Replacing FETs on Classic 150

[Hockey]

Hello,

Long time lurker, first time poster.

I picked up an 'as-is' Classic 150.   Connected a 24V battery and it started up fine.  Brought it home, reconnected the battery and connected my bench supply (32V, 10A) to the PV input, and voltage was dragged down to 1.2V, full 10A being drawn.  Ughhh, now I know why it was basically free.  For fun, I used a boost converter and brought input voltage up to 55V, and still the Classic dragged voltage down to sub 2V.   MNGP registered 0.0V. 

Further testing reveals 0 Ohms between PV- and PV+ terminals.   In searching around this forum, I found suggestions that some or all of the FETs are toast.   

Resistance between Source and Drain is 0.1 Ohms on all FETs when tested on board (which did not really tell me much)

The FETs are labeled as follows, observed resistance between Gate and Drain (tested on board):

Q6  20 Ohms
Q9  10K Ohms
Q7  0.1 Ohms
Q10  10K Ohms
Q8  20 Ohms
Q12   10 Ohms
Q11  0.1 Ohms
Q13  0.1 Ohms


I desoldered one FET (Q9) and tested it with my DMM in diode mode:
0.L across Drain-Source, 2.065V after charging the Gate (I think this one was fine?)

Before I order some new IFRB4115 mosfets, anything else I should think about/do?  I rarely get deep into PCBs, so wondering if there is something obvious I should check or consider before proceeding. 

Thanks!

[ClassicCrazy]

Check with Midnite for parts they may be able to sell you.
You may want to change out some of the capacitors and fans while you are at it. But maybe after you know you have it fixed via the mosfets.
I don't know the details of repairing this but others probably do .
Hope you get it running again !
Larry

[Hockey]

Thanks Larry,

 New fans are definitely in order when i get it going.

Question for those who have repaired these in the past.  If I  understand correctly,  the FETs are configured in 2 sets of 4 in parallel?  If correct, Which ones are grouped together?  They are a serious pain to get out,so i would like to snip them all, and replace just 1 FET in each set, then connect a low power PV input to see if problem solved.  If all good, I'll replace the other 6 FETs.

Another question...  what is the purpose of the blue insulating sleeve on the drain pin on every other FET?

Larry,
I learned a lot from your posts regarding Serial/Uart communication with MCUs.  I have a Classic 150 and an FM60 in use currently.  If I revive this second Classic, it will replace the FM60. I plan to turn on Follow Me.   I might have a solution to using serial comms with Follow Me.   Will keep you posted.

[ClassicCrazy]

Have you seen this project for the Classic ?
https://github.com/ClassicDIY/ModbusAdapter
Also look at the Classic to MQTT project.
Larry

[ClassicCrazy]

Did you look at this post on Classic repairs ?
https://midniteftp.com/forum/index.php?topic=5987.0
Jim had suggested to check the drivers that may be blown too.  Bob said you need to use a very large soldering iron at least 200 watts since there is so much heat sink on the copper board.
Larry

[boB]

Hockey, from your FET measurements, that is the way to tell !

I tend to measure from gate to source but of drain to source is shorted, which is what normally happens when they blow, you would see similar maeasurements.

0.1 Ohms is definitely shorted.

20 Ohms would mean that FET is possibly good and each FET would have 10 Ohms to Gate which would be why you would see twice 10 Ohms there.

10K Ohms is the gate to source pull-down resistor value so that FET is probably good.

Either way, the insides of the FET have turned to slag basically, as one old wise engineer used to call it.

[Hockey]

Thanks for the reply Bob.  I'll order some new FETs.  I saw some mention of checking gate drivers in another thread.  Is there anything else I should check? 

Using my hobby soldering irons, I needed 4 arms to get the one FET out.  One iron on top, one underneath...  with my son holding the board with one hand, and tugging with needle nose pliers with the other hand.  The scene looked like something out of an open heart surgery.

As an aside, when looking at the Infineon datasheets, I found FET IRFB4127 with similar specs, except breakdown was 200V vs 150V on the IRFB4115.  Max current is lower as well.  Makes me wonder if using the IRFB4127 would bring the unit closer to a Classic 200?  Probably many other things to consider and change (including reprogramming HyperVOC threshold).  So, not planning on doing it, just more curiosity.

Link to IRFB4127

[boB]

Yes, that FET is good but a 150V FET will have almost half the RdsOn of 17 milliOhms.

As far as removing and replacing FETs, what I do is use an iron with some good amount of power (150W ?) and a wide enough tip to straddle all 3 FET pins.  Then, after possibly using a bit of solder or flux, pull the part straight out.  Replace it the same way.  Then you don't have to clip any leads or clear any holes.

Hopefully you can put the Amo beads (the blue things) back on the high-side FET drain leads.

Not a huge deal if you break one or two of those but try not to put them back on the low-side FETs.

The drivers are fairly easy to get.  HCPL-3120 or the Fairchild (On-Semi now) equivalent. Cant remember the number right off.
The drivers do not always break.  Check the gate resistors too.  Alternating 5 and 10 Ohms I think



boB

[Hockey]

As best as I can tell, the Onsemi version of this octocoupler is FOD3120.   Screen snapshot attached from the datasheet.

Link to Onsemi FOD3120

As for testing the existing 3120s on board...

Bench supply voltage was set to 2.5V, and current limited to 14mA between pin 2 & pin 3. When the supply was switched on, Voltage dropped to around 1.5 (as expected). Current was 14mA,  and resistance on pin 6/7 dropped dramatically.

Not the most robust test, but it was the best I could do with:
1. My limited skill, knowledge, experience and test gear.
2. Having to test the chips on board which makes me suspicious of any results given all the other on board components that may affect results.

I found 4 3120s, all of which  exhibited results that makes me, with some skepticism, give them a clean bill of health.

FedEX with FETs from digikey arrives tomorrow.

Will update status soon.

[boB]

FOD3120  is correct.

There are a few manufacturers of that part I seem to remember.  Toshiba was another one that we used.

I hope that the iron works for you.

boB

[Hockey]

4 FETs replaced:  Q9, Q7, Q11, Q13.

After replacing these 4, PV+/PV- impedance = 1MegaOhm.   A good sign!

Connected a 12v Battery, set my bench supply to 32V, 4A and connected it to  PV terminals.  Success!   Photo attached.

Will give it a better workout tomorrow.

I'm going to bed.  It's 2:30am. 

[boB]

4 FETs replaced:  Q9, Q7, Q11, Q13.

After replacing these 4, PV+/PV- impedance = 1MegaOhm.   A good sign!

Connected a 12v Battery, set my bench supply to 32V, 4A and connected it to  PV terminals.  Success!   Photo attached.

Will give it a better workout tomorrow.

I'm going to bed.  It's 2:30am. 

This is great news !

You might get the general temperature of the FETs to make sure they are all getting similarly warm ?

Every-Other FET is a High-Low-High-Low type of thing so there may be alternating differences in temperature.

One of those pointed thermal sensors work for that if you have one in your medicine cabinet.
boB

[Hockey]

Thank you very much for the hand-holding.  It wouldn't have happened without.

Unless advised otherwise, I am re-using the thermal FET pad.  It was pliable with no visible damage or dryness.
I was also able to re-use the blue beads on two of the the drain leads.

There is only so far I can push the testing at the moment.  I boosted my 32V bench supply to various higher voltages but was only getting ~280W at best.   Tomorrow, under close supervision, I will connect it to one of my arrays. It is supposed to be a sunny day.   I have an IR temperature sensor, so will check temps periodically.

Now...  where the heck is my torque screwdriver for the 18 in-lbs required on the FET Clamp.

[Vic]

Hockey,  GREAT WORK!!

Vic

[boB]

Thank you very much for the hand-holding.  It wouldn't have happened without.

Unless advised otherwise, I am re-using the thermal FET pad.  It was pliable with no visible damage or dryness.
I was also able to re-use the blue beads on two of the the drain leads.

There is only so far I can push the testing at the moment.  I boosted my 32V bench supply to various higher voltages but was only getting ~280W at best.   Tomorrow, under close supervision, I will connect it to one of my arrays. It is supposed to be a sunny day.   I have an IR temperature sensor, so will check temps periodically.

Now...  where the heck is my torque screwdriver for the 18 in-lbs required on the FET Clamp.

I was going to say use a torque wrench if you have one.

And use the tried and true star pattern method.  Well, as best as you can because it isn't a car tire.

Good news on being able to reuse the amo beads and the insulator.

We could probably have provided those if they were not in decent shape I would think.

Glad the FET drivers and gate resistors didn't need replacing either.

Good job !

boB