Author Topic: [FIXED] Venerable AIM TTI PL303QMD-P needs help - Overvoltage trip error  (Read 6134 times)

0 Members and 1 Guest are viewing this topic.

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Dang, too late.
My guess was that OUP is too wide as output.

In case you have enough courage you can try creating an OCP trip using the functioning side of the machine.

The feedback can be on/off style controller pins or analog inputs and then controller decides what is bad.
Older version has ILIMIT as a single pin, but can't say what it actually is.
Triple model 5V has OVP trip signal, others have OPON.

Swapping boards and using brute force measuring tactics you can "easily" narrow possible problem pins.
But schematics would be better.

For PJ6 my guess is that it doesn't help.
But who knows, maybe it's constantly outputting some console info.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Thank m k to stay with me in this madness.

After some emails, TTi support told me they believe/think the uP must be defective and that's it, they did not respond to my last email.... I let them go in peace...
They helped me a lot for free!

Despite that I think we have still cards to play, now that we know that it's an OVP problem we can dig into the right direction.
Please be aware, if you see a OVP malfunction in your device, I would say 90% the bad guy is the power board, my case is an unicorn IMHO.

According to my rev eng, there are 2 places where the output voltage (aka vsense+ and vsense-) comes into the control board:
1 - into the ADC
2 - into the VC generation analog part, basically to close the output voltage regulation feedback loop:

     Vset ----> Vc --> Output ----
              |                               |
              --------------------------

 - Vc is the control voltage
 - Vset is the target voltage coming from the uP
 Please note the Vset to Vc analog part is on the control board, the Vc to Output is on the power board.

Now, two facts:
a) I can't find any problem in the analog control board [CB] part
b) after powering up the unit, immediately the OVP is triggered on the side (master or slave does not matter) where the offending control board is installed.
Here below the startup sequence on the offending side

Green is the 5V rail
Blue is Vc
Yellow is Vset

this means to me:
1) Assuming the uP is still alive since I see "OUP tRIP" in the display, we need to check what it is going on between Vsenses lines before and after the ADC.
2) I need to compare the same startup sequence on the good side with output off, to see if Vc going to the roof for about 85ms after startup is normal.
I think this is not a problem... Vset stay at zero.... so...

[oh I did not check if the output is spiking at startup as well... but since Vset is 0 all the time, I do not expect any movement there.]

Before the ADC this is what is happening:


where A17 and A16 are the two ADC inputs, simulated in QSpice we get:


it's too late now, I need to sleep.... more to come.

« Last Edit: August 02, 2024, 03:53:04 am by Zucca »
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Chopped startup of PID controller.
First D goes through the roof, then 'I' start kicking in and finally everything is steady again.

Maybe this startup shouldn't go through the feedback.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Back to the ADC section, this is the correct output sense circuit. The previous one is not good.



basically the Vsense+ is connected to GND (when sense is set to local of course).

And this is the output voltages to the ADC (sweep 0V to 35V output voltage):


And this make sense, because the ADC input range is 0V-2.5V, in my previous one input was above 2.5V so it would had not work well.
In my previous measurements, I confirmed those ADC inputs lines are sitting at 2.4V when the outputs are disabled, which again matches the simulation.
« Last Edit: August 07, 2024, 05:15:51 pm by Zucca »
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Startup voltages, yellow output, green 5V rail







since the output doesn't move, again either the ADC or the uP (unlikely) are bad.

Next step sniff the ADC SPI

« Last Edit: August 09, 2024, 01:35:40 pm by Zucca »
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
The chip has 8 inputs, have you tried comparing them?
It probably has the same stuff for left and right.

It's also possible that OVP is the first message and so misleading.
Means that only if there's no OVP other messages are appearing.
A bit far fetched, surely stuff like missing ADC is messaged before it.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Well I did not compare the good/bad side ADC inputs because the ADC inputs on the bad side make sense coming from the Sense lines...
More over I dig into the analog part and I coudn't find any error.

Tonight I compared at the SPIs good ADC and bad ADC, no major differences just looking at the signals/edge... still have to decode the traffic.

One more interesting find, if I swap the control boards again, the good one (was slave) at the master place report a FAIL 4 and won't turn on the output.
Once I reverted it, the error disappears and at least one side (slave) is working as expected.

Did not see a fail 4 on the bad one at the wrong place (master in slave)... mmmm hope the uP is not toasted.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Maybe you got a bad connection.
I don't like those IDC style connectors, too many times pin coatings are sub-standard.
And ribbon cable side knife connection is what ever after few twists.

Are there many possible serial bus connections out from MCU?
Maybe you can map most of them and find out how the thing is initialized.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
thx m k, I am too tired to work on it tonight.

Here the startup video,

https://www.dropbox.com/scl/fi/1rs3myawu33dif7xj68vr/20240811_205119.mp4?rlkey=wenaqjdetib6d5u4kn7ly6tae&st=wsuwg4s9&dl=0

uP is booting, then goes in error mode ovp immediately.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
SPI madness...





Now I just have to do some homework...

FUUUU I was not expecting such a hard job, it is just a damm power supply!
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Dang, I though it has only one MCU.

Well, then you just compare 8 ADC inputs to other 8 ADC inputs.
Very simple.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
ADS1241 high nibble of opcode can be only 0, 1, 5 or F.

So 5 (1+4) bytes is fine, but 48 or 40 can't be any opcode.

3 times the same set indicates it's close.
And 1st byte is recognized as an opcode, or is it.

Answer could be one from negative limit of 24bit signed value, if MSB is right.
But then opcode must be 1.

     4       8      4
0100 1000 0100 0000 0000 0000 0000 0000 0000 0000

MSB still right and shifting 2 bits would be xx 01.
But then 1st byte doesn't fit.

E,
obviously not -2.

E2,
boot has 8 frames of right CC LED, later left CC LED on is a bit shorter.
May that CC LED can be a trigger for checking the fault test moment.
« Last Edit: August 14, 2024, 04:59:50 pm by m k »
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 
The following users thanked this post: Zucca

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Thanks m K! So cool

I will dig into it the next days... I hope there are no noise that are causing some bit flipping... I have also a Picoscope that for decoding SPI maybe it is better...
Crucial will be the startup phase... what I posted above was just a proof dirty try/proof of concept...

Tonight I compared the good and bad ADC inputs regarding the Vsense lines (the one causing the OVP)

Good one
PIN   
16 Vsense- 2.402V
15 Vsense+ 2.402V
9 REF+ 2.510V

Bad one
PIN   
16 Vsense- 2.402V
15 Vsense+ 2.402V
9 REF+ 2.511V

as expected the analog part works and matchs my Spice sim.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Was your measuring GND far away?
Maybe the chip has some problems with it.

If those boot LEDs are what they are marked to be it's like master side has equally bad voltage and current.
When booting two slave side LEDs come on together, but master side has a delay.
Delay can mean that boot is waiting that tested voltage is settling down, that never happens, and then does the same wait for current.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
ADS1241 high nibble of opcode can be only 0, 1, 5 or F.

m k is right!

I had a wrong MSB/LSB setting in the previous screenshots.

Let me dig a little bit further and I will report my findings soon!

Thanks again m k!
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Here we go... here what I could sniff on the SPI on the GOOD working board:

Sent:
12 02 00 00 00
Received:
00 00 80 00 01

the above repeats a lot

Sporadically I get

Sent:
51 00 38
OR Sent:
51 00 58
Sent:
51 00 48
OR Sent:
51 00 28

always Received:
FF FF FF


followed by

Sent:
01 FF FF FF FF
here I get multiple possible responses

Received A:
FF 00 00 25
Received B:
00 00 15 C1

and so on, attached the sniffed traffic in .csv files.

Now dive into the ADC IC07 ADS1241 datasheet to understand what it is going on.

cool.

« Last Edit: August 28, 2024, 03:05:15 am by Zucca »
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
02 after 12 is a counter, same with 00 after 51.

FF can be a waiting thingy, but after 01 (read data) should be only 24 bits.
So last received is a byte too long, maybe it's just a hiccup.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
sniffed the SPI from the bad uC board...



There are a lot more FF FF spitted out from the ADC to the uC.

Since I measured the inputs, and they are okay... I think the next reasonable step is to swap the two ADCs from the two good/bad board and see what happens.

Such a unicorn failure...

PS: the Picoscope for decoding serial is such a champion.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
New ADS1241E/1K‎ ordered, finger crossed.
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
So single ended input channel 3 and differential 6-7 are open or pulled up.

Channel 5 is sending 32 bits, maybe nothing, but it's also positive once.
Did you check those inputs with a scope?

But anyway, ADC change is pretty affirmative thing.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 

Offline ZuccaTopic starter

  • Supporter
  • ****
  • Posts: 4611
  • Country: it
  • EE meid in Itali
Re: Venerable AIM TTI PL303QMD-P needs help - [was Output] Overvoltage trip error
« Reply #45 on: September 04, 2024, 03:23:06 am »
DISCLAIMER: I never reworked a SSOP28.

I did it without microscope.... long story.... you don't wanna know...

Here the awful result: not centered at all.



I did cleaned the pads, eyeballed the chip and soldered it down
Next time I will presolder the pads and then let the hot air do its magic.
Yes I used tacky flux, no worries.

Then I give it a shot and FUUUU YES! No more OVP Trip!

Now I have to fully test my repaired  PL303QMD.

Finally I just want to thank m k and Gertjan for helping me in this madness.

PS: C18, C17 and C15 will be installed once I am sure that ADS1241 is installed correctly..

More to come! FW Upgrade? Calibration? yummm...

« Last Edit: September 04, 2024, 03:36:34 am by Zucca »
Can't know what you don't love. St. Augustine
Can't love what you don't know. Zucca
 

Offline mk_

  • Regular Contributor
  • *
  • Posts: 243
  • Country: at
Re: Venerable AIM TTI PL303QMD-P needs help - [was Output] Overvoltage trip error
« Reply #46 on: September 04, 2024, 05:27:12 am »
DISCLAIMER: I never reworked a SSOP28.
 
Here the awful result: not centered at all.
 

Well done... and who cares about not perfect alligned pads as long as they don`t short to others? The lid hides those little lapsi
 

Offline ArdWar

  • Frequent Contributor
  • **
  • Posts: 721
  • Country: sc
Re: [FIXED] Venerable AIM TTI PL303QMD-P needs help - Overvoltage trip error
« Reply #47 on: September 04, 2024, 05:44:28 am »
No bridges, no overhanging pins, no one notices if not for the shiny lead.
 

Offline Gertjan

  • Regular Contributor
  • *
  • Posts: 147
  • Country: nl
Re: [FIXED] Venerable AIM TTI PL303QMD-P needs help - Overvoltage trip error
« Reply #48 on: September 04, 2024, 07:32:05 am »
Hi Zucca,

Well done!  :-+ :-+
Congratulations with a well deserved successful repair! You did the extra mile, and it paid off.  :)

Quote
Next time I will presolder the pads and then let the hot air do its magic.
I would not do that. Hot air is splendid for small SMD IC packages, like TQFN. You heat the whole package, and it will float in place when all solder has melted.

In my experience with a (relative) larger package, like this ADS1241, it is best to solder it just like you did:
- Clean the pads
- Add some flux
- Position the new IC, and solder two opposite corner pads.
  (After tacking the first corner pad, check position of the IC, and adjust if necessary, then solder opposite corner pad)
- The IC is now held in place, solder the rest of the pads.

This is how I do it without much problems. Only difference is that I use a microscope.
(But your handy-work does not look too shabby too... :) )

Lastly, do not tinker too much with your PL303QMD. Just use it, and start to enjoy and appreciate it!

regards, Gertjan.
 

Offline m k

  • Super Contributor
  • ***
  • Posts: 2648
  • Country: fi
Re: Venerable AIM TTI PL303QMD-P needs help - [was Output] Overvoltage trip error
« Reply #49 on: September 04, 2024, 10:00:20 am »
Here the awful result: not centered at all.


It's a camera effect, see C19.

If you ever start having intermittent issues, check through holes.
Factory tin is thin.
Advance-Aneng-Appa-AVO-Beckman-Danbridge-Data Tech-Fluke-General Radio-H. W. Sullivan-Heathkit-HP-Kaise-Kyoritsu-Leeds & Northrup-Mastech-OR-X-REO-Simpson-Sinclair-Tektronix-Tokyo Rikosha-Topward-Triplett-Tritron-YFE
(plus lesser brands from the work shop of the world)
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf