ISA215-HYD module reverse engineering => involved in p1a15-error-condenser-charge-timeout

[iso14000]

Hi Folks

I would like to reverse this little board. I'm a skilled electronic engineer that bough a Ion (Peugeot) same as Imiev
I don't have yet the issue (crossing fingers).
The subject starts to be sensitive in France as cars are getting older. Here peoples are only implementing the "resistor patch" as a solution.
I personnaly don't like it because the root cause isn't clear and the situation can be worth as time goes by.
I also would like to thanks members from this forum for all the very enteresting posts : pictures and diagrams are very usefull.

let's start about the core subject :
this daughter board is a voltage isolator with a fixed gain =1, everybody here know that now
transfering an analogical value across a isolation fence isn't very straight forward. Classically we have only two media : light and magnetism (yes radio waves also )
using light is possible with very specific optocoupler but precision suffer with aging and generally must be calibrated (not good in automotive industry) , they are also forbidden in space application but it is out of focus.
So it is not a surprise to fin transformer onboard : One to transfer energy from the "process side" to the "measurment size" and the seconde one to transfert voltage.
after a glance on your picture I can tell that energy tranfert is performed by a push-pull SMPS topology , almost for sure.
transfering an analogical value thru a tranformer is a little bit more tricky because of the volt-second balance rule, that's why more circuitery is necessary , the solution used here seems to be a synchrone demodulation.
but I'm not sure....

My goal is to sketch the electrical diagram by analysis. What I need for this job is a good picture of each side of the board _without_ components to catch the hidden via.

Does someone having a defective board sleeping in a drawer and ready for the adventure?

best regards

 

[iso14000]

if you're interested about magnetic feedback you can find a white paper here : https://ieeexplore.ieee.org/document/4729638
I can send it to you by PM if you don't want to sign in

 

[iso14000]

another ressource I found , it is interesting because it explain the way a signal can be modulated to feed the transformer and how it is detected :
https://www.analog.com/en/resources...nchronous-detectors-facilitate-precision.html
see §

Practical Lock-In Implementations​

 

[kiev]

Placeholder for Index Post for reversing the hybrid board. This thread is not for troubleshooting or solving P1A15 issues--use the other thread for that.

INDEX:

1. Scope Traces--see posts #32-36


=====
Here is the Index Post from the P1A15 thread. The references to page numbers is incorrect due to the forum update that changed to 20 posts per page from 10, so be aware when trying to find posts by page.

https://myimiev.com/threads/no-ready-p1a15-error-condenser-charge-timeout.4225/post-38264

 

[kiev]

This is good that you have a dedicated thread to solving this board.

A tip i learned from member coulomb to use an Index in a technical thread in order to quickly find the pictures, schematics, datasheets, etc. The editing permissions are likely too restrictive but we can try.

There is a post with pictures in the P1A15 thread of both sides of the board, maybe those would work or we can take some fresh photos.

Some thoughts and questions to ponder:
What electrical or electronic component degrades but doesn't just fail completely?

This circuit still works somewhat but the output value is less than expected or specified by the original circuit design.

Mitsubishi issued a recall of the MCU for certain cars produced during specific months, and claimed it was due to some defective capacitor(s). The production date of these hybrid boards is marked with codes on the board, so they can trace VINs back to MCUs back to Hybrid boards. Is this a clue that capacitors on the Hybrid board are suspect?

 

[iso14000]

What electrical or electronic component degrades but doesn't just fail completely?

almost all of them. The aging is sometime also a subject : simply leaving a part in tube make it wear out after years .. for example if one day you order MSL 4 devices you will receive a "dry pack" and you have to fullfill a baking procedure before solder them on a board.

for our subject I can see several things :
-ceramic caps are very stable is used properly (working voltage and ripple) BUT they are very sensitive to vibration and mechanical constrain , because cracks can propagate inside and create shorts or open circuits
-resistors are also stable (when max power dissipation are fullfilled of course), less subject to cracks
-Integrated circuit are suspicious because they are complex and numerous issues can occur : the package can capture humidity , aluminium metallisation can migrate, etc etc , The IC can deviate slowly from the specs and can simply stop working...

Globally temperature is a starter or an activation for failure process , you will find that in MIL HDBK publication or FIDES .
TEmperature cycling is also the enemy because of delta dilatation coef in materials

-something as "simple" as a diode used the wrong way is also subject to defect .

here for what I can see , the power transfer from the "measurment side" to "isolated side" the dessigner used a push pull topology with a center tap for the transformer. But he didn't find usefull to have the real "forward" structure because he only used a diode to rectify the wave.

it as accepatable for very low power systems but the diode have to sustain large current pulses : it is a coulomb charge transfert.
It is cheap and effective because it generates easylly two voltages (+5V -5V ?) ....

only a clue,

 

[iso14000]

This circuit still works somewhat but the output value is less than expected or specified by the original circuit design.

got it. for sure!

 

[iso14000]

Mitsubishi issued a recall of the MCU for certain cars produced during specific months, and claimed it was due to some defective capacitor(s). The production date of these hybrid boards is marked with codes on the board, so they can trace VINs back to MCUs back to Hybrid boards. Is this a clue that hybrid capacitors are suspect?

it is a classical "interblocking" process in automotive industry.
I don't know what stands for "hybrid capacitor"

 

[kiev]

The timing capacitors (Ct, Cs) used for the Hybrid Board circuit--these are suspect.

From some testing done by Boo with applying hot air with heat gun and seeing output voltage go lower, then voltage goes back up as the temperature cools.

 

[iso14000]

ah ok ... ok ,I beilived you talked about a specific capacitor technology.
So... during the chase against a root cause , we must be very carefull...

What is very important is to stay focus on the REAL conclusion you have...

Using a heat gun isn't very demonstrative because... You cannot say precisely what was the temperature delivered by the tool ... 100°C? 50°C?... you don't know ...
IT is quite normal to see a shift because ALL componenent are sensitive to T.
resistor are at least 150ppm and caps also (may be more!)... needless to say that IC are also sensitive...
finally if you irradiate the board with an air witch is too hot , you probably are out of specs...
if I believe what I red from Nihon pulse , the isolator is specified from -40 to 85°C

and moreover it is the OVERALL temperature , heating only ONE component isn't correct on system that compensate by a way ore another , the drift .

it is essential to draw a little schematic, and extract the way it works to find the root cause... according to me of course.....

best regards

 

[iso14000]

to be perfectly clear TL494 is an old IC, used WW , but performance are nowadays outdated .

requency change with temperature(4)

ΔTA = MIN to MAX

10

Hz/kHz

means for industrial version (I) -40 to +85 , frequency drift is for a 100Khz : 1khz

if you had ppm from resistor involved in relaxation oscillator and capacitance drift ... it is worth

best regards

 

[iso14000]

Hi all

I've started to sketch the schematic with the data I can extract from your pictures.

It is clear for me now how this circuitery works :
on the right you have an OpAmp that manage a Gain and high impedance input , and also a low impedance output (as an Opamp should do)
the analog switches creates a square waves that feed the transformer , this sq wave amplitude is 2 times the voltage delivered by the Opamp : let's say Vamp
because it is alternatively switched on the transformer Volt.second balance is agreed, +Vamp, -Vamp, +Vamp etc...

I don't have any data to sketch the left side but I'm almost sure that the remain 4066 is used to synchroneously rectify the wave tranfered by the transformer. Added with a low pass filter module's output is proportionnal to the input ... in theory

by the way the little SOT23-5 IC is for sure an inverter....

 

[iso14000]

done.... almost

 

[iso14000]

need to know what's is the circled sot23, and where are connected the two resistance

 

[kiev]

i think the 3-pin device is a dual diode pair, that acts like a pair of zeners to bleed off [or clamp] any excess voltage on the "high" side [pin 2] of the "M23:" demod transformer secondary that feeds the input pins 1 and 4 of the 4066. That voltage passes straight thru to the output pins 2,3 that feed the ~3.5kHz low pass filter on the Output. 141nF, 275R,171nF

The "low" side of the transformer secondary feeds pins 8, 11, which passes straight thru to pins 9,10 to the Return stud 4 of the Hybrid board.

The Control pins of the 4066 are always active, held high by a voltage divider described in a post #19 and 32.

i'll have to find my notebook to get more details on the circuit but it has been a while since i looked at it.

My eyes are bad and i can't see tiny diagrams very well anymore...

 

[iso14000]

HI Kiev

If you get a look on the schematics I posted, you will see that wether a it is a zener diode , it is connected at the 4066 input. saying it should clamp voltage delivered by transformer , on only _one_ wave.
don't know why.
but for sure , only diodes and transistor are integrated in this type of package, and designer'd used very classic components , I cannot imagine them selecting an exotic one for a function.

after a small review I found some weak things on this design :
-the power supply topology , with coulomb transfert : a close control with a scope is necessary. diodes sustain large current pulse , it is less a concern for the caps
-the _very_ strange R//C on the power line , the guy how draw that should have been drunk! jeopardise intentionnaly an impedance like that is mad! it is not even a for EMC purpose because of the cap in //
any degradation in the power supply quality leads to bad behaviour.
-4066 are subject to large characteristics span and there are , on each wave, 4 switches in series, it will affect the frequency response module to module stability, hopefully we almost don't care because voltage to transfert's got a slow steep.
-even wether the module is embeded in EV, it will have to sustain vibrations , module is tighten by pins at each flange and large area is loaded by transfomers. It is not impossible to find a mechanical wave ringing in sucg a circumstance.
and what do we found exactly in the middle of the board? yes! two ceramics caps ! I don't like that ... at all! moreover for isolation creepage purpose they'd milled the board... precisely next to the caps... great!

what do you think?

 

[kiev]

Oh yeah that's a great observation about the board and layout.

The flexing mode of the board is aggravated by putting the two heavy components in the middle and milling slots that effectively reduce the stiffness. The ceramic capacitors are already notorious for thermal cracking, but now there is a physical vibration element to consider.

Years ago i saw this pcb vibration mode while repairing old cathode tube TVs, which caused cracked solder joints, and intermittent operation that the owner could 'fix' by slapping the side of the TV--until slapping no longer worked.

 

[iso14000]

old cathode tube TVs

so .. we probably got the same age ... 54 in a few days...


I'm waiting for a used ISA215 board from a french forum member.

if , by accident , somebody had one, here, I'll be very happy to have a glance on it ...

I saw so many defect in embeded electronics because of mechanic flexing mode (cars, planes ...)

best regards

 

[kiev]

Those 2 resistors (each are 8.6k) make a voltage divider across a switching supply and return. The mid point is fed to the control pins of the 4 bilateral switches of the 4066 (pins 5,6,12,13).

[Made edit to post 15 and 23 to add details of 4066 circuit]

 

[iso14000]

are you 100% sure? control pins of sw cannot be connected altogether... two by two yes.