Repairing Switched Mode Power Supplies (SMPS)

Thread Starter

GS3

Joined Sep 21, 2007
408
Very often when I attempt to repair electronic devices, either mine or from friends, I find out the problem is with the power supply which these days is most often a Switched Mode Power Supply (SMPS). Most of them are flyback, regulated, SMPS (but not all!). In this post I will be discussing the repair of flyback SMPS and a basic understanding of how flyback power supplies work is assumed.

In my experience, most of the time the output electrolytic capacitors failed and if you are lucky this is all that needs to be replaced but sometimes this has lead to other components also failing.

Electrolytic output capacitors fail a lot and SMPS designers need to catch up with this problem. These capacitors are being subjected to high currents of high frequency and are stressed beyond their designed capabilities. Not to mention that there are cheap, bad quality, caps being used out there. This has lead to the so-called capacitor plague.

A very useful tool in checking and diagnosing electrolytic capacitors is an ESR meter. This tool allows for a quick check of all electrolytic capacitors without removing them from the circuit. I built myself an ESR meter based on this article which is itself a practical building of a design published by the Italian magazine Nuova Elettronica N212 as kit 1518. I made some modifications (which I think are improvements) to that design and the device has been very useful to me in repairing power supplies. An ESR meter does not measure capacitance but ESR (duh!) which is a very good indication of a capacitor's health and well-being. I recommend this tool and understanding how it works to anyone who will be repairing power supplies. Besides this one there are plenty of designs out there which can be easily and cheaply built by the amateur. One thing I like about this one is that it signals if the cap is shorted. This is very important because a shorted cap looks ideal with regards to ESR. The building of such an ESR meter would be a worthwhile project for anyone who does not own one.

So, when confronted with a non-working SMPS the first step is checking all output capacitors and replacing any defective ones. Replace them with good quality, low ESR capacitors. As the hobbyist may not have access to high quality capacitors ad, especially, to ESR information, my advice is to generally use higher capacity than originally installed which means lower ESR and the cap generally being less stressed. Bigger capacitors will have smaller ESR than smaller capacitors. Two 1000 uF caps in parallel will have lower ESR than one 2000 uF cap.

The (generally schottky) output diodes should also be checked and replaced if necessary. Up to here it is all very straightforward.

At this point, if we are very lucky, the SMPS might work again but it may be that other components were damaged. The first one to look for is the switching element on the primary side of the transformer. Here things can start getting tricky because it will be mounted on a heat sink, it may be difficult or impossible to find a replacement, etc. Other components, including the control IC may be damaged as well. Sometimes the switching element and the control circuit are on the same IC like the STRT-G6153.

If the switching element shorted then the fuse blew and maybe the diodes of the rectifier bridge also blew. These elements are easy to check and replace if necessary. If the switching transistor and/or bridge diodes were blown it makes sense to insert a lightbulb in series with the mains when testing because that limits the current and the damage should anything else still be wrong. I remember in my first attempts to repair blowing one transistor after another...

So sometimes we get to the point where we have checked the output side and the input rectifier bridge and capacitor but the SMPS is still not working due to the switching element and/or other feedback/control parts. Now things can get tricky. We can test all components individually one by one but this can get tiring and may not be possible with all components. Many would have to be removed from the circuit. And the problem is that there is no way to test parts of the power supply because it need the feedback of the regulation. if something, anything, is not working, then with no feedback the power supply will just blow.

So I am considering building a tool which would have certain parts of a power supply and which could be connected to a power supply under test. This tool would provide some of the functions like the feedback and the switching element so that it could temporarily replace those parts of the SMPS being repaired. Maybe even provide manually adjustable duty cycle switching.

The idea is to provide external, protected, DC voltage to the primary side and a switching element which can be adjusted in duty cycle (I can see a 555 coming). Plus a load on the secondary side. The idea is something like this: with voltage on the primary and load on the output side we slowly start switching pulses which start from zero duty cycle very slowly up and see if power is being transferred to the load. Feedback could be used to adjust the duty cycle. An isolation transformer would isolate from the mains and a lightbulb in series would limit the current (and the damage). This tool would be designed for SMPS which have mains voltage input and low voltage output (say anywhere between 5 to 25) which seems to cover most cases. I am working on this idea which is intended to be extremely simple and, hopefully, useful. All comments are welcome.
 

chesart1

Joined Jan 23, 2006
269
So I am considering building a tool which would have certain parts of a power supply and which could be connected to a power supply under test. This tool would provide some of the functions like the feedback and the switching element so that it could temporarily replace those parts of the SMPS being repaired. Maybe even provide manually adjustable duty cycle switching.

The idea is to provide external, protected, DC voltage to the primary side and a switching element which can be adjusted in duty cycle (I can see a 555 coming). Plus a load on the secondary side. The idea is something like this: with voltage on the primary and load on the output side we slowly start switching pulses which start from zero duty cycle very slowly up and see if power is being transferred to the load. Feedback could be used to adjust the duty cycle. An isolation transformer would isolate from the mains and a lightbulb in series would limit the current (and the damage). This tool would be designed for SMPS which have mains voltage input and low voltage output (say anywhere between 5 to 25) which seems to cover most cases. I am working on this idea which is intended to be extremely simple and, hopefully, useful. All comments are welcome.


There is a tool for checking components in a circuit without disconnecting the components. You might want to consider using a Huntron Tracker.

Huntron Power-Off Diagnostics - Enhance Automatic Test Equipment with Component Level Testing

On the switched mode power supply I viewed in the ARRL 2005 Handbook for Radio Amatuers [page 11.30], the elements of the switching circuit are not easily identified. I would have difficulty disconnecting the switching circuit without disconnecting the feedback.

You said and I quote "we slowly start switching pulses which start from zero duty cycle very slowly up and see if power is being transferred to the load." My difficulty is the filtering circuit which converts the pulses back to DC at the output of the power supply. What about the maximum ripple current the capacitors are able to handle? I'm not sure about the effects of varying the duty cycle. If the duty cycle is not within the tolerance of the circuit, then the results may lead you on a wild goose chase.
 

Thread Starter

GS3

Joined Sep 21, 2007
408
There is a tool for checking components in a circuit without disconnecting the components. You might want to consider using a Huntron Tracker. Huntron Power-Off Diagnostics - Enhance Automatic Test Equipment with Component Level Testing
For my purposes that is overpriced and overkill. I am looking for something cheaper, more limited and more focused. Something I can build cheaply and easily. Like the ESR meter I built which was extremely cheap and has proved very useful. (Maybe I should present that as a separate project.)
On the switched mode power supply I viewed in the ARRL 2005 Handbook for Radio Amatuers [page 11.30], the elements of the switching circuit are not easily identified. I would have difficulty disconnecting the switching circuit without disconnecting the feedback.
I have not seen that circuit but most (if not all) the SMPS I have seen use a transistor (bipolar or FET) as a switching element. This is *easily* identifiable. On all but the tiniest SMPS it will be on a heat-sink and, in any case, it is connected to the transformer primary. When it fails it can either be (1) shorted or (2) open. If it is shorted then you remove it. Easy enough. And if it is open you can leave it in place as it is as if it were not there. It is really pretty simple. Once you remove or disable the switching element the feedback and control circuits which control it become irrelevant as they control nothing.

My idea is to provide temporary external switching for testing before doing a permanent job.
You said and I quote "we slowly start switching pulses which start from zero duty cycle very slowly up and see if power is being transferred to the load." My difficulty is the filtering circuit which converts the pulses back to DC at the output of the power supply. What about the maximum ripple current the capacitors are able to handle? I'm not sure about the effects of varying the duty cycle. If the duty cycle is not within the tolerance of the circuit, then the results may lead you on a wild goose chase.
You are missing something. Either you did not understand well my idea or you do not understand well how flyback SMPS work. What you question is a non-existant problem. This is a good moment to explain the basics as others may also benefit.

A flyback SMPS works by storing magnetic energy in the transformer during part of the cycle and transferring it to the secondary during another part. (This is in contraposition to direct mode PS where the energy is transferred directly from primary to secondary.)

Suppose I energise the primary during 2us. Magnetic energy builds up during that time in the core and when the primary is shut of there is a certain amount of energy stored there which is transferred to the secondary. Each pulse of equal width transfers equal amount of energy "W".

If I give one pulse per second I am transferring W/1 watts. If I give 1000 pulses per second I am transferring 1000 * W Watts. By regulating the energy of each pulse and the frequency of the pulses I am controlling the amount of energy transferred. By placing a suitable load on the secondary I am ensuring that the energy is disipated and voltages do not exceed dangerous levels.

So, to repeat and address your questions one by one:
My difficulty is the filtering circuit which converts the pulses back to DC at the output of the power supply. What about the maximum ripple current the capacitors are able to handle?
I am not sure I understand the problem. As I said, we start the pulses short and spaced so we are at the lowest limit of energy transmission. The output capacitors and diodes have already been checked and are capable of handling way more.
I'm not sure about the effects of varying the duty cycle. If the duty cycle is not within the tolerance of the circuit, then the results may lead you on a wild goose chase.
No. A circuit will have a maximum power it can safely transfer but there is no lower limit. You just start from zero upwards to a limit which is way lower than the maximum the circuit can handle.

Let us talk about an imaginary, completely made-up, example. But before we go any further we all need to understand that a flyback SMPS transfers energy in such a way that the input and output voltages are not directly related in any direct way by the ratio of turns in the transformer. This is essential to understand. In a flyback converter

Vout=Vin x (n2/n1) x (Ton x f) x (1/(1-(Ton x f)))
where:
n2 = secondary turns on T1
n1 = primary turns on T1
Ton = conduction time of Q1
(reference link). So, we see that the output voltage depends on the ratio of turns but also on the conduction times. This is notably different from a direct forward transformer.


My Imaginary and Simplified (tm) SMPS (100% efficiency) has 300 V on the primary side and outputs 15 V at 2 A = 30 W. The ratio of turns is 20:1 (just to make it easy but it could be something else and the output voltage would still be the same). At full load the duty cycle looks like this:

Frequency: 40 Khz ~ T= 25 us
Ton: 40% (10 us) the primary is ON and current is building up. At the end the magnetic energy stored is 750 uJ.
As the primary is switched OFF the secondary starts conducting and for 10 us it outputs 5 A at 15 V which is equal (what a coincidence!) to the 750 uJ which was stored.
Then there are 5 us of dead time before the next cycle begins.

Each cycle transfers 750 uJ which repeated 40K times/s = 30 W.

Normally with lesser loads the switching frequency will be the same but the Ton duty cycle will be shorter as well as the secondary conduction time. At 15 V, 2 A the load can be a 7.5 ohm resistor.

Ok, now my Imaginary and Simplified (tm) SMPS has stopped working. I have checked the output diodes and capacitors and replaced any that were defective but the switching transistor is open and there are no pulses on the primary of the transformer. (Right now I have a SMPS for repair which is in this situation: link)

Keeping the 7.5 ohm load I can place an external switching element which I can control manually. I can start at a constant 40 Khz but starting from Ton=0 slowly upwards and, if everything is well, I should observe the voltage on the load slowly rise. I could even start with the pulses more spaced out, say 20 or 10 or 5 Khz and the voltage on the load would just be proportionally lower. There is no problem in doing this at all. You can always transfer less energy than the maximum the SMPS was designed for. What you cannot do is transfer more.

I hope this helps anyone who is interested in learning about the workings of SMPS.
 

espire

Joined Dec 18, 2007
24
Hi TS,

I had recently started working on smps too.
I had read quite a number of articles on controlling smps.
One of the method that bring my attention is the one cycle control method, whereby only 2 inputs are required, the output voltage feedback and input current sensing. The output pwm is able to maintain the output voltage and also ensure almost unity power factor. Do you happen to have any experiences in this, I would like to know more about it. Tks
 

SgtWookie

Joined Jul 17, 2007
22,230
Have you heard of an Octopus? It's a handy diagnostic tool that you can use in conjunction with a display device, preferably an O-scope that has X and Y inputs.

It consists of a center tapped 6.3v transformer, a few resistors, a test lead and connections to your O-scope. Very simple and useful. It's a predecessor to the Huntron; it dates back to 1935.

Here's a useful page:
http://octopus.freeyellow.com/octopus.html
Links to more useful Octopus reference pages:
http://octopus.freeyellow.com/octopus2.html

Note that with cheap transformers, you'll experience saturation of the transformer core and resulting distortion of the waveforms. One way you might get around that is to simply use two 6.3v transformers with the primaries in series, and only use the output from one of them.
 

mladen82

Joined Mar 2, 2009
3
Hi to all! i am new here
i found this thread interesting as i am wondering about SMPS failures,
can too high line voltage (240-320V) produce failure in secondary side, let say output electrolytic capacitor failure like plague? with no eefcts on primary side
 

pravardhan

Joined Jul 30, 2009
10
Hi GS3,

Very Good Thread...! :cool:

I wanted an e-book on how SMPS works with some practical examples and also an e-book on how to repair an SMPS. :D

Bye,
Pravardhan
 

Nelson2001

Joined Nov 10, 2011
27
P1060551.JPG Can You help me I want to lean about high frequency transformers in psu, but I was unable to find the datasheets with the windings. this is my psu. thank tou in advance. the transformers are tm8350-erl35, 8350 ee16dx, 8350ee19dx.
 

korchoi

Joined Jun 5, 2015
59
That's actually a pretty good idea, to test a power supply with temporary replacement hardware, just like Some embedded systems can have external hardware fake some real features, for testing.
JTAG for power supplies, soon in your next 1000$ PSU.
 

ian field

Joined Oct 27, 2012
6,536
That's actually a pretty good idea, to test a power supply with temporary replacement hardware, just like Some embedded systems can have external hardware fake some real features, for testing.
JTAG for power supplies, soon in your next 1000$ PSU.
The old trick of the trade when I serviced the likes of TVs monitors etc, was to take out the mains fuse and wire in a 60W lightbulb.

In most cases it would limit the current so the fault didn't make it go bang, with at least some power being converted it was possible to observe symptoms and take measurements.

It doesn't always work so well - the PTC characteristic of the bulb filament means it has a low cold resistance, that causes a turn on surge. Some TVs could blow unless you used the weaker 40W bulb.
 
Top