TPS6123x burn

Hi Jasper, thanks for answer.

The burn conditions are unknown, but my investigation and research can reproduced burn, but it depend on chip.....

in my device returned from terrain (burned TPS)  i was replace this chip an continue research,

so replaced chip was burned in following condition:

The converter was launched, but no lead , after few minutes i made short circuit VCC to GND (less than 200ms) , the chip immediately burn.

so i was replace one again damaged chip, and continue test short circuit with poly-switch (2,6A) and next with out poly switch, and no effect, all are OK, I can't reproduce burn out...... in none case.

It look like that , it depend on sort of chip tolerance.....

please look in schematic (partly simplified for better understanding)   and PCB layout (.png) and Atium PCB

Hi Jasper

 COUT, are you thinking about 3x capacitor 22u/10V X5R (marked on my schematic as c15,c65,c99)?

if yes, capacitors are connected on the same layer, please take a look that i have additionally tantalum capacitors connected on in input (2x 100u)  and output 100u +470u, also very close on chip and connected on the same layer.

from my experience:
series 20 pcs devices as prototypes was launched with positive results (all was tested on maximum lead DCDC converter >1,5A /5.15V ) . it works exactly  as expected.

at the moment i have 4 cases burnout DC/DC converter

My research take unstable results:

in laboratory i have 1 case, where short circuit  damage TPS immediately , and one when NOT (even if it is permanently long time short  circuit VCC to GND )

please notice that DC/DC converter is started only on demand where main VCC (external for other sources ) was dropped  bellow setting, and measured by MCU (in this case <4.9V)

So, maybe important is that TPS have 5VDC on back side, form external sources?

in some cases to major functionality is increase (stabilise -fill)  main VCC  5VDC (from external supplier) when this dropped bellow mentioned value......

I research also case when DC/Dc are launched without lead, i tried  (multiply time) connect to output , large area of capacitors  20X1000u (discharged) , for simulate current hit. but this no causes damage DC converter,  in none cases.

Now, i am afraid start  production with more quantity series.......

this device doesn't guarantee short circuit protection. the device could damage in some short circuit scenario.

in your application, if there is not short circuit condition, the damage may cause by other reason. normally, the damage could cause by high current or high voltage, so
1. could you please to measure inductor current in all the condition, especially in the transient condition, to check if there is a scenario that the inductor is out of control.
2. could you measure the voltage spike between SW pin and the GND pin with probe (set the probe full bandwidth) to check if the damage is caused by the voltage spike? make sure the measurement loop is small to avoid noise coupling.

Hello Jasper/I will prepare mentioned waveform soon.

You prefer with lead, or without?

hello Jasper i t thought  about conditions for  waveform:

please take me rules for measuring, my DC/DC converter should be  be work  in range  0-1,5A/5,15V.

I need to know current output for waveform..... and input voltage.....

regards Mariusz

Hello Jasper.

I the probe was done. In case voltage between GND and SW pin

You can see two case 

1) voltage drop (current on track between accu and chip) (unknown resistance but probably  ~~40mohm)

2) SW-GND normal work with full lead

3) SW-GND in start converter

4) Out - GND  voltage output on start

Do you have an idea how to measure current inductor without increase inductor circuit resistance ?

current coil is difficult to measure so i made measured voltage droop on track from accu to chip. 

Hi Jsaper.

in this example DC/DC converter will start only from accu. (no external supply on back side). Load are other electronic devices with large capacitors.......   

Unfortunate I don't have current probe,

even if I will have then this measure can be  very difficult  (this need de-solder coil)

what i can do at this moment...  buy very small resistance resistor (1mOhm), and measure via scope voltage droop. but this will change coil circuits.

1) VIN (blue)  vs Vout (yellow)

    Vout  vs SW

the same but.

Hi Jasper.
please look in my measure circuit. I think that it is OK.please confirm.

What signal i can use for trigger probe?  in mentioned document is used load current  (in my circuit I don't have this possibility)

what should be expected  time-line in oscilloscope ?

In this case i used slope for trigger

within mentioned document my R1=430K R2=102K C=470p

in this post You can see simplified block diagram my module with DC/DC converter  for better understanding whole work condition:

My module is marked in breaking line:

CPU control  5V line from external AC adapter, when 5V is dropped or no 5V from Ac adapter, then CPU do:

left switch turn OFF

TPS switch ON

right switch turn ON

CPU is supplied directly from accu.

normally when main 5V is ok (99,999% time of work) , left and right switch are in ON state.....

so, in normal state, TPS have 5V  from main 5V whole time..... (back side)

Is possible install module in HOTPLUG mode when

main 5V is present from AC adapter, but TPS is working mode, because module is before plug-in to circuit and not "see" yet main 5V from AC adapter.

In this case, is possible that in short time (5ms) that AC adapter (by schottky diode)  and TPS supplied right load, together with AC adapter  ,

After plug in module to circuit  (hot-plug)   then CPU measure 5V from ac adapter , and switch OFF TPS converter, turn "ON" botch switcher, in this mode right load will supplied directly from AC adaptor.       

Hi Jasper.

point 2) AC adapter powered 5V - 5.2V and module with TPS delivery 5.1-5.2 V depending on load (this are  very similar  voltage)

point 3)

The idea is: powered right load in all condition (if AC adapter are dropped or gone) by AC adapter or by TPS6123.

we have three case:

first), normal working, module is plugged in circuit,  AC adaptor is OK and measured 5V by CPU is OK, then both switch are in state "ON", TPS in "OFF" state , left load is powered by AC adopter directly , right load is powered from AC adapter trough  by both switches.

second), normal working, module is plugged in circuit, AC adapter is fail (measured 5V by CPU in module reports that is less than 5V or 0). then module go to state: left switch go to "OFF", right switch continue state "ON" and TPS go to ON,  in this case right load have always 5V whole time.

there) install mode in this case module isn't on the circuit, but is running. 

left load are powered from AC adapter, right load isn't not powered because module is plugged OFF

in this mode TPS is in "On" state right switch is in "ON" state, but TPS are working without load. (no connection with right load)

next the working module is plugged-in circuit  in this time right load will powered together  for AC adaptor by diode in module, and from TPS by right switch. 

when module is plugged in circuit CPU measure 5V form AC adapter and module continue normal working.

in question large capacitors You have right  is possible current hit, because in load right i have 4000uF

but in my module i have also 4000uF connected directly to output TPS for storage energy  when 5V dropped, and balance current hit.

It is second generation module, I think  that idea is corrected  (used the same solution switch, diode, etc)

in first generation was used MAX1708,

in population more than 5000 pcs and 5 year experience working isn't unknown case damage DC/DC converter

I decided use TI solution, because maxim is expensive and take more space PCB.

What i can do something else ? 

i need resolve problem... Are You sure that magnetic parameters are OK?

I hope that this description should clarify idea and all condition for TPS.

Best regards Mariusz

Hi Jasper.

OK.  I understand,

when is install mode always right load is 0V......

if the chip isn't immunity to short-circuit, then i don't see solution.

some chip based on non representative parameters work correctly, and some will damage.

The main problem is that module doesn't know when is mode install. and always exist risk current hit from right load.......



Is no possible "delay start" because module doesn't recognise when is install mode or when is 5V dropped, from point off view module, the are the same condition.

This know only customer who will install module.


Do you know maybe about other TI solution, protected by short-circuit (best 1:1 replace).

or how to protect TPS in this case:

polyswith for right load? (i have place on PCB to install this, now 0R is mount)

What do you think about increase capacitors on output TPS from 4000uF to 6000uF?

other solution?

In second hand, I tested short-circuit  by connect discharged to 0V large capacitors  area (more than 20000uf as right load) when TPS was in on mode.....

but I know, that in some cases (none representative) this test can pass..... or not

please look on updated block diagram (it is fill on real capacitors value, on side module and right load)

What can be maximum capacitor on output TPS? it is limited ?

Hi Jasper.

within your opinion, You claim that damage TPS can only occur when module is installed  in circuit (TPS is working),

so the load capacitance do short-circuit it can caused burnout the TPS. Yes ?

the solution will be

detect that module is installed or not, and if no installed the TPS could be in off state. Yes?


other question.

Is possible that module installed on circuit, and AC-adapter is switch off state. In left, and right load are 0V

Can i turn ON TPS (where) left switch is in state OFF, right switch is in state ON,  right load is 0V, total load capacitance is 10000uF?  is it safe for TPS ?


regards Mariusz

Hi Jasper.

I have idea how to recognize that module on circuit or not, So it should be resolve problem Hot-plug module, and exclude short-circuits event.

abuot:  "please make sure the device is stable."

My scope  waveform was as attached pictures . I don't see typical Waveform pattern within SLVA document (by DC/Dc started)

What  do you need else, to confirm that the device are stable or not ?

blue line = measured voltage output when DC/DC are starting (full load 1,5A capacitance  4000uF from 0V )

Hi Jasper.

I need make sure.

1) You need as load pure resistance load 3 ohm without output large capacitance ?  only  3x  22uF like in reference ?

if yes follow is waveform in mentioned condition (pure 3 ohm resistance only 3x22uF on output TPS).

it is right measure as You expect  ? (within SLVA condition)

Hi Jasper

I made few pictures with different  base time

so first three, presents when module isn't on circuits.(without right side load)
in ths case on TPS output is connected 3x1000uF (aluminium)  and 470uF (tantalum) and 100uF (tantalum),  in this case, load for other PCB elements is very low..... <100mA

next pictures presents waveform when module is on circuits (right load side is connected for TPS)

Attachment is my measurement with the TPS61230EVM and 10000uF output capacitor. it seem good.

what you attachment should be startup waveform, i don't see problem from the waveform

Hi Jasper.

thanks for answers, What i can improve else to avoid burn:

Do You think  that  exclude short-circuit (large capacitance- connected  when TPS is running) should be resolved problem?

Regards Mariusz.