which is in spec. And a small motor on the output draw 50mA with no volt drop. What i dont understand is that it works fine powered via serial adapter but putting the converter on screws it up - telling me its the converter thats the problem!

OK thank you Papabravo for your comments, ill do some testing tonight.
 

Beau Schwabe

Joined Nov 7, 2019
155
Does the DC-DC converter have a slow start option? Any caps on your main board (especially low ESR) can look like a dead short when first applying power. If the DC-Dc converter fails it can allow full voltage on the output which can over voltage all of your 5V components. A slow start option can alleviate any initial startup issues.
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
Does the DC-DC converter have a slow start option? Any caps on your main board (especially low ESR)
Yes has soft start with a 50uS blank time then a slow ramp up, actually the switching chip (mp1584) looks a capable piece for a so8 and looks like it has "precise current limiting" - will have to look into that. As for Caps there's nothing discernable few 100nF on power rails and smallish tant on the converter output.
 

Rasterist

Joined Sep 15, 2020
4
It's not clear from your posts if you've independently tested the DC-DC converter. Ideally this would be done with a scope and an electronic load so you can detect ripple and spikes in both static and, especially, dynamic load cases. Have you done any of this? Are you able to?
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
It's not clear from your posts if you've independently tested the DC-DC converter. Ideally this would be done with a scope and an electronic load so you can detect ripple and spikes in both static and, especially, dynamic load cases. Have you done any of this? Are you able to?
No, I don't have a scope, so it looks like I'll bite the bullet and start fresh with new converter and board. And see if that holds up. I've never had a problem with these converters before so maybe it's a one off. Thanks anyway
 

Thread Starter

dandy1

Joined Sep 30, 2017
178
On another note, because I have a shared ground between 12 volts and Vcc due to the converter, is there any point in constructing a star point for the 12 volt devices (such as optocouplers) and run it back to the negative input? At this point im only really using optos as switches rather than isolation so they're a bit pointless really other than you can swap them out easily (via dip socket) if they're damaged
 

MIS42N

Joined Jan 25, 2013
22
You could put a small 5.6V zener in place or as well as the MCU. If it fries, you have a power supply problem. If it doesn't fry (1) there isn't a power supply problem OR (2) there's a small spike that fries the MCU but the zener has clamped it. If it were something I built, I'd be looking at the programming to see if there's a chance to create a short on one of the ports. But that's me.
 
Top