Author Topic: More x-ray images of XX1117 linear voltage regulators  (Read 4650 times)

0 Members and 1 Guest are viewing this topic.

Offline Chipguy

  • Supporter
  • ****
  • Posts: 294
  • Country: de
More x-ray images of XX1117 linear voltage regulators
« on: October 08, 2013, 08:14:53 am »
Hello,

here are more X-ray images of xx1117 linear voltage regulators from different manufacturers.
I started wondering after a TS1117 died in one of my designs for no obvious reason.



So i investigated and came up with a simple way to torture the regulator.
I wanted a realistic test, so the test circuit i made is not exceeding the limits of the regulator itself.
What i did was using insufficient cooling to make the thermal shutdown kick in.
The circuit supplies the IC with 12V, while the load is 200mA at 3.3V output.

After 3 days the TS1117 died, no burns, no pop, just gone.
After 11 days the LT1117 was still alive.

I have been told that you can see the suffering of the IC by a change of the output voltage with no load at room
temperature so i measured that before and after the tests. Apparently when the reference inside is drifting that can be a

sign of stress.

The TS1117 had 3.328V before and 0 after the tests.
The LT1117 had 3.309V before and 3.310V after the tests.

So i wondered how they look inside and guessed that the die in the LT is bigger than in the TS.
Which was confirmed.

The prices I quoted in the picture are all Farnell/element14 prices for 1 piece each.

It's interesting to see how different they are inside.
1) Looking at die size vs. price I would say the TLV1117 is the winner.

2) The LT1117 is the most expensive one but if price is not the issue and you need something really reliable thats what you
should go for. Notice the thick bond wires.

3) The TS1117 is only cheap in volumes and die size/experience wise the looser of the lot. Look at the thin bond wires,
LOL.  >:(

4) The NX1117 is also very cheap in volumes, the die is small, so maybe if you dont need much power this is a cost
effective part for a design.

5) The ON Semi Conducter one looks good too, mid range i would say. Slighly more elaborate then the TLV.

6) THE ST LD1117 looks nice too, seems to have a nice stable case.

I hope you enjoyed the x-ray images and that they help you to choose the right manufacturer for your design.
Below is the high-res version for download (I hope).

Cheers
Chipguy  :)
Where is that smoke coming from?
 

Offline BravoV

  • Super Contributor
  • ***
  • Posts: 5252
  • Country: id
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #1 on: October 08, 2013, 05:55:36 pm »
Thanks for sharing this, great post !  :-+

Any chance of doing it again at this kind x-ray photo on other popular power ic ?
 

Offline Paul Price

  • Super Contributor
  • ***
  • Posts: 1419
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #2 on: October 08, 2013, 09:22:45 pm »
Regardless of your guess of quality based on one failure and multiple X-Rays, your X-Ray analysis of bonding wires is probably not as good a judgement method as your thermal stress test, but it does give us a hint on robustness of construction.

The proof of a regulator is in the stressing. 

I would have liked to see more extended tests using thermal stress and time before any solid conclusions.

A single test is always just a single test..

Thank you for your posting, maybe you can be so thoughtful as to stress test the other brands of these chips and post results.
Some of the brands were quite reasonable in price..perhaps they are also just as reliable as the LT?
 

Offline dannyf

  • Super Contributor
  • ***
  • Posts: 8229
  • Country: 00
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #3 on: October 08, 2013, 10:30:19 pm »
Quote
The circuit supplies the IC with 12V, while the load is 200mA at 3.3V output.

I don't quite understand the purpose of the test. The chips are clearly being pushed outside of their performance envelope. If you intend to test for failure mode, sure. But if the chips were subject to this kind of operating conditions, someone has made a mistake.
================================
https://dannyelectronics.wordpress.com/
 

Offline Paul Price

  • Super Contributor
  • ***
  • Posts: 1419
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #4 on: October 08, 2013, 10:46:07 pm »
Dannyl,

A well-designed circuit must work and survive overload conditions if it is to be considered well-engineered and reliable.

There is a need and a want to test for failure to test the quality of a device, the device must operate reliably in overload mode according to the device spec.

Reliability is extremely important and thermal protection operation can show if a chip is robust and well manufactured.

Thermal protection is part of the the device specification. It must live up to its specs.

Overload is always  a possibility in any power supply circuit. A good power supply circuit must survive unexpected conditions or deal with other possible connected circuit's faults in long-term operation.
 

Offline digsys

  • Supporter
  • ****
  • Posts: 1710
  • Country: au
    • DIGSYS
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #5 on: October 08, 2013, 11:08:14 pm »
Great report - even though "everyone" should know already be aware of the crap that's out there - it's good to be reminded !
Hello <tap> <tap> .. is this thing on?
 

Offline amyk

  • Super Contributor
  • ***
  • Posts: 5736
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #6 on: October 09, 2013, 12:05:02 am »
I think it's perfectly acceptable for a part to fail if pushed outside manufacturer's operating conditions, and left there for an extended period of time. LT don't guarantee it works either - from their datasheet:
Quote
Stresses beyond those listed under Absolute Maximum Ratings may cause permanent damage to the device. Exposure to any Absolute Maximum Rating condition for extended periods may affect device reliability and lifetime.
Have you tried this "stress test" on the other 4 brands?

If you order them by die size/price, the LT and TS are actually the 2nd-worse and worse value... the best is the TLV1117.
 

Offline algorath

  • Contributor
  • Posts: 19
  • Country: at
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #7 on: October 09, 2013, 02:42:51 am »
can you xray some larger chips, ICs?
 

Offline Chipguy

  • Supporter
  • ****
  • Posts: 294
  • Country: de
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #8 on: October 09, 2013, 04:08:56 am »
Hello

thanks for all your feedback.
First of all : No, i think i can't X-ray much more IC's like that.
The images have been taken by my PCB assembler (Gohlke CCS in Hildesheim, Germany)
All this was part of a design of which they built 3 prototypes for me.

All this started when 1 prototype failed for no obvious reason. The regulator is used in there to generate 3.3V out of 5V.
It is well cooled and when under use barely warm.
So I went to them and asked for a little favor to see which would be the best choice.

However in theory I could do something similar in the future for other components for the same reason.
If I do, then I will share it here, promised.

In one of the feedback it was mentioned that I used the part out of spec.
So I had a look again on the datasheet of the TS1117 and figured that this is the only one of the lot with a maximum input voltage of 12V while it is even mentioned that the suggested operating range is only up to 7 volt.

Stressing the components is a valid test because insufficient cooling is one of the most things to go wrong in a design.
See that Atten power supply teardown (or was is Rigol) Dave made.

All other regulators of the lot can take either up to 15V or 20V with no restrictions.
So that actually proves my point, stay clear of the TS1117.
It should not even be named/numbered as 1117 because it does not even take 15V in.

Other news as of today:
It even seems that the second prototype using TS1117 is failing.
There was a "crash" of the hardware today which could mean the 3.3V dropped out.
The circuit is now under surveillance with the 5V and 3.3V monitored for drops and overshoots.
Maybe I should not even bother, and just replace the damn thing.

I would like to do more test but unfortunately I don't have the time and resources to conduct them.

Cheers
Chipguy


Where is that smoke coming from?
 

Offline dannyf

  • Super Contributor
  • ***
  • Posts: 8229
  • Country: 00
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #9 on: October 09, 2013, 04:11:36 am »
Quote
All other regulators of the lot can take either up to 15V or 20V with no restrictions.
So that actually proves my point, stay clear of the TS1117.
It should not even be named/numbered as 1117 because it does not even take 15V in.

Voltage didn't kill those chips - it is the dissipated power that did the job.

================================
https://dannyelectronics.wordpress.com/
 

Online NANDBlog

  • Super Contributor
  • ***
  • Posts: 4047
  • Country: nl
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #10 on: October 09, 2013, 12:57:23 pm »
Thanks for sharing the findings. However. If you are so concerned about reliability, than why do you go with the SOT-223 package? Slightly bigger, but there is the DPACK, which has half of the thermal resistance, which gives you 12K lower  temperature for that load (on the die). Also, you probably know that the PCBis used for heatsink, so the design of that has a very much effect on the reliability. If you must dissipate that ~2W heat, you need about 30 cm2 area if you aim 40K temperature rise, in still air, depending on many factors.
Also, if you aim reliability, I dont know why you started with the 1117. They don't work well with ceramic capacitors, only with electrolyte, and they could fail after time, and you need the reverse current diode. Sure they are cheap.
 

Offline ejeffrey

  • Super Contributor
  • ***
  • Posts: 1615
  • Country: us
Re: More x-ray images of XX1117 linear voltage regulators
« Reply #11 on: October 09, 2013, 02:57:24 pm »
Quote
The circuit supplies the IC with 12V, while the load is 200mA at 3.3V output.

I don't quite understand the purpose of the test. The chips are clearly being pushed outside of their performance envelope. If you intend to test for failure mode, sure. But if the chips were subject to this kind of operating conditions, someone has made a mistake.

These regulators are supposed to have thermal overload protection that keeps them from exceeding their maximum ratings as long as the input voltage is in spec.
 


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf