MCC-152 problems again

[Donk]

Hi again,

I now have two more bad MCC-152s on my desk since I asked this question.  This is the same machine as the previous post.  I seem to be having a similar issue to what is described here.  The i2c expander is getting red hot when W3 is set to 5v.  I did change to 3.3v which seemed to slow down the failure, but I may have already weakened the chip by overheating before changing to 3.3.

I currently use 3 inputs and 4 outputs.  The outputs are sent directly to a ULN2801.  Two of the inputs are connected to the open collector outputs of a LM393 comparator.  The other to an open collector opto-isolator.  The inputs are set as pull up, only internal pull ups are used.

I also notice an oddity when polling the inputs, but I can't determine any pattern causing it's occurrence, all the digital inputs read high.  This is physically impossible, a spare unused input tied low also reads high.  The only way to "clear" this is to toggle an output.  I'm working around this now, but I am wondering is it somehow related.  I also have an identical system, except using a Pi 4, which hasn't exhibited the problem.

 

Thanks 

[Fausto]

Hello @Donk.

Were you using a RPi 5 module with the two MCC 152 daqhats?

Does your RPi 4 module and MCC 152 have the same external components and wiring as the RPi 5?

On the RPi 4, is the MCC 152's W3 jumper set to 3.3V or 5V? 

What are the voltage levels connected to the digital input lines, on either system?

 

Regards,

Fausto

[Donk]

No, I'm having problems with RPi 3 B+.  

Both are in identical systems.  The only reason some are RPi 3 is the shortage a few years back. We had spare unused 3's for a different project.

The RPi 4 is set to 5V, the chip is very hot, but is still working.  The 3's were setup the same way, but I saw Display Names post and decided to see if his suggestions would help.  So I changed to 3.3 (seeing the open collector outputs won't care either way)

Digital in's are open collector, internal pull up (internal to MCC-152).

 

Thanks

donk

[Nick Wright]

It is likely that the IO interface chip has been damaged, most likely due to some form of over or under voltage spike similar to what was reported here.  A system schematic would be helpful for further analysis to avoid future damage.  Are you driving relays with the ULN2801

[Donk]

Thanks @Nick Wright.

The ULN2801 outputs are driving solenoids.  Unfortunately, the schematic terminates at a connector without showing the coils.  However I'm just using the built in suppression diodes from the ULN2801.  The coils are  24VDC. 

I'm almost done for the day, but if I can find time, would a schematic without the connectors, and showing the coils help?  I could probably have one to you on Monday.

[Nick Wright]

Just now, Donk said:

Thanks @Nick Wright.

The ULN2801 outputs are driving solenoids.  Unfortunately, the schematic terminates at a connector without showing the coils.  However I'm just using the built in suppression diodes from the ULN2801.  The coils are  24VDC. 

I'm almost done for the day, but if I can find time, would a schematic without the connectors, and showing the coils help?  I could probably have one to you on Monday.

Yes.  We mainly need to check that the freewheeling diode output from the ULN2801 is connected back to the coil supply so you don't get large spikes when turning the coils off.  You can also check the MCC 152 outputs with a scope while switching the solenoids to look for spikes.

[Donk]

Ok.  I only have one working MCC 152 in house and it's currently being used.  I should be able to check some time Monday.  I should have a schematic for you by then also.

 

thanks for your help.

 

[Donk]

On 5/17/2024 at 12:28 PM, Donk said:

Ok.  I only have one working MCC 152 in house and it's currently being used.  I should be able to check some time Monday.  I should have a schematic for you by then also.

 

thanks for your help.

 

I haven't been able to pry the only working unit from production yet.  Here's the drawing.  There are 4.7K resistors as pull ups, but they aren't populated, as denoted by the X.  

 

AFS_DIOa.pdf

Edited May 21 by Donk
wrong attachment

[Nick Wright]

@Donk Is GND1 a different ground than the other ground symbols in the schematic?  is it at the same potential or separated?

[Donk]

@Nick Wright Same potential.  I missed hiding the text.  

 

 

[Nick Wright]

2 minutes ago, Donk said:

@Nick Wright Same potential.  I missed hiding the text.  

 

 

Some things seem odd in the design unless I misunderstand something:

• LED D1 is always off (has GND on both sides)
• The ULN2801A O5 is tied to GND
• There is no current limiting component for the U2 diode (unless the capacitive pushbutton switch limits current)

However, I do not see anything that would damage the MCC 152 if the connections on J7 are made correctly to the 152.  What is the pin mapping from J7 to the 152?  If you are using the 1x10 header footprint on the 152, the pinout is:

[Donk]

@Nick Wright  I'm sorry, I hastily modified the current drawing and made some errors.   The enclosed capacitive touch switch, has it's output internally limited.

Somehow when moving chunks of the drawing O5 (U3.14) got connected to ground.  I've fixed the drawing as attached. 

DIO1 -> J7.2

DIO2 -> J7.3

DIO3 -> J7.4

DIO4 -> NC

DIO5 -> J7.7

DIO6 -> J7.8

DIO 7 -> J7.9

 

AFS_DIOb.pdf

[Nick Wright]

I think your best course of action is probing the system to look for spikes or unexpected voltages.  There are no obvious issues in the schematic, but I would focus on the high voltage portion around U3 and the switch.  You may find unexpected behavior from things like the leakage current through the clamp diode in the ULN2081A on O5.