Repair of 2413S PLM When the Power Supply Fails

[jtara92101]

My parts are to arrive today!

 

(Just noticed, though, that Mouser ships "signature required". 8 bucks to ship, AND they want a signature on a small order...)

 

classicmac, I see your parts list varies from the "canonical" parts list. Can you explain the differences?

 

1. You have two extra 10uF/16V. What capacitors do those replace? Or just spares?

 

2. I see you preserved the original values of C7/C13 (10uF) vs 100uF in the "canonical" replacement list.

 

(2) Got me thinking. Can somebody explain the reason for the "upgrade" from 10uF to 100uF? I looked at the schematic, and see these are part of a pi filter on the (power? Or is this detected signal?) line to the riser board. What is the reason for the radical change of value for these caps? Assume not a typo, as I'd hope somebody would have noticed by now if so.

 

Could not find any (useful) information on the LNK345GN. A handful of sites in Chinese that, when translated... suggest that it is some kind of IC. (Which I already knew!)

Edited February 20, 2017 by jtara92101

[Brian H]

Not a typo. All of my hardware 1 versions have the 10uf/35V capacitors in C7 and C13.

Smartlabs changed them to 100uf/35V starting with hardware revision 2 and above.

Maybe to beef up the main power supply or they would last longer.

As it is the unregulated 12 volts {usually around 18.5-19V} DC that runs everything.

 

The original 1.0 had rework on it to add the second capacitor and coil. To add the pi filter. All three hanging off the PCB in the air and a fly wire to the back of the PCB.

 

The PC run you traced to the serial daughter board powers it. There is a small 5 volt regulator on the serial daughter board. Run by the unregulated 12 volts from the main board.

 

Both my 1.? and 2.? hardware have the same brand and series capacitors in them.

Attached LNK345GN Data Sheet,

lnk353_354.pdf

Edited February 20, 2017 by Brian H

[jtara92101]

Not a typo. All of my hardware 1 versions have the 10uf/35V capacitors in C7 and C13

Smartlabs changed them to 100uf/35V starting with hardware revision 2 and above.

Maybe to beef up the main power supply or they would last longer.

As it is the unregulated 12 volts {usually around 18.5-19V} DC that runs everything.

 

 

And do we know that the inductor value (filter with C7/C13) is the same from V1 to V2?

 

Got the parts, BTW. The UPS guy didn't actually require signature, left it at my door.

[Brian H]

Off hand I do not know if the inductor value was changed.

If I can read the values between my revisions I will post back.

[apostolakisl]

I assumed they'd make other changes than just the caps to make a very high reliability unit. If they didn't get the power supply right, there's probably other things they didn't get right as well. So I just sort assumed somewhere around 150% of the current price.

Based on the fact that this thread is 2.5 years old and I haven't seen anyone have a failure after replacing the caps, I'm thinking the caps are pretty much it as far as problem parts?

 

I have a repaired board that is running now for about a year or so.

 

Only time will tell for sure.  Not sure how long you would say these should last before feeling like "OK, that's pretty much all you need".  100 years is no doubt over-kill, but what about 10?  20?  I'm going to say 10 would put it at a number that I would say is plenty.  There is a reasonable chance that after 10 years, you'll be ready to move on to new technology.  Maybe not.  

[jtara92101]

Please add +1 to the "successfully repaired" column!

 

Was able to query entire system, no longer getting "blinky" on KPLs that are configured to run a program, etc. etc. (I was having those problems prior to the PLM dying completely.)

 

I found the best way to remove is to use solder wick to first remove the bulk of the solder, then position the board so you can grab the cap with pliers and wiggle and pull as you heat the pads from the back. Easy does it, or you might pull a pad off the board. First one I tried to do it all with solder wick, and it was very time-consuming. If you get the cap pulled up enough, it would be helpful to clip the leads once you can get at them.

 

I do not like/trust manual "solder suckers". Too many past incidences where they pulled the pad off of the board! If you have access to a professional desoldering station, that would be best, though. The manual spring-loaded solder suckers are too violent. Plus that hitting-yourself-in-the-chin thing....

 

The Hakko soldering station is an indulgence - they are the best! But you can get similar knock-offs for much less. Tips are expensive, as they not just a blob of machined metal, but contain both a matched heating element and temperature sensor. The controller maintains tip temperature very accurately! Have never needed an alternative or replacement tip, so not worried about tip cost.

 

In any case, a temperature-controlled soldering station with sensor in the tip will help prevent damage to pads and through-holes.

 

I could have sworn I had a can of flux remover. Couldn't find it - used Goof Off and Q-tips. Original soldering was sloppy, and they didn't remove flux. It was a mix of machine and hand soldering, or maybe they did the reflow work on surface mount components with a heat gun (they have tiny heat guns just for this). The through-hold soldering was obviously done by hand, and sloppily, but no horrible flaws, so did not "correct" any of the other joints.

 

The fit of some of the capacitors is tight, but you can wiggle things around to get them to fit. There were no lead-spacing mismatches.

 

 

Yes, my "workbench" is my kitchen counter, and, yes, that is a pizza peel. I have no idea where the plastic jaws for the PanaVise are - I was careful!

 

I used the Kemet ESX106M400AH4AA 10uF/400V which is back in stock.

 

I used UTT1C100MDD1TP as substitute for discontinued UTS10100MDD instead of USV that some others have used, as UTT is recommended replacement for UTS line.

 

Some tips:

 

- Use a magnifier headlamp if you have one, to help avoid solder bridges, accidentally wicking solder from adjacent pads, etc.

 

- These are plated through-holes, always make sure to heat the wire and pad both and then apply solder to the pad. It will wick down into the plated hole.

 

- Clean both sides of pads with solder wick after removal. Place down wick, soldering iron on top of wick.

 

- Wet (with solder, silly!) and clean the tip OFTEN. ESPECIALLY when desoldering/using solder wick.

 

- The "brillo pad" is the best for cleaning the tip! Please no wet sponges.

 

Tomorrow I will attempt a LampLinc and an Access Point. I think I am lacking a zero-crossing cap recommended here, will need to make a trip to Fry's.

Edited February 22, 2017 by jtara92101

[MarkB]

Please add +1 to the "successfully repaired" column!

Thanks so much for posting this mine is back up and running.

[jtara92101]

A wonked-out PLM can corrupt link tables in device controllers. A depressing syndrome I've witnessed first-hand. There also was one wretched iteration of switchlinc (V.35?) firmware that had the same magical capability to corrupt unrelated controllers at random. And corrupt link tables anywhere naturally open the door to all kinds of irrational mischief anywhere else. (Attention SH: how about a damn CHECKSUM?)

 

 

Interesting.

 

I have a bunch of "flakey" v.35 SLs that I've never been able to pin-down the exact problem with. All I know is that they are toxic to the entire network. It will work for a couple days, and then things start going south. As well, some KPLs built around the same time. Newer or older ones work fine.

[Winter Dragoness]

I've got a 2413U manufactured in late 2012 with firmware v1.A.  I bought it April 7, 2013 so it lasted 4 years before failing.  For a few weeks I found it had stopped working periodically (usually after a house power failure) but power cycling fixed it.  Then its LED went dark and would not turn on again.  After having it unplugged for awhile I opened the case and found C7 hit 125F in my FLIROne camera after being plugged in only a couple minutes.  C7 and C13 were still the older 10uF, 35V models.

 

I ordered new caps from Mouser but found they ship from Texas instead of California so decided to temporarily use a 50V 10uF cap from a capacitor kit I have.  The new cap runs cool in the IR camera and the PLM worked fine 6 days till I had time to replace all caps with Mouser versions.

 

The original, failing C7 cap measures 3.1uF on a UT58D but 9.2uF on a UT210E meter.  Both meters measure a new 10uF cap just above 10uF.  I wonder what that means?  Maybe the cap has slowed in how fast it builds a charge and one meter is testing too quickly?

 

I'd say a temperature-controlled soldering station like the Hakko FX888D or similar is more of a necessity than an indulgence.  I've tried various soldering guns in the past and always had enormous frustration and poor results.  They might sit there not melting or barely melting solder, or they might briefly melt it, then solidify as heat spreads out through components and the iron doesn't increase power to keep the temperature up.  I tried following various guides and advice online but only the Hakko actually works.  A desoldering braid without Hakko is the worst: the braid pulls heat down its copper length and I could almost never get it to melt the target solder into the braids and pull it away while still wet.

 

I'm not sure why jtara92101 says not to use a wet sponge, but Hakko comes with a stand to hold the iron and the stand holds a special wet sponge with slits in it.  Wipe the iron through either slit and it keeps it nice and silver as you work so heat isn't blocked by oxidation on the tip.  Bits of solder are wiped into the slit and fall below the sponge.  Works great.  The stand also holds a brassy brillo pad but I never feel it cleans the tip as well.  Rarely, a bit of solder doesn't seem to want to come off with the sponge so I knock it off with the brillo.  I also wipe the tip on the wet sponge after turning off the iron to cool it a bit faster so it doesn't oxidize so much as it cools.  It's the heat that makes it oxidize (get darker) so quickly.

 

3-29-17: Replaced all caps on main board with the following Mouser parts:

C7 and C13:   661-EKZM500D101MHB5D   (EKZM500ETD101MHB5D)   50volts 100uF

C8:   647-USV1C100MFD   (USV1C100MFD)   16volts 10uF

C11:   647-UTT1E101MPD   (UTT1E101MPD)   25volts 100uF

C3:   647-UPW2G100MHD1TO   (UPW2G100MHD1TO)   400volts 10UF

 

These caps have all been mentioned in previous posts but it doesn't hurt to re-post their latest model numbers as some have changed since the original post or been mistyped in various places.  First number is Mouser part number, second number in parenthesis is manufacturer part number which might be found on other sites like Digikey.

 

Note the new C3 is huge and its base needs to be pushed down to within 2mm of the circuit board or its top will prevent the cover from closing.  All the caps are larger than the originals and may fit a little funny or wedge against things next to them.

 

I used a desoldering braid dipped in a tiny bit of paste flux to remove the bulk of solder on wires of old caps.  Hold braid against solder, then hold iron against braid to draw solder into braid and pull away.  Cut off end of braid when it has too much solder in it.

 

Heated one leg of capacitor and pulled out a few mm.  Heated other leg and pulled out a few mm.  Repeat until capacitor is removed.

 

Used a Greenlee 1700 Desoldering Tool With Standard Tip to completely clear holes of solder after removing each capacitor.  Hold the tool against hole on one side, then touch iron to hole on other side until solder goes shiny, then spring tool to suck all solder out of hole.  Repeat if needed.  New caps slid easily into cleared holes.

 

Be sure to put the shorter leg of the new cap in the round hole.  The cap will have a stripe with zeros (or sometimes minus symbols) on the side with the shorter leg.  I found it convenient to remember to put the side with zeros in the round hole because the round hole looks like a zero.

 

When finished, I used rubbing alcohol on a toothbrush to clean off some whiteish residue and any flux I'd left around solder points.  I've found in the past that flux is slightly conductive and can make circuits malfunction so I made doubly sure to scrape it off from between contacts.

Edited March 30, 2017 by cdragon

[Teken]

Best practices for those of us in the board level repair business has always been to place a glob of solder on the tip.

 

This is done mainly for long term storage and tip protection. Those who live in the rust belt know heat, salt, and humid air will quickly break down metals.

[fmk2]

Followed the instructions on this thread to replace the capacitors in my 2413S today, but no joy.  The LED doesn't light up at all.  I now suspect that there was more than just a power supply failure.

 

If it is indeed a catastrophic failure, it's probably the result of our latest power failure.  A nearby transformer failed and took out all the power for several blocks, but the worst part came when they turned the power back on; it didn't immediately come up to 117V, but instead ramped up slowly.  That can be very hard on electronics, not to mention things like refrigerators and freezers.  Fortunately, the 2413S appears to be the only thing in the house that failed.

 

I'll poke around a bit to see if I can verify that the 5V regulator is working, but beyond that I'm not sure what I can do without a schematic as a reference.  It may be time for a new unit.  (sigh)

[Brian H]

Look at post #282 of this thread. There is a schematic of the power supply portion of the PLM in that message.

 

I would double check for things like small solder bridges or a broken PCB run.

I damaged a run in one of my repairs. By tearing out a plated through hole and small piece of the run attached to it.

Edited April 1, 2017 by Brian H

[danu1964]

Followed the instructions on this thread to replace the capacitors in my 2413S today, but no joy.  The LED doesn't light up at all.  I now suspect that there was more than just a power supply failure.

 

If it is indeed a catastrophic failure, it's probably the result of our latest power failure.  A nearby transformer failed and took out all the power for several blocks, but the worst part came when they turned the power back on; it didn't immediately come up to 117V, but instead ramped up slowly.  That can be very hard on electronics, not to mention things like refrigerators and freezers.  Fortunately, the 2413S appears to be the only thing in the house that failed.

 

I'll poke around a bit to see if I can verify that the 5V regulator is working, but beyond that I'm not sure what I can do without a schematic as a reference.  It may be time for a new unit.  (sigh)

There is a possibility the power failure/power up condition you have described may have taken out the off-line switcher IC.  Let me know if you need one as I have a few on hand.

[fmk2]

There is a possibility the power failure/power up condition you have described may have taken out the off-line switcher IC.  Let me know if you need one as I have a few on hand.

It looks like the switcher is ok.  I can see a bit over 21V across C13, which -- according to the post by Breezyken in post 282 -- is about right.  But there's no voltage coming out of the 5V regulator, so it is likely toast.

 

I've also noticed that the unit gets quite hot in the area of the switcher IC.  I know it's having to dissipate power to get the voltage down to 21V, but it's hotter than I would have expected.  Is that normal, or is that possibly an indication that I have more than one issue to deal with?

[Breezyken]

21 volts on C7 or C13 is within the correct range.  It's likely that you have a bad U3 5 volt regulator or there is excessive load on the 5 volt line.  That might also explain the excessive heat on the line switcher, but in that case I would expect the output there to be lower than 21 volts.  You should also check the regulator output on the daughter board.

It's been almost three months and my repaired PLM is still working.

[fmk2]

21 volts on C7 or C13 is within the correct range.  It's likely that you have a bad U3 5 volt regulator or there is excessive load on the 5 volt line.  That might also explain the excessive heat on the line switcher, but in that case I would expect the output there to be lower than 21 volts.  You should also check the regulator output on the daughter board.

It's been almost three months and my repaired PLM is still working.

Thanks for the quick reply.

 

I did those voltage checks on the main board without the daughter board installed, but I had checked the daughter board's regulator earlier -- before finding you schematic and voltage measurements.  It appears to be working as expected.

 

Depending on how the 5V regulator on the main board failed, it could be responsible for the excessive heat.  I'll see if I can find a replacement for it... hopefully that's what is needed.

[fmk2]

Never mind...

 

The excessive heat really nagged at me, so I decided to do one more check of the connections on the main board to see if I could find the cause.  Sure enough, there was the tiniest solder bridge -- a fraction of a hair's width -- across C8... which of course shorted the output of the 5V regulator.  Ugh.

 

I cleaned up that bridge, and the unit fired right up.

 

I've been building and working on electronics since I built my first Heathkit in the late 70's so you'd think I could avoid something as simple as a solder bridge by now, but apparently not.  I blame my aging eyes for not catching it sooner:  I didn't need bright lighting and high powered magnifiers to see such things in those days!  

 

Thanks to all that have contributed to this thread, and especially to those that responded when I ran into what turned out to be self-inflicted problems.

[Breezyken]

Congratulations on reviving another PLM.  I hear you with the eyes and back when we were building Heathkits the parts were nowhere near as small as they are today!

[Brian H]

Glad you found the problem. My eyes are not what they use to be either. My large magnifying glass gets more use everyday

 

Those 5 volt regulators can usually take a beating. I believe they have a shorted output shut down to a safe level built in.

[Brian H]

I see in the 2413S Reviews Tab.

 

Smarthome's reply to failures.

Indicates that they have identified a component failing in any revision below 2.3 and that as of January 2017 revision 2.3 should have the revised component or components  in it. They also tried to gloss over the failures as only in some situations.

Edited April 4, 2017 by Brian H

[kioolt]

I see in the 2413S Reviews Tab.

 

Smarthome's reply to failures.

Indicates that they have identified a component failing in any revision below 2.3 and that as of January 2017 revision 2.3 should have the revised component or components  in it. They also tried to gloss over the failures as only in some situations.

Could you please be a little more specific on where you saw this.  I'd like to read it.

[Teken]

Could you please be a little more specific on where you saw this.  I'd like to read it.

 

http://www.smarthome.com/powerlinc-modem-insteon-2413s-serial-interface-dual-band.html#reviews

[Brian H]

Top of the review tab that Teken provided a link to.

Now it does not address what part or parts they changed.

I suspect power supply but it could be something else.

[larryllix]

I think the part that couldnt take the heat was the upper management.

Maybe they replaced him?

[fmk2]

If anyone on this list has one of the post-January 2017 units, please crack that puppy open and let us know what's different.  As a minimum, it would be nice to know what values they used for the capacitors we've all had to replace (C3, C7, C8, C11 and C13).