[EE]: SMPSU repair -- secondary schottky shorts and similar stuff
Philip Pemberton
(Pardon the alliteration in the subject, I think I'm suffering from
excessive solder smoke exposure)
Here's a "tale from the trenches" for you...
I have -- on my bench -- a "Nichimen" branded DVR. It belongs to my
grandfather, who is quite attached to it (and the stuff he's recorded on
it). They don't sell 'em any more (gee, I wonder why) and he doesn't
want to buy a new one and have to learn how to use it all over again.
So I offered to fix it. Looks like a fairly simple PSU problem --
multimeter says 0V on all rails, even though the B+ is up (+330V
off-line, which is about right for the peak voltage of a 240V AC line).
The power supply is a cheap-junk little Chinese switcher based off a
Philips TEA1522 chip.
A bit of debugging revealed a ton of junky "Capxon" and "JPCon"
capacitors, which were all duly replaced with Matsushita-Panasonic FC
and FM-series low-ESR electrolytics of the same or greater voltage
rating. I also found a dead Schottky in the 12V rail regulator -- this
was replaced with a higher-rated substitute.
I'm now in the situation where when I power the thing up, it kicks in
for about a second, then the output Schottky blows dead-short, and the
Philips chip goes BANG. Literally -- as in, sparks flying, and its
output power-FET blows open. There's SUPPOSED to be an over-current
limiter for the FET built into the chip -- but this never works. I've
lost all three of my spare chips already. So I guess it's fair to say
this chip is about as good as everything else I've seen Philips/NXP
produce: that is to say, it's a pile of steaming crap.
Now onto the real question: what could cause the output Schottky to blow
shorted in this manner? All that's on the output is a Pi filter:
SB650 =====
XFMR >----|>|-----*----^^^^^----*------> Vout
| |
--- ---
### ###
| |
=== ===
The two capacitors are 1000uH 25V low-ESR electrolytics, the inductor is
a 10uH "wire round a cylindrical former" thing.
So far the best theory I've come up with is that the design was marginal
to begin with, and expects that the capacitors have a higher ESR than
those I've got installed. That seems silly though -- surely you'd want
the ESR to be as low as possible to keep the ripple down?
Any comments (or enlightenment) would be most appreciated... I've spent
two days on this thing and it's still being a complete ass. If it wasn't
for the HV rail it uses to drive the VFD, I'd have probably replaced it
with one of the 5V/12V mixed-mode PSUs from my junk box...
Thanks,
--
Phil.
pic...@philpem.me.uk
http://www.philpem.me.uk/
--
http://www.piclist.com PIC/SX FAQ & list archive
View/change your membership options at
http://mailman.mit.edu/mailman/listinfo/piclist
Steve Smith
Check how the voltage regulation takes place... it could be either a tl431
and an opto or it could be from the aux winding that feeds the chip... if
the diode on the aux winding is poorly (open) this may cause the op voltage
to exceed the PIV of the diode... and also check the snubber RC across the
flyback fet... open circuit R is also a possibility.
Steve
Philip Pemberton
> Phil:
> Check how the voltage regulation takes place... it could be either a tl431
> and an opto or it could be from the aux winding that feeds the chip... if
> the diode on the aux winding is poorly (open) this may cause the op voltage
> to exceed the PIV of the diode... and also check the snubber RC across the
> flyback fet... open circuit R is also a possibility.
It's a TL431 and an optoisolator (I've already replaced both of them).
In failure mode, the +12V rail does seem a bit high -- around +20V by
the time I get my meter across it.
There's a HER107 series diode on the Aux winding which appears to be
used to power the TEA1522 (diode - resistor - capacitor to ground). The
Aux winding also gets wired to the AUX pin on the 1522 via a 75k resistor.
Dwayne Reid
>I'm now in the situation where when I power the thing up, it kicks in
>for about a second, then the output Schottky blows dead-short, and the
>Philips chip goes BANG. Literally -- as in, sparks flying, and its
>output power-FET blows open.
I've successfully fixed problems like that but its painful.
Your going to need an isolation transformer to safely troubleshoot
this. Also disconnect the PSU from the remainder of the DVD recorder
if possible (pull the connector if one exists).
Start by disconnecting the bottom end of the primary winding from the
chip. Connect the Drain pin on the chip to Vunreg with a high-value
resistor (100k or so). Also connect a variable supply to the Vdd pin
on the chip and another supply to whichever output rail is monitored
by the voltage feedback network.
With both variable supplies turned to 0V, apply primary
power. Monitor the waveform on the Drain pin of the chip.
The chip should be operating in burp mode at this time - the voltage
on Vdd rises slowly until it reaches the start threshold, then you
should see PWM on the Drain pin. That will last until the charge
stored on the Vdd capacitor drops too low, then PWM stops and the
cycle repeats.
Now bring up the variable supply on the Vdd line. You will reach a
point where chip stays working, rather than burping. PWM duty-cycle
should be at max (whatever that happens to be for this particular chip).
Now start to bring up the supply on the output rail so as to provide
feedback for the chip. You should reach a point where the PWM
duty-cycle starts to decrease. That point should be about the rated
output voltage for that rail.
This should allow you to get close to the area that isn't working.
Couple of other suggestions:
1) check the resistor on the Source pin on the chip. Is it open or shorted?
2) When you think that you've found the problem, re-connect the
bottom of the primary winding back to the Drain pin on the chip via a
resistor. Start with 1k 1/4W and see if it burns up. If it does,
you haven't found the problem.
Whatever you do find to be the problem, do let us know what it
was. Stuff like that is always useful to hear about and to get added
to the 'how to fix smps problems file.
dwayne
--
Dwayne Reid <dwa...@planet.eon.net>
Trinity Electronics Systems Ltd Edmonton, AB, CANADA
(780) 489-3199 voice (780) 487-6397 fax
www.trinity-electronics.com
Custom Electronics Design and Manufacturing
Richard Prosser
I'd also suggest checking / replacing any lower value capacitor that
look even slightly "dodgy". You could have some RF instability
occurring.
RP
Philip Pemberton
> I've successfully fixed problems like that but its painful.
Indeed. I think it's safe to say the designer of this PSU was an
absolute masochist...
> Your going to need an isolation transformer to safely troubleshoot
> this. Also disconnect the PSU from the remainder of the DVD recorder
> if possible (pull the connector if one exists).
There are three connectors. The 12V rail only goes to the hard drive
(which is disconnected).
An isolation transformer might be a tricky thing to get hold of unless I
can bodge one together out of a couple of identical AC transformers.
> Start by disconnecting the bottom end of the primary winding from the
> chip. Connect the Drain pin on the chip to Vunreg with a high-value
> resistor (100k or so). Also connect a variable supply to the Vdd pin
> on the chip and another supply to whichever output rail is monitored
> by the voltage feedback network.
I tried that with a 10-ohm resistor -- that got the PSU to start up,
then a second later the HDD tried to spin up and blew out the 12V rail
Schottky again.
Speaking of which, I seem to have found an error in my diode cross
reference. It listed the diode I was using (a surface-mount
Vishay-Gensemi thing) as a "near identical replacement" to the SB560 in
the secondary circuit. Seems "near identical" isn't close enough; I'm
going to replace it with a proper SB560 "of reputable manufacture"
(Fairchild Semi.) tonight and try again.
I'm probably going to fit an IC socket for the TEA1522, too...
> 1) check the resistor on the Source pin on the chip. Is it open or shorted?
It measures (in-circuit) as a one-ohm resistor.
> 2) When you think that you've found the problem, re-connect the
> bottom of the primary winding back to the Drain pin on the chip via a
> resistor. Start with 1k 1/4W and see if it burns up. If it does,
> you haven't found the problem.
OK, well I've got a reel full of 1k 0.25W metal-film resistors. Should
be fun.
> Whatever you do find to be the problem, do let us know what it
> was. Stuff like that is always useful to hear about and to get added
> to the 'how to fix smps problems file.
I'm thinking 'output schottky not up to spec' and 'controller chip...
again'. I'm also somewhat suspicious of the new capacitors, maybe the
inrush current is just too much for the PSU to handle?
Olin Lathrop
> Indeed. I think it's safe to say the designer of this PSU was an
> absolute masochist...
No, probably quite clever. This is a high volume consumer item. A lot of
design effort probably went into making the price as cheap as absolutely
possible while keeping the lifetime just past the warranty limit. People
pretty much buy these things on price, so this is what the manufacturers are
forced to produce. The technology also changes fast enough and osbsolecense
is expected, so most consumers won't think much of it if the unit fails in
five years. That will just be a excuse to update to the latest anyway.
I admit I pretty much buy such things on price too. I might be willing to
pay more for one if I could be convinced it was actually better, but I have
no confidence that the retail price I see has much correlation with quality.
So I roll the dice and minimize my exposure because there is little else I
can do.
********************************************************************
Embed Inc, Littleton Massachusetts, http://www.embedinc.com/products
(978) 742-9014. Gold level PIC consultants since 2000.
John Chung
> I'm thinking 'output schottky not up to spec' and
> 'controller chip...
> again'. I'm also somewhat suspicious of the new capacitors,
> maybe the
> inrush current is just too much for the PSU to handle?
>
> --
the schottky.... I would check anything related to the inrush current
for the hard disk. You can use a variable load instead of the
hard disk as a test.
John
Philip Pemberton
> forced to produce. The technology also changes fast enough and osbsolecense
> is expected, so most consumers won't think much of it if the unit fails in
> five years. That will just be a excuse to update to the latest anyway.
What annoys me is that it hasn't even lasted a year, there are no
contact details for the manufacturer in the manual, and the store wasn't
even remotely interested in helping out.
A rough transcript:
"So what do you expect me to do with it? You need to talk to the
manufacturer."
"What I expected was either a repair or a replacement. As for the
manufacturer, do you have their contact details?"
"No and no."
"Excuse me?"
"No replacement, no refund, no repair, and no I don't have the
manufacturer's contact details. They're in the instructions."
"No they're not, I have the instructions right here..."
"Not my problem. Google it or something. Now in the meantime, unless
you're buying something, get out."
*sigh*
sergio masci
On Sat, 12 Jun 2010, Philip Pemberton wrote:
> On 12/06/10 13:29, Olin Lathrop wrote:
> > forced to produce. The technology also changes fast enough and osbsolecense
> > is expected, so most consumers won't think much of it if the unit fails in
> > five years. That will just be a excuse to update to the latest anyway.
>
> What annoys me is that it hasn't even lasted a year, there are no
> contact details for the manufacturer in the manual, and the store wasn't
> even remotely interested in helping out.
>
> A rough transcript:
>
> "So what do you expect me to do with it? You need to talk to the
> manufacturer."
> "What I expected was either a repair or a replacement. As for the
> manufacturer, do you have their contact details?"
> "No and no."
> "Excuse me?"
> "No replacement, no refund, no repair, and no I don't have the
> manufacturer's contact details. They're in the instructions."
> "No they're not, I have the instructions right here..."
> "Not my problem. Google it or something. Now in the meantime, unless
> you're buying something, get out."
>
> *sigh*
As I understand it, you as a consumer have an automatic 12 month guarantee
from the supplier (the guy that sold you the DVR). If he wont play ball
you can complain to the trading standards people. Shame you've already
started the repair yourself, you don't have a leg to stand on now.
I did a quick search and this was my first hit:
http://www.lawguarantee.com/consumer-protection-articles/guarantee-warranty-law-uk.html
maybe it will inspire you to do a little more research regarding your
rights in case you get bitten again.
Regards
Sergio Masci
Mike Harrison
Assuming this was a retail sale, they are wrong - your agreement is with the retailer, not the
manufacturer. If you were sold something advertised as having a warranty, the retailer is ultimately
responsible for honouring it.
Step 1 : Trading Standards. Step 2 : Small Claims.
peter green
> On 12/06/10 13:29, Olin Lathrop wrote:
>
>> forced to produce. The technology also changes fast enough and osbsolecense
>> is expected, so most consumers won't think much of it if the unit fails in
>> five years. That will just be a excuse to update to the latest anyway.
>>
>
> What annoys me is that it hasn't even lasted a year, there are no
> contact details for the manufacturer in the manual, and the store wasn't
> even remotely interested in helping out.
>
> A rough transcript:
>
> "So what do you expect me to do with it? You need to talk to the
> manufacturer."
> "What I expected was either a repair or a replacement. As for the
> manufacturer, do you have their contact details?"
> "No and no."
> "Excuse me?"
> "No replacement, no refund, no repair, and no I don't have the
> manufacturer's contact details. They're in the instructions."
> "No they're not, I have the instructions right here..."
> "Not my problem. Google it or something. Now in the meantime, unless
> you're buying something, get out."
>
RIGHTS?
Philip Pemberton
> As I understand it, you as a consumer have an automatic 12 month guarantee
> from the supplier (the guy that sold you the DVR). If he wont play ball
> you can complain to the trading standards people. Shame you've already
> started the repair yourself, you don't have a leg to stand on now.
I've had experience dealing with idiots like this guy before... usually
even with trading standards getting involved, all you get is the fun of
having someone (verbally or physically) spit in your face.
And I'm not going to small claims court over the matter of an £80 DVR.
Something like a DMM (~£300), LCD TV (~£600) or an oscilloscope (~£1400)
then I'll be more than happy to give the bastards hell. But when this
happens it's easier just to say "I'm not buying from that guy again,
he's a git." and chalk it up to experience.
--
ivp
> manufacturer's contact details
In NZ the initial contract is between the customer and the retailer.
Then the retailer and the manufacturer. Retailers try to absolve
themselves from responsibility by displaying their own rules and
conditions for returns and repairs but the real law is the only thing
that matters
"OK, see you on Fair Go" is not a phrase most NZ retailers want
to hear
http://en.wikipedia.org/wiki/Fair_Go
Philip Pemberton
> you did mention that when the HD started to spin up THEN is blew
> the schottky.... I would check anything related to the inrush current
> for the hard disk. You can use a variable load instead of the
> hard disk as a test.
Turns out it was the Schottky on the 12V rail which died, and took out
the TEA1522. Replacing the Schottky with the exact specified part solved
the problem.
Philip Pemberton
> In NZ the initial contract is between the customer and the retailer.
It's the same over here, though a couple of retailers are quite picky
about which parts of the Sale of Goods Act they'll obey. Usually the
ones which are more advantageous to the seller are kept and the ones
that aid the buyer are ignored...
A couple of electrical retailers got nailed by Trading Standards for
this a few years ago... Dixons (DSGi, PC World, Currys) are no stranger
to getting a good whack from the Powers That Be, and the other
"warehouse-style" electrical chains are scarcely any better.
> "OK, see you on Fair Go" is not a phrase most NZ retailers want
> to hear
>
> http://en.wikipedia.org/wiki/Fair_Go
Sounds similar to Watchdog, though they seem to have become a bit
toothless as of late...
http://en.wikipedia.org/wiki/Watchdog_%28TV_series%29
Michael Watterson
> On 13/06/10 00:20, ivp wrote:
>
>> In NZ the initial contract is between the customer and the retailer.
>>
>
> It's the same over here, though a couple of retailers are quite picky
> about which parts of the Sale of Goods Act they'll obey. Usually the
> ones which are more advantageous to the seller are kept and the ones
> that aid the buyer are ignored...
>
> A couple of electrical retailers got nailed by Trading Standards for
> this a few years ago... Dixons (DSGi, PC World, Currys) are no stranger
> to getting a good whack from the Powers That Be, and the other
> "warehouse-style" electrical chains are scarcely any better.
>
>
(those are with Manufacturer usually). The goods have to be "fit for
purpose " which means working for 1 to 6 years depending on king of
product.
Manufacturer's warranty or returns policies are of course important if
you purchase for Manufacturing, You are a Business buying it retail or
you ARE the retailer. :-)
alan.b...@stfc.ac.uk
> > I've successfully fixed problems like that but its painful.
>
> Indeed. I think it's safe to say the designer of this PSU was an
> absolute masochist...
Yeah, he used a chip that starts with TEA....
I remember when SMPS chips were just starting to come on the market one
of my colleagues had a project where such a chip looked like a useful
device to use. He started off with a Philips TEA-something that seemed
to be the first such chip available where we were, and he had no end of
trouble with it. There were that many current limit-voltage limit-duty
cycle limit- whatever else the designers could think of to limit pins
that he never ever got it working. Later on he got hold of a Motorola
chip and had it running perfectly happily in about 2 days.
> An isolation transformer might be a tricky thing to get hold of unless
I
> can bodge one together out of a couple of identical AC transformers.
For an item of that current draw that is probably the ideal way to do
it. Also adds some source impedance on the mains side so you don't quite
have your local mains company attempting to be an ideal voltage source
when things do want to go bang.
> OK, well I've got a reel full of 1k 0.25W metal-film resistors. Should
> be fun.
They do make good 'glow in the dark' lamps ... ;)))
--
Scanned by iCritical.
alan.b...@stfc.ac.uk
> contact details for the manufacturer in the manual, and the store
wasn't
> even remotely interested in helping out.
>
> A rough transcript:
>
> "So what do you expect me to do with it? You need to talk to the
> manufacturer."
> "What I expected was either a repair or a replacement. As for the
> manufacturer, do you have their contact details?"
> "No and no."
> "Excuse me?"
> "No replacement, no refund, no repair, and no I don't have the
> manufacturer's contact details. They're in the instructions."
> "No they're not, I have the instructions right here..."
> "Not my problem. Google it or something. Now in the meantime, unless
> you're buying something, get out."
>
> *sigh*
Sounds like time to get a local Trading Standards Officer to call on
them and remind them of their obligations to consumers. Also known as
're-education' ... ;)))
Even just the threat of doing so seems to produce results often.
--
Scanned by iCritical.
alan.b...@stfc.ac.uk
> Assuming this was a retail sale, they are wrong - your agreement is
with
> the retailer, not the
> manufacturer. If you were sold something advertised as having a
> warranty, the retailer is ultimately
> responsible for honouring it.
> Step 1 : Trading Standards. Step 2 : Small Claims.
To me there is a missing step - 1A: Watchdog
http://www.bbc.co.uk/watchdog/
(for those not in the UK, this is our local Consumer Rights TV program.
You don't want to get on the wrong side of Annie when she is in full
flight, or be a Rogue Trader in Matt Alwrights sights).
--
Scanned by iCritical.
Picbits Sales
Lidl here in the UK.
I bought 2 shower mixers from one of them which had a "5 year" warranty on
it.
My father in law bought one of the units off me but it broke a few months
later. The store told him to contact the manufacturer which he did (in
Italy) and then spent the next couple of months running round in circles
trying to get it sorted.
He repeatedly returned to the store to try and get it sorted and was sent
away each time. They have now totally boycotted the stores along with most
of their friends and family in their town.
After 3 months of having no shower I gave him my other unit as a replacement
which is still (touch wood) working.
Apalling customer service over a £30 item and I've not set foot in their
stores since.
----- Original Message -----
From: "Philip Pemberton" <pic...@philpem.me.uk>
> A rough transcript:
>
> "So what do you expect me to do with it? You need to talk to the
> manufacturer."
> "What I expected was either a repair or a replacement. As for the
> manufacturer, do you have their contact details?"
> "No and no."
> "Excuse me?"
> "No replacement, no refund, no repair, and no I don't have the
> manufacturer's contact details. They're in the instructions."
> "No they're not, I have the instructions right here..."
> "Not my problem. Google it or something. Now in the meantime, unless
> you're buying something, get out."
>
> *sigh*
>
> --
> Phil.
--
RussellMc
> it. Also adds some source impedance on the mains side so you don't quite
> have your local mains company attempting to be an ideal voltage source
> when things do want to go bang.
Putting a mains light bulb in aq socket in series with the input can
be a useful trick.
Start with a low wattage bulb and then if all seems OKish change to a
larger wattage one.
*** DO ** use a filament (incandescent) lamp for this
CFL's are not suitable :-).
As tungsten filaments have positive temperature coefficient of
resistance, if the lamp current is small relative to rated current
then the resistance will be much lower than when running at full
power. Reductions of 10:1 occur in practice (depends somewhat on lamp,
construction etc)
You could also use a PolySwitch or PolyFuse. They tend to be slower
than a lamp on turn on and much slower to "reset" but have a wide
off:on resistance ratio. Wi8kipedia PolySwitch entry is talking about
something else which may warrant the name but which is a much less
common use of the term. Somebody may wish to disambiguate it (123
down).
Good oil: http://en.wikipedia.org/wiki/Resettable_fuse
Carbon filament lamps need not apply :-).
Russell
Michael Watterson
> Now this sounds very like a similar conversation I've had with both Aldi and
> Lidl here in the UK.
>
> I bought 2 shower mixers from one of them which had a "5 year" warranty on
> it.
>
> My father in law bought one of the units off me but it broke a few months
> later. The store told him to contact the manufacturer which he did (in
> Italy) and then spent the next couple of months running round in circles
> trying to get it sorted.
>
> He repeatedly returned to the store to try and get it sorted and was sent
> away each time. They have now totally boycotted the stores along with most
> of their friends and family in their town.
>
> After 3 months of having no shower I gave him my other unit as a replacement
> which is still (touch wood) working.
>
> Apalling customer service over a £30 item and I've not set foot in their
> stores since.
>
>
I've had to return several things to Lidl. They refunded me without quibble.
(Ireland)
Picbits Sales
One was for the showers - they were bought from Lidl who refused to get
involved when the first one broke.
I've also had the same problem with Aldi - I bought an electric impact
wrench which broke after the first couple of uses. I took it back and got
another which lasted another month or so but broke. I then asked for a
refund on the grounds of "unfit for purpose" which was refused and I was
told to call the number for the manufacturer (who were actually pretty
good). I had a new one sent over but daren't really use it as I know it will
break. I was able to keep the old one and dismantled it to find out what was
wrong and its just a really weak design.
The Lidl store in question with the shower problem is the one in Saltash /
Cornwall - I'm quite happy to name and shame considering the amount of
problems my inlaws have had with them.
----- Original Message -----
From: "Michael Watterson" <mi...@radioway.org>
To: "Microcontroller discussion list - Public." <pic...@mit.edu>
Sent: Monday, June 14, 2010 2:50 PM
Subject: Re: [EE]: SMPSU repair -- secondary schottky shorts and similar
stuff