Nespresso Delonghi Coffee Machine Repair

[WannabeGeek]

I've been pulling my hair out with this one.  The coffee machine just quit working one day, a little pop then no sounds, no lights, nothing, so I had a look inside.  A quick look at the circuit board revealed a blown triac, so I tested all triacs and they were all good except the one in the red circle (see photo).  Following the wires to see what that Triac was powering, I noticed that the solenoid valve looked a bit rusty.  A quick test revealed that it was also faulty.  The replacement solenoid valve is hard to find and costs around €25 so I wanted to make sure everything else was working before ordering it. I replaced the faulty triac and removed the solenoid valve from the circuit. Powered up, the pump is now on (it shouldn’t be) but there are still no lights or anything else.  I couldn’t find a schematic anywhere, so I had a closer look at the power section (see my schematic).  At first, it looked like a voltage doubler but that didn’t seem to make sense since there’s plenty of voltage to go around.  Since the x capacitor should act as a large resistor to limit current, this must be some sort of rectifier.  The 2 zener diodes are a bit confusing.  Why clamp the voltages and then use a linear regulator?  Anyway, I started testing voltages.  The AC voltages were all normal, but getting into the section that powers the ICs and the lights, there are some issues.  Across the input of the 78M05, DZ1, and C3 there are only 2.8 volts.  I was expecting 8.2v.  Of course, there’s no output voltage on the 78M05 which explains why there are no LEDs lighting up. Across C4 and DZ2 are .2 volts! Needless to say, I tested all the Diodes with my multimeter and they all seem to be good.  I desoldered DZ2 and tested it out of circuit with load and it is functioning ok although the zener voltage should be 5.1v and I only get 4.5v.  I even replaced it temporarily with a through hole 5.1v zener, but the resulting voltage in circuit was the same.  I tested all the caps with my ESR meter and they’re all good.  I took out C4 just to make sure, and it checks out with no leakage or issues. With all the parts back in (except the solenoid valve) I turned it on and decided to inject 9v across the 78M05 input.  Suddenly the pump stopped and the LEDs lit up and went through what looked like a start up sequence.  Sadly, with this little trick there’s still no control over the buttons or anything else.  Since the logic section after the 78M05 seems ok, I suspect that if I can get the power section working properly, the machine should perform as usual except for the solenoid valve which I could then justify replacing. Unfortunately, I’m running out of ideas since all the caps, diodes, resistors, triacs, and MOVs test good, as far as I can test them.  There are no fuses to be seen anywhere.  Does anybody have any ideas?  The Nespresso Delonghi is quite expensive in this country at about €300 so I really don’t want to just throw it away and buy a new one.

[Paul Moir]

It sounds like something is loading down the -5v rail made by D2 and maybe the 8.2v rail.  What's the resistance across it like?  When you injected 9v into the 8.2v rail, did anything get warm? 

I would be suspecting the ULN2004 driver since it appears to feed the blown TRIAC, but that would require further testing.  Also, could you post a pic of the other side of the board so we can follow the traces?

EDIT:  the power supply is a capacitor dropper type.  They're usually designed to supply the maximum current required by the circuit, so they can't handle much additional load without dropping in voltage.  Replace the C2 capacitor's impedance with a big power resistor in your mind to get the idea. 

[WannabeGeek]

Thanks for the help!  I wasn’t even considering negative rails ;-).  The resistance across DZ1 starts at around 35M and counts down slowly like a capacitor is charging.  Discharging C3 starts the process over again.  The resistance across DZ2/C4 is a stable 613K. When you refer to the ULN2004 driver, are you talking about the transistors and capacitors directly above it?  The front side of the board is covered with wires and heatsinks so it’s very difficult to make out the traces (see photo).  I’ve been shining a bright light through it at different places and using my continuity tester to get the rest of the connections. Nothing got hot when I injected 9v into the 78M05.  I was using a battery.

[Paul Moir]

Ok, so it doesn't look like anything is loading down the power supply rails, I would have expected quite a bit less.  Really that only leaves R100 and C2 in the supply side.  You should be able to check R100 in circuit easily enough.  Do you have any way of checking C2?  Again, you should be able to measure this one in circuit.  What I don't understand is why either would have failed:  they don't have much in common with that TRIAC.

The ULN2004 I suspected is U3, which has one of it's outputs wired via a resistor into the gate of the blown TRIAC.  It's pretty close to a bunch of transistors in a package.  It's a very common and useful device.

Thank you for the picture of the other side.  Even with the wires in the way it's very helpful. 

[WannabeGeek]

I haven't pulled out the R100, but in circuit it measures 99 Ohms which is pretty much spot on.  It's not shorted or open.  I could pull it out but I doubt that will make a difference.  The C2 tests very low ESR with my meter and it's right on 1.2 uF.  I actually removed it initially, because it was one of the first things I suspected. Out of circuit, it tested the same but I haven't loaded it with 200+ AC.  It's rated at 275v. When I had it out of circuit, I powered up the machine just to see what would happen and the pump (which is now always on) was suddenly silent.  After putting it back in, the pump was making its usual purr when the Delonghi was turned on.  When powered up, I can measure mains voltage across that xcap.  I'm assuming that's normal. Also the MOVs are all open circuit, which I'm assuming is normal.  I can't test them with high voltage either.  There's also mains voltage at all the triacs.  Actually, I think I was getting around 120v AC on all of them except the TY5 which was reading 228v.  I was assuming that it had something to do with the fact that the TY5 wasn't connected to the solenoid valve anymore.  Replacing that part didn't change the voltage reading.  Perhaps that's a clue that could lead me to another problem, but I'm not that well educated about mains AC circuits.

Update:  That 270 Ohm resistor you mentioned between pin 12 of the ULN2004 and the gate of the triac is reading 20K!  I haven't replaced it yet (I don't even have a surface mount resistor) but it also looks damaged under a magnifying glass.  I'll replace it temporarily with a through hole resistor (I hope 1/4 Watt will do) and let you know what happened. 

[WannabeGeek]

I replaced the resistor but there was no change when I powered up the machine.  I injected some voltage (this time around 7.5v because my battery is getting flat) into the 78M05 to turn off the pump and I started measuring voltages again.  The voltage of the new triac TY5 (z0107sn) with the new resistor between that and the ULN2004, reads 8v AC and 0v DC across gate and A2.  Across A2 and A1 is 0v AC and 0 DC.  The voltage of TY3 (identical to TY5) is 110v DC and 122v AC across either gate and A2 or A2 and A1.  The voltage across DZ1 (and my battery when connected) is still 2.8v.  The voltage across C2 is 128v AC, no DC of course and resistance is OL.

[mikerj]

What the resistance of the R100 resistor?  (value printed on component, and measured value).

[WannabeGeek]

Sorry, I should have labeled it better.  The R100 resistor has 100 Ohms, 3w printed on it.  I measured it at 99 Ohms with my meter.

[mikerj]

The voltage across C2 is 128v AC, no DC of course and resistance is OL.

That voltage seems way too low if you are powering this with a 230v supply, I would expect it to be quite close to the mains voltage, maybe 220v or so.

What voltage do you get across the 100 ohm resistor, and what voltage do you get across the rectifier/zener circuit (i.e. between the junction of D2/D5 and the junction of DZ1/DZ2)?

[WannabeGeek]

The voltage across the junction of D2/D5 and DZ1/DZ2 starts at around 9v AC and goes up slowly. Discharging caps starts this process from around 9v AC again.  I'm getting no DC voltage across this junction.  The voltage across R100 is 8.71v AC!  This seems very strange to me, since the voltage across C2 is 228v AC. Where is C2 getting its voltage from?  Other than R100 there's only a MOV connected to that pad on the board and that MOV is connected to neutral, not phase like R100.  The way I understand MOVs is that they're supposed to conduct only when there's a large spike in the AC line.  The reading across the MOV is 228v AC.  This would make sense if I were actually reading the voltage between phase and neutral with the MOV being OL, but the only connection that MOV has to phase is R100.  This would suggest that R100 is conducting mains voltage to that pad (and C2), so why am I only getting 8.7v across R100?  The MOV reads OL when the machine is turned off. Since it's a double sided board, I may be missing something but (set to continuity) my meter isn't beeping anywhere else if one probe is on the phase side of C2, so I still think the only path to C2 must be through R100.  What am I missing here? 

[mikerj]

The voltage across the junction of D2/D5 and DZ1/DZ2 starts at around 9v AC and goes up slowly. Discharging caps starts this process from around 9v AC again.  I'm getting no DC voltage across this junction.  The voltage across R100 is 8.71v AC!  This seems very strange to me, since the voltage across C2 is 228v AC.

That makes far more sense, I presume your previous post about the voltage on C2 was a typo (128v)?  The voltages around this circuit all seem fine.  The current in the circuit (i.e. through R100, C2, D2/D5 and DZ1/DZ2) will be roughly 85mA with a 230v RMS supply, the the voltage you are seeing across R100 is pretty much exactly what you should be seeing.

C2 is simply being used as a current limiter; since it is a reactance rather than a resistance you can drop a lot of voltage without having to dissipate a lot of power.  There is a clear current path from live, through R100 and C2, then through either D2 and DZ1 or D3 and DZ2 (depending on which half cycle of the main we are on) back down to neutral.  All the time this circuit is connected, current is flowing around it and most of the voltage is being dropped across C2.  R100 is present to limit the current through the circuit if there are any transient voltages on the mains.  Since the reactance of C2 drops as frequency increases, a fast voltage transient could cause destructive currents to flow, and R100 limits this.

[WannabeGeek]

Yes, the 128v in my previous post was a typo.  Sorry about that 
That was a nice explanation of the circuit and it clears up a few things, but I'm still stumped as to why the coffee machine doesn't work.
Is it likely that the ULN2004 is fried seeing as how the triac and the resistor going into pin 12 were also blown?  Could the ULN2004 even have an effect on the voltage going into the 78M05?  My whole premise was that for some reason the 78M05 wasn't getting enough voltage to make the ICs work.  Since there are relatively few parts between the mains and the 78M05, I thought one of them must be faulty, but I haven't been able to find it yet.

[mikerj]

Yes, the 128v in my previous post was a typo.  Sorry about that 
That was a nice explanation of the circuit and it clears up a few things, but I'm still stumped as to why the coffee machine doesn't work.
Is it likely that the ULN2004 is fried seeing as how the triac and the resistor going into pin 12 were also blown?  Could the ULN2004 even have an effect on the voltage going into the 78M05?  My whole premise was that for some reason the 78M05 wasn't getting enough voltage to make the ICs work.  Since there are relatively few parts between the mains and the 78M05, I thought one of them must be faulty, but I haven't been able to find it yet.

A shorted component on the output side of the regulator will pull the voltage down on the input side (since the input is inherently current limited by C2).  In this case the voltage on the input side of the regulator would be roughly the dropout voltage of the regulator which is about 2v for a 78M05. 

Because your battery can deliver more current than the capacitive supply, it can raise the voltage on the output of the regulator enough to restore partial operation.  As someone else mentioned, the ULN2004 would be immediately suspect because it's connected to a triac controlling mains voltages that blew up.

[WannabeGeek]

Thanks for the help.  I guess the consensus here is that the ULN2004 is most likely faulty due to the bad solenoid valve which blew the triac, which blew the resistor going to its output.  I had a look at the datasheet and it seems to consist of a bunch of Darlington pairs with suppression diodes.  Since there's a common cathode at pin 9, I had a look at the output pins 10 to 16 (with respect to pin 9) with my multimeter set to diode check.  I would expect some sort of forward voltage on every output.  Strangely, the voltages are all over the place and I'm getting the same readings in either direction:
10 = .701
11 = .487
12 = OL
13 = .008
14 = .478
15 = .310
16 = .309
Of course, the most interesting part is that pin 12 (the one connected to the triad that blew) is OL.  I don't know why pin 13 is so low or why I'm getting readings in both directions, but I would assume there's no way pin 12 could be OL in both directions if that diode were working at all.  Would you say that proves that the ULN2004 is faulty?  I'm not even sure if I have the technology to replace that chip. I've replaced surface mount parts before, but never an SO-16 narrow package IC.  Do you have any experience with this?  There's also that 270 Ohm SMD resistor that I temporarily replaced with a through hole one.  Since I have no real replacement yet, I was wondering if I could just remove R37 on the board and transplant it into the R33 slot.  Since R37 is connected to the gate of the (theoretical) TY6, I'm thinking that won't have an effect on anything.  I'm assuming TY6 is installed in some other version of the board for different machines.  Does that make sense?

[mikerj]

The voltages you measure will depend on what else the pins are connected in the circuit, but it certainly sounds broken if you can't get any diode action on pin 12.

You ideally need a hot air pencil to remove these with minimal risk.  There are other methods though...

You can make a mask out of thin aluminium to protect components around the IC and then use a hot air gun(paint stripper).

Another option is to carefully cut through the pins with a scalpel/xacto knife or even a dremel with cutting disc and then individually remove the remains with solder braid or a solder sucker.  Extreme care required to prevent accidentally cutting through traces or ripping the pads off the PCB.

Soldering an SO package by hand is easier than you might expect providing you have a reasonably fine tipped iron.

[WannabeGeek]

Thanks for the tips.  I'll try the hot air gun method.  I'm assuming that cutting a small rectangle in a piece of aluminum foil will suffice.  I'd hate to desolder all those tiny caps.  After the parts that I ordered get delivered, I'll follow up with this thread and let you know whether I was successful or not.

[WannabeGeek]

Success!  I desoldered the ULN2004 using a clever method that is floating around youtube: 
which involves bending a copper wire to use as as a heat sink to heat all the pads on the IC at the same time.  It worked like a charm for me (see photo).  After waiting a long time for the IC to come in the mail, I soldered the new one in and the coffee machine was working again with the exception of the part that heats the milk.  After waiting for the new solenoid valve to arrive in the mail, I was finally able to install everything and the Nespresso machine works like new again.  Thanks everybody for the help!  Special thanks to Paul Moir and Mikerj for taking the time to help with the troubleshooting ;-)