Continue to Site

Welcome to our site!

Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

  • Welcome to our site! Electro Tech is an online community (with over 170,000 members) who enjoy talking about and building electronic circuits, projects and gadgets. To participate you need to register. Registration is free. Click here to register now.

Question about SMPS fault? Real Baffler!

Status
Not open for further replies.

bigal_scorpio

Active Member
Hi to all,

I just found a strange fault with a batch of SMPS for laptop computers that we had delivered!

We ordered 20 of them, they are the type that has switchable 12v to 24v output at 3.75A(24v) and 2 USB power outputs. They are model number SC-9049LED.

The problem is that they will unerringly power up most laptops without any problem, windows will load perfectly and any program that you care to use will function OK. But!!!!!!!! The mouse will not work correctly, it will not move about the screen as it should and instead jerks and sticks randomly so badly that it is unusable.

This effects the touchpad, ministick pointer or indeed external USB mouse, infact anything that moves the mouse pointer is affected! On ANY laptop!

How can a PSU be stable enough to power up the laptop perfectly every time yet fail to enable the mouse to move correctly?

This problem is definately the PSU causing it and they are all the same from this batch but if one from another batch or the laptops original supply is used the mouse will work OK! Anyone got the answer to this puzzle?

BTW I have had the scope on the ouput of a few of these and there are no spikes or dips apparent whatsoever. A pure straight trace within 1% of the set voltage.

Bet I have you all puzzled with this one! ;)

Regards.........Al
 
Sounds like some EMI is getting into the computers. The output oscilloscope trace is likely not as pure as you think. Look at the output with the scope set to AC input and high sensitivity (10mV/div) at about 1µs/div sweep speed. Adjust the sync control until you see the switching spikes. Compare a bad unit with a good unit to see if there is any difference.

Switching power supplies (which this is) can put out a lot of noise if not properly designed and filtered. Perhaps that batch of power supplies has high EMI due to a missing filter or bad design.
 
What's the scope bandwidth? Sometimes the EMI is up in the hundreds of MHz or GHz range. And it may not show up on the output at all, but rather somewhere in the PSU itself like a circuit trace carrying high di/dt or dv/dt signals. Are these PSU's in plastic casing? (I assume they are like most laptop PSU's and are plastic cased, double insulated?)
 
Last edited:
What's the scope bandwidth? Sometimes the EMI is up in the hundreds of MHz or GHz range. And it may not show up on the output at all, but rather somewhere in the PSU itself like a circuit trace carrying high di/dt or dv/dt signals. Are these PSU's in plastic casing? (I assume they are like most laptop PSU's and are plastic cased, double insulated?)

Hi,

The scope bandwidth is (AC) 3 to 10mh.

The PSU is in a plastic case screened internally on 3 sides by a screen/heatsink but not below the board or either end, the board is also single sided which seems a bit unusual as most today seem to be double and incorporate an earth plane.

I also noticed the filter cap on the output is only 1000uF which seems fairly small for a precision PSU. I have added another 1000uF and will try it on some laptops tomorrow.

Still I think that its odd to be able to load the whole of windows and other progs error free, I would expect at least the odd glitch program wise if interference was the cause. I have been working with computers since their humble ancestors ran in 16kB and came as kits but they never cease to surprise me! ;)

Al
 
Wrap a whole bunch of tinfoil all the way around the plastic-encased PSU. I mean just wrap it over and over and over. If you think you've done enough, double that amount. Make it so the only thing going in is wall power, only thing coming out is laptop power, and everything is tightly wrapped in foil with no gaps.

Then test. If you still have problems, it is not radiated EMI and that's one thing we take off the list of possible culprits.

The PSU is in a plastic case screened internally on 3 sides by a screen/heatsink but not below the board or either end, the board is also single sided which seems a bit unusual as most today seem to be double and incorporate an earth plane.
I took apart an electronic neon sign power supply that was exactly like that. The two output FET's were on pieces of metal that were vertical for the connection to the FET's and then bent back over and covered the rest of the board area. Everything single sided through hole. I guess it's the cheaper way of doing it still, though I can't imagine how much slower through hole stuff is to populate compared to a pick and place machine.
 
Last edited:
Update on PSU

Hi to all,

I tried the PSU modified with double the smoothing - still the same.

Then thied Speakerguys idea about EMI and covered with a full roll of foil - still the same again. I'm stumped.

What could I do to remove UHF frequency spikes that may be there but not showing on the scope? Any ideas? :confused:
 
Since it's not radiated EMI from the PSU itself, it's noise radiating from the cable OR noise on the line getting into the PC.

They make clamp-on ferrites for EMI and RFI suppression that can go on power cables. They have some effectiveness but are not great. These are like the thick cylinders that are molded in to VGA cables, power cords etc but they just clamp onto the cable. They have some effectiveness and are worth a try. They will introduce some impedance to higher frequencies from the switch mode PSU. One place that makes them is:

https://www.fair-rite.com/newfair/index.htm

Put one right at the PSU on both the input and output sides. If it's not a PFC unit it can also be throwing noise back onto the input power line. If it's got power factor correction that won't be an issue.

I'll try and think of some more ideas.
 
Last edited:
If a clamp-on ferrite over the whole DC cable doesn't help, then try splitting the cable and clamping another ferrite to one of the lines.

I just did a search of a web site that sells this supply. The letters 'fcc', 'ul', 'csa', 'ce', 'listed', or 'approved' don't appear anywhere... I did a Google search of 'SC-9049LED fcc' and those two never appear together in any web site anywhere in the world. Good luck.
 
Last edited:
If it's not a PFC unit it can also be throwing noise back onto the input power line. If it's got power factor correction that won't be an issue.
Since the device is rated at 150VA in, 80% efficient, and 90W out; it's safe to say it's not PFC.
 
Yeah,with no PFC it could be throwing a metric crap-ton of EM back on the power line. Crap-ton being one of the EM compliance measurements :)
 
Last edited:
Hi to all,

Thanks for all the ideas so far. :)

These devices are made to such a high spec that the manufacturers don't even put a makers name on them! ;)

I don't think anyone wants to admit to them, but saying that the other batches of these have all been OK.

I will try the ferrites anyway and I was just wondering if it's worth putting a 100n cap across the output as I can't see any small capacitance across it at the moment, just the 1000uF.

Keep up the ideas guys...........Al
 
The 555 won't double the voltage, you'll loose 0.7V per diode and 0.6V in the output stage of the 555. If the input voltage is 9V, then the output will be about 16V unloaded and drop a couple of voltage when fully loaded.
 
The 555 won't double the voltage, you'll loose 0.7V per diode and 0.6V in the output stage of the 555. If the input voltage is 9V, then the output will be about 16V unloaded and drop a couple of voltage when fully loaded.

hi hero,
I think that CERN accelerator has mis directed your post.:D
 
The 555 won't double the voltage, you'll loose 0.7V per diode and 0.6V in the output stage of the 555. If the input voltage is 9V, then the output will be about 16V unloaded and drop a couple of voltage when fully loaded.

Eh? :confused:

Al
 
CERN Again. LHC Post Interrupter.

hi hero,
I think that CERN accelerator has mis directed your post.:D



Yup I think so, :p they must have accidentally bumped it again today. Maybe it was the cleaning people with the broom handle to the on Button. :D


kv
 
Last edited:
Status
Not open for further replies.

Latest threads

New Articles From Microcontroller Tips

Back
Top