Case of failed AC adapter and interesting device

[Alexei.Polkhanov]

On Friday I lost my internet connection abruptly. When I went to check modem (2WIRE ADSL modem) I found dead AC adapter. modem was damaged at phone line input but that is another story.

It was 5V DC, 2A rated adapter made out of two glued plastic shells and PCB inside. I never cared about power supplies - when I need one for project or something else I buy a brick from DigiKey. I cracked my failed adapter open with screwdriver and immediately found busted 100uF/10V electrolytic aluminum capacitor inside. I replaced the capacitor, taped the shells back together and plugged it into socket to test this is when I started finding out strange things about it. Strange only to me of course 

First of all it suppose to generate 5V DC output and according to specs on the case it is rated to up to 2A. I hooked it up to large 10W, 8 Ohm resistor and I got ~4.54V, when I added another one in parallel I got 4.1V or so. 5V/4ohm = 1.25A, not good, but modem was OK with it I guess. When I looked at it again I found the mystery device at the output of an adapter.

It is a TO-92 package marked as "431-B  SI 9G". I did some googling and I found this datasheet http://www.fairchildsemi.com/ds/TL/TL431A.pdf. I am not entirely sure it is the right datasheet. I de-soldered it and tried to test it, but it is either this device is busted too or I don't understand how it suppose to work. According to a datasheet it emulates a simple zener diode, but in addition to basic zener functionality it allows to control breakdown voltage by applying reference voltage to REF pin.

According to a datasheet I expect reverse breakdown voltage of this "emulated zener" to change when I apply something around 2.495V  to a REF pin. What I got instead is that it works as a gate - it breaks down when I apply ~1.8V to REF. Is this happen because

A. I do not understand how it works?
B. It is damaged device?
C. both?

[amyk]

1.8V sounds too much like a standard value to be a coincidence, but a regular '431 should have 2.5V internal reference. You'll need to see how it's hooked up to sense the output voltage; usually it's a voltage divider, for 5V and 2.5Vref it'll be halving the output.

[megajocke]

According to a datasheet I expect reverse breakdown voltage of this "emulated zener" to change when I apply something around 2.495V  to a REF pin. What I got instead is that it works as a gate - it breaks down when I apply ~1.8V to REF. Is this happen because

A. I do not understand how it works?
B. It is damaged device?
C. both?

It isn't really clear what you mean by "break down", but see the test circuit in Figure 1 and 2 in the datasheet on how to test the device in its zener mode. Connected like in Figure 1 it should act like a 2.5 V zener diode.

[Alexei.Polkhanov]

Nope, 1.86V   
Here is the schematics of low voltage portion of AC adapter that I have reverse engineered.

As you can see resistors R14 and R18 would generate reference 2.389V at 5V output. If I want this circuit to provide 5V out I need R18 to be ~2.9K if reference is 1.86. I assembled the circuit annd replaced R18 by potentiometer and surely it was 2.93Kohm when I adjusted it for 5V out. Another interesting observation is that if I add 10K series resistor to REF pin to limit Iref then that 1.86 would be more like 2.36V.

But what puzzles me even more now after looking at this app note http://www.onsemi.com/pub_link/Collateral/TND381-D.PDF and others - RLED or current limiting resistor for optocoupler is MISSING !? Or maybe whoever designed it found that is is not needed for 5V output?

[Rufus]

As you can see resistors R14 and R18 would generate reference 2.389V at 5V output. If I want this circuit to provide 5V out I need R18 to be ~2.9K if reference is 1.86. I assembled the circuit annd replaced R18 by potentiometer and surely it was 2.93Kohm when I adjusted it for 5V out. Another interesting observation is that if I add 10K series resistor to REF pin to limit Iref then that 1.86 would be more like 2.36V.

Newark in Canada list TL431s in TO92 at 9.7 cents. Why waste your time experimenting with one that is probably damaged?

[megajocke]

Well, then it sounds like it's broken.

[Alexei.Polkhanov]

Newark in Canada list TL431s in TO92 at 9.7 cents. Why waste your time experimenting with one that is probably damaged?

I just got large order of parts from DigiKey, I added few TL431 to new order so that I can compare, but I have not submitted it cuz it is not worth the shipping.

I did not plan on this investigation into AC adapter - it just kind of happened that two of them failed in 6 months, same model, and I started getting curious what is wrong with them. 1000uF electrolytic caps failed in both cases, but now I am thinking maybe it is not just as simple ...

[free_electron]

tl431 is a shunt regulator.

take 5 volts, via 1 kiloohm resistor to kathode. anode to ground. connect ref to cathode . measure voltage over device : it should be 2.5 volts.

you need to send a good 1 milliampere through it to work. this thing is NOT a zenerdiode. it is an opamp with a built in bandgap , divider and a shunt transistor ( not a pass transisotr like in a classic regulator like a 317 or 7805.

this device compares the voltage in its vref pin with an internal 2.5 volt reference. it drives its transisotr in conduction so that there is 2.5 volts across it.

so basically that vref is the feedback pin.
you need a series resistor as this shunts the output to ground...

see it this way : a classic regulator takes an input voltage and control how much is passed form input to output. (series regulation)
this thing controls how much is deviated to ground.... it is essentially a voltage divider where you set the voltage over the bottom tap by programming vref.

the advantage of this thing is that it has a very low output impedance. a series regulator doesn't. if you have a series regulator making 5 volts , and you apply 5.1 volts at the output the regulator stops working. you will have 5.1 and the pass transistor will turn off. so you can 'overdrive' the output of a series regulator.
this thing wont et you. if it is set to make 5 volts and you apply 5.1 it is going to throw its transisotr hard in conduction trying to make that 5.1 go down to 5 by deviating electrons into ground. (if you dont limit the current the poor thing will fry itself trying.

so these things make excellent voltage references as they are a 'hard' output impedance.

[chickenHeadKnob]

Hi Alexei.

Your problem sounds somewhat similar to one I had back in 2000. I just received a Nortel ADSL modem (ya I know, Nortel). MTS (Manitoba telephone systems) issue or whatever they are called now. First power-up the Led flickers and dies after 2 seconds. Back to the MTS service centre for a new wall wart. New one dies instantly. Back to the MTS service centre, this time I take the modem and 2nd wart, tell the guy I think the modem is drawing too much current. He didn't believe me. I   suggested we replace the floor display unit which was running for customers to play with, with my own. He didn't want to, but after much pestering, relented. Plugged it in, it of course powers up normally and runs like a champ for at least 15 minutes. See he says, it was just the transformer. In the meantime we had also tested the 2nd transformer and found no output. I look stupid.

 Ok I say, I guess I will take it back. Just as he is reaching to unplug my modem (and not looking at the front leds) I see the power led die. There! it just died!. He doesn't believe me. Just humour me and plug it back in, you'll see! We were both getting pissed off with each other at this point. So he plugged it back in like it was the most onerous thing he had to do all month and then I ended up with another modem.

Moral of my story don't always blame the power supply, those ADSL modem are pretty shifty.

[Experimentonomen]

Dlink are also notorious for beeing delivered with undersized wallwarts.

[Alexei.Polkhanov]

Here is the video on Youtube that where guy investigates exactly same problem with his 2WIRE router and AC adapter: . He like myself had failed capacitor inside AC adapter. I also found failed capacitor exactly like on video, but I decided to investigate it little further.

Let's assume that it is built with best brand name parts and look at schematics. As I mentioned I think that there should be a current limiting resistor between output and opto-coupler.

[SpiderElectronics]

Here in the UK a few years back, BT (British Telecom) were shipping ADSL modems with the 2-Wire branded adapters and they recalled the entire lot and sent every customer a free replacement. That must have been millions of units.

[Rufus]

and others - RLED or current limiting resistor for optocoupler is MISSING !?

A current limiting resistor is not required, the TL431 controls the current in the LED so that the current transferred to the hot side is what is required to produce +5v output.

If something else fails in the circuit so it can't quite output 5v or a marginally excessive load is applied then current in the LED could be excessive and that is probably what damaged the TL431. A resistor is not required for it to work, it might limit damage when it doesn't work.

The circuit arrangement is very common. This http://www.ti.com/product/tps5904 is an opto-isolator with a TL431 integrated.

[amyk]

If something else fails in the circuit so it can't quite output 5v or a marginally excessive load is applied then current in the LED could be excessive and that is probably what damaged the TL431. A resistor is not required for it to work, it might limit damage when it doesn't work.

Actually the opposite for this circuit - an open feedback loop causes the voltage to go up; the LED is off until the TL431 starts conducting, which occurs when the output reaches 5V.

The primary side probably is some variant on the basic ringing-choke converter.

[Alexei.Polkhanov]

Here I think I have almost complete, reverse-engineered schematics of the adapter. I was unable to identify emitter and collector of the Q2 BJT transistor although it is clear that it is PNP. One p-n junction is 0.6V another - 1.4V. Not much else I can get from looking at it - it has markings "1AM". Also I found strange zener diode - Z4, it should be a zener, but marking on is "2B1", but it has ~0.7 forward drop and ONLY 1.61V in reverse - does that make sense? So Z4 is probably dead as well. I did get new TL431-s and, yes I do get ~2.45V at REF pin, so one I found on the failed adapter was definitely damaged.  WFF2N60 MOSFET appears to be ok.

Assuming that I did get the schematics right I don't understand the purpose of Z4.

[amyk]

Secondary side is basically correct, primary side looks a little odd. Here are some things you might want to confirm:

* Snubber diode is backwards
* Auxillary coil should have a rectifier and reservoir cap
* Zener in the wrong place
* Orientation of Q2 doesn't make sense - increasing optocoupler current should pull down the base of Q1, reducing the voltage

Check the schematic against the one in here: http://www.st.com/st-web-ui/static/active/en/resource/technical/document/application_note/CD00066269.pdf
It looks like you have all the components of that one (even the dual source resistors in parallel) but not the same connections.

[Alexei.Polkhanov]

amyk you right - I had the flyback diode backwards, zener (Z4) was wired incorrectly so was a a Q2. I probed Q2 while still on PCB and I did get it wrong, also it is NPN (not PNP). That app note does help to make sense of it indeed.

Regarding the rectifier and reservoir cup for secondary coil - I am not sure.
I have corrected schematics attached here. I got measured values for all caps as well.

I also checked the transformer trying to get it parameters.
L12 = 7.46uH, R12 = 0.015ohm
L36 = 2.318mH, R36=2.183ohm
L45 = 13.62 uH, R45 = 0.394ohm

So here 3-6 will be primary, 1-2 is secondary.

[Alexei.Polkhanov]

Well finally did get replacement AC adapter for my modem and well they gave me completely new model which is much better quality. Well since I ended up with 2 adapters I decided to open one and check what is inside. By looking at components and data sheets I concluded that it is a PWM-type. Schematic in datasheet for PWM controller pretty much explain everything except few things:

What are those components that I have marked by red/purple squares? Yellow plastic box looks to me as Polypropylene capacitor (?), black box next to it is filter? That blue thing maybe an EMI suppression cap? Almost any power supply I ever looked at has one of these. All these square/triangular and round compliance signs printed on them also make me think that these are for EMI suppression, or maybe I am wrong?

[mariush]

Yeah the blue cap is a Y cap (and it looks like it's proper safety rated) ... ties the grounds between primary and secondary for EMI supression.
pink ... it's a resistor masquerading as a fuse, you can see the fuse text on the pcb.
yellow square is an x2 capacitor, for filtering AC line along with that small inductor ... kinda like this one : http://www.digikey.com/product-detail/en/PHE840MY6470MD14R06L2/399-5429-ND/1927374

Read this about class x1, x2 ,y capacitors etc : http://my.execpc.com/~endlr/line-filter.html

The gray square thing could be another capacitor ... it could also be a sort of thermistor/varistor (limit rush current when plugging in but I doubt it - i'm only saying it because I see the text on it starting with T and then some numbers)

[Alexei.Polkhanov]

The gray square thing could be another capacitor ... it could also be a sort of thermistor/varistor (limit rush current when plugging in but I doubt it - i'm only saying it because I see the text on it starting with T and then some numbers)

I think I have correctly identified it as a fuse: http://www.farnell.com/datasheets/1520844.pdf, http://uk.farnell.com/multicomp/mst-2a-250v/fuse-radial-slow-blow-2a/dp/1566103