Problems with LDO regulator AZ1117lH-TRG3.3

Go To Last Post
38 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm having heavy headaches with the LDO regulator of the title... they work when initially tested (either dead bug or on an otherwise empty PCB) but I've had a better than 50% failure rate where subsequent soldering on the board will suddenly cause them to emit 10.7v, 12.5v, 4.8v and indeed any voltage other than the expected 3.3v. Which basically takes out the processor, which is a pain because (a) they're a few quid each and I only have a few left and (b) with 64 legs they're a pain to remove and replace.

 

So I just built a board with everything on it except the regulator; added the regulator, tested the voltage, all fine. Programmed the processor. Fine. Added six more solder joins to connect a display (I'm adding the expensive bits last as they cost more than the rest of the board in total, and I have a very limited supply of them) and suddenly the thing is pulling half an amp from the bench PSU and throwing ten volts out.

 

I'm beginning to lose my sense of humour; the soldering iron is earthed at the tip; I'm anti-staticed, the part is wired per the datasheet and well within its load parameters; working units get warm - it's dissipating 600mW but into a nice solid Vcc plane and never gets too hot to touch.

 

Right now I'm looking for alternatives for the part with a SOT223 pinout...

 

Neil (grumpy mode=on)

 

Edit: changed title to correct part number

Last Edited: Thu. Jul 8, 2021 - 06:27 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
and suddenly the thing is pulling half an amp from the bench PSU and throwing ten volts out.

What is the bench supply set to when this happens?

voltage and current limit settings?

 

What type solder are you using, and with what flux?   water soluble fluxes must be rinsed and dried before power is applied, as they are conductive!

 

Jim

edit:  LDO, so I would not supply more then 5v to the input of the regulator

 

 

 

Keys to wealth:

Invest for cash flow, not capital gains!

Wealth is attracted, not chased! 

Income is proportional to how many you serve!

If you want something you've never had...

...you must be willing to do something you've never done!

Lets go Brandon!

Last Edited: Wed. Jul 7, 2021 - 02:43 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
they're a few quid each and I only have a few left
only a few quid? (rhetorical of course) cheeky

Ross McKenzie, Melbourne Australia

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

They're being fed 12v-13v; leaded solder; no-clean flux or no flux beyond what's in the multicore. They're rated for 20v input max.

 

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

At the failure time, they're not delivering more than a couple of mA; the heavier currents - 40mA or so - are in the display which hasn't been connected at that point.

 

The PSU is limiting at a nominal 1A.

 

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

That's a pretty scary story!  I would test a bunch with just a resistive load drawing maybe twice or more what your board should be drawing.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Are you using a low ESR such as a ceramic capacitor on the output side of the LDO?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Yes.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I think that is the cause.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Hmm... charging current?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Search for "LDO" and "ESR".
You can get information about output stability.
The LDO data sheet, which can use ceramic capacitors for the output, clearly states this.
It cannot be used with older types of LDOs such as the AZ1117.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

kabasan wrote:

Search for "LDO" and "ESR".
You can get information about output stability.
The LDO data sheet, which can use ceramic capacitors for the output, clearly states this.
It cannot be used with older types of LDOs such as the AZ1117.

This is very interesting.  One datasheet I found specifies tantalum or aluminum electrolytic caps, 0.2 to 10 Ohm ESR.  Another datasheet says nothing about this.

 

OP, do you have a link to the datasheet for your part?  (there seem to be different variants on the net)

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


The TLV1117LV I use says:

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Did you allow enough heatsinking on your pads?  (do a polygon fill).   A basic SOT-23 is may be rather poor at more than 200mW or so. However, overheating typically causes shutdown, not violent overvoltage  output. Scope the Vreg, to see if you have cap instability.

 

Some regulators from China are ripoff junk---so beware of bad goods.

 

 

the soldering iron is earthed at the tip; ---you certainly shut off PCB during any soldering

 

 

Is there some other sneak path for HV onto the rail...say through some output driver transistor assembly?

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I would set your bench supply down to 5v unless there is another reason for it to be at 12v, and set the max current to twice what you expect your board to draw.

At least while testing.

Jim

 

 

Keys to wealth:

Invest for cash flow, not capital gains!

Wealth is attracted, not chased! 

Income is proportional to how many you serve!

If you want something you've never had...

...you must be willing to do something you've never done!

Lets go Brandon!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


barnacle wrote:
... will suddenly cause them to emit 10.7v, 12.5v, 4.8v and indeed any voltage other than the expected 3.3v.
1117

  • require pre-load else over-volt
  • are conditionally stable (output capacitor value greater than a minimum and ESR within a range)
  • an open in the voltage divider will result in an over-volt

barnacle wrote:
Which basically takes out the processor, which is a pain ...
Consider a crowbar or clamp.

Reason : one of several is a shorted pass transistor

barnacle wrote:
... and suddenly the thing is pulling half an amp from the bench PSU and throwing ten volts out.
Possibly an oscillation (an instability); regulators are very typically conditionally stable (dependent on input impedance, output impedance, di/dt; a load can upset the regulator)

barnacle wrote:
Right now I'm looking for alternatives for the part with a SOT223 pinout...
The ones at Linear Technology created a 1117 follow-on that's a series pass PNP, instead of series pass NPN, with a loop gain of one (LT3080 series)

Similar method for recent CMOS voltage regulators though with a series pass DMOS PFET (reduced drop-out); these are relatively inexpensive, simple stabilization, some are low noise and/or high PSRR, with low quiescent current.

 


note, revision I instead of H :

AZ1117I (Low Dropout Regulators (LDOs))

 

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 120, lower left]

9.13.1 Overvoltage crowbars

XR21B1420 Data Sheet (MaxLinear)

[page 59]

VCC Clamp Circuits

 

https://www.analog.com/en/parametricsearch/11536#/p5576=Current%20Source%20Reference&sort=5576,asc (feature of current source reference)

LT3088 - 800mA Single Resistor Rugged Linear Regulator

[page 1]

Pin Compatible Upgrade to LT1117

[page 10, top of right column]

Using the LT3088 as a Replacement for the LT1117

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A regulator's stability does vary by load current (iow, gain and phase margins are dependent on load current, etc)

Engineer It - How to test power supplies - Measuring Stability | TI.com Video

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

kabasan wrote:
The LDO data sheet, which can use ceramic capacitors for the output, clearly states this.

NCP1117LP - Voltage Regulators, 1.0 A Low-Dropout Positive, Fixed and Adjustable (ON Semiconductor)

[page 10, top of right column]

Frequency compensation for the regulator is provided by capacitor Cout and its use is mandatory to ensure output stability. A minimum capacitance value of 4.7 F with an equivalent series resistance (ESR) that is within the limits of 20 m to 20 is required. The capacitor type can be ceramic, tantalum, or aluminum electrolytic as long as it meets the minimum capacitance value and ESR limits over the circuit’s entire operating temperature range. Higher values of output capacitance can be used to enhance loop stability and transient response with the additional benefit of reducing output noise.

[bottom of left column]

Input bypass capacitor Cin may be required for regulator stability if the device is located more than a few inches from the power source.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I am suspicious on 20V data. My AMS1117 works up to 12V (15V absolute max).

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

There are a number of variants with Vmax at 15v, 18v, or 20v... I think it's the cap that's the issue. I'm about to order some tants...

 

I'd also like to order some more processors, but at the moment nobody has any STM32L073RZT with deliveries November if you're lucky or May if you're not. I might have to get creative...

 

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:

At the failure time, they're not delivering more than a couple of mA; the heavier currents - 40mA or so - are in the display which hasn't been connected at that point.

 

You could try just varying the load on a less populated design (but with all decoupling in place) ? 

Did you scope the power, to see if it oscillating ?  If the load is reduced, does it work properly again ?

How do you power the unit - by direct plug in ? 

I've seen comments before about live-power connections giving issues with the inductive overshoot, where the 1uF or so input side cap, resonates with the lead inductance to spike above the rated VMAX.

Those ferrite beads and PTCs you often see, can serve multiple purposes :)

 

We had one issue here on a test bench, that a change to a modest series R in the lower lead cured what looked like live-connect power spike issues.

A connect-then-apply-power order was always ok, but who has the time to do that ? ;)

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

"AZ1117H-TRG 3.3" cannot be found.
AZ1117 has many second sources.
You should show the datasheet for the device you are using.
This is the cause of the increase in false advice that refers to the wrong data sheet.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
There are a number of variants with Vmax at 15v, 18v, or 20v...

The 3.3V type should definitely be with 15V absolute max, and the input voltage is 5V in most use cases.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

https://eu.mouser.com/datasheet/... - it claims vmax as 18v. My apologies for missing the 'l' suffix.

 

There are only three part-built loads I can test at: regulator alone; regulator + processor + decoupling caps; and plus the OLED display. Scope investigation is tricky as I don't own a scope, but I'm looking into it.

 

Though the whole thing may become moot due to the absence of processor chips...

 

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


There would have been less confusion if you showed the datasheet first.

 

There seems to be no ESR problem I pointed out.

 

 

 

 

Absolute Maximum Ratings and Recommended Operating Conditions are often confused and misleading.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

show ye schematic & pcb!

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
I'm beginning to lose my sense of humou

Where I work there would be other Engineers to laugh at my misfortune. I guess this doesn't apply to you.

 

I found linear Vregs to be quite reliable so long as you don't overvolt the input. Which leads me on to the story of a crappy bench PSU.

 

In the office Lab we have a Dual Channel Bench PSU (with standard 2A / 30V independent or tracking mode outputs, LED meters for both current & voltage and output disconnect switch). A fellow Engineer was blowing up development boards at the rate of approx 1 every 2 or 3 days. So because he was now beginning to lose his sense of humour we eventually took a look a the PSU output (Normally this is the one thing you trust) but found it was accurate. He carried on with another board.

 

On the next instance of board destruction he associated the blowing up event to switching on the bench supply using the Power switch with the outputs connected to his board. A scope trace showed a brief but significant excursion to the maximum output voltage of 30V before stabilising at the correct level.

 

That cursed PSU now resides on the bottom shelf of dust in our lab.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A scope trace showed a brief but significant excursion to the maximum output voltage of 30V before stabilising at the correct level.

That is actually power supply option -XVST  (voltage stress test)...you probably paid an extra $500 just to have that angel  At least it is now on the "shelf of honor"

 

At our company, a QA tech was running a 48 hr accelerated storage temperature eval on a dozen relatively expensive handheld units.  So on Friday he set the oven to 200 & went home for the weekend.  Upon returning Monday, he was met with a puddle of rather slumped / melted units. What!!! 

Turns out that 200C is not the same as 200F !!  We put one on his desk for some good laughs.  All the electronics actually did power up just fine after the "test".

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Who-me wrote:
I've seen comments before about live-power connections giving issues with the inductive overshoot, where the 1uF or so input side cap, resonates with the lead inductance to spike above the rated VMAX.
Likewise read.

Who-me wrote:
Those ferrite beads and PTCs you often see, can serve multiple purposes :)
Indeed (some ESR)

Who-me wrote:
A connect-then-apply-power order was always ok, but who has the time to do that ? ;)
smiley a hot plug capability is handy though below a certain voltage (connector arcing, may not be an issue for 12Vdc but is for 24Vdc medium truck/semi-tractor/bus)

 


LT3088 - 800mA Single Resistor Rugged Linear Regulator

[page 11, bottom of right column]

Stability and Input Capacitance

...

However, applications connecting a power supply to an LT3088 circuit’s IN and GND pins with long input wires combined with low ESR, ceramic input capacitors are prone to voltage spikes, reliability concerns and application-specific board oscillations.

...

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
Scope investigation is tricky as I don't own a scope,
jelly bean voltage regulator's error amplifier gain-bandwidth is only a few 100KHz; so, a very inexpensive scope.

barnacle wrote:
but I'm looking into it.
Consider a low frequency network analyzer though a college may prefer a scope with Frequency Response Analysis (FRA)

barnacle wrote:
... due to the absence of processor chips...
A generalization though is specific in this matter.

 


Haasoscope Boards - Crowd Supply | Mouser

Haasoscope | Crowd Supply

 

Using the Network Analyzer - Digilent Reference

Easy Control Loop Measurement and Verification (ON Semiconductor)

 

ATMEL parts are running out of stock due to wafer shortage ? | AVR Freaks

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

N.Winterbottom wrote:
On the next instance of board destruction he associated the blowing up event to switching on the bench supply using the Power switch with the outputs connected to his board. A scope trace showed a brief but significant excursion to the maximum output voltage of 30V before stabilising at the correct level.
Likewise for power off.

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 120, right colum]

We’ve encountered some aberrant bench supplies that soar to their full output voltage, briefly, when you switch them off. But “briefly” is all it takes to ruin your whole day!

Some power supplies have a low voltage lock-out.

 

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

My company blacklisted Diodes Inc; LDO and all of their voltage regulars after receiving a bad 3000pcs reel from them 5 years ago. About 60% failure rate on reel. They confirmed a defect in the die, offered to ship us replacements. No thanks. 👎 .... very interesting to see another person having an issue again. I recommend never buying any regulators from them unless you like doing lots of field RMA's

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 1

gchapman wrote:

N.Winterbottom wrote:
On the next instance of board destruction he associated the blowing up event to switching on the bench supply using the Power switch with the outputs connected to his board. A scope trace showed a brief but significant excursion to the maximum output voltage of 30V before stabilising at the correct level.
Likewise for power off.

The Art of Electronics 3rd Edition | by Horowitz and Hill

Download a sample chapter

[page 120, right colum]

We’ve encountered some aberrant bench supplies that soar to their full output voltage, briefly, when you switch them off. But “briefly” is all it takes to ruin your whole day!

Some power supplies have a low voltage lock-out.

 

"Stick with vacuum tubes!" I said, but nobody would listen.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:

There are only three part-built loads I can test at: regulator alone; regulator + processor + decoupling caps; and plus the OLED display. Scope investigation is tricky as I don't own a scope, but I'm looking into it.

You can easily add dummy loads tho, to shift the current up and down.

You also did not say if the regulation recovered after load removal ? - a Load dependent oscillation will go crazy, but you would expect it to restore, when the load is back in the stable area.

You could also try different vendor's 1117 variants,  to see just how much this is brand dependent ?

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

No, the device did not recover after load removal. I'm becoming suspicious of the bench power supply; overshooting on switch on could explain a lot.

 

I have a batch of different 1117s arriving in the next day or so.

 

Neil

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

N.Winterbottom wrote:

.... Which leads me on to the story of a crappy bench PSU.

 

In the office Lab we have a Dual Channel Bench PSU (with standard 2A / 30V independent or tracking mode outputs, LED meters for both current & voltage and output disconnect switch). A fellow Engineer was blowing up development boards at the rate of approx 1 every 2 or 3 days. So because he was now beginning to lose his sense of humour we eventually took a look a the PSU output (Normally this is the one thing you trust) but found it was accurate. He carried on with another board.

 

On the next instance of board destruction he associated the blowing up event to switching on the bench supply using the Power switch with the outputs connected to his board. A scope trace showed a brief but significant excursion to the maximum output voltage of 30V before stabilising at the correct level.

 

That cursed PSU now resides on the bottom shelf of dust in our lab.

 

Another variant on that story - 2 identical Bench PSU units were built by the apprentice, and only one sometimes popped the linear controller chip - after a lot of miss-wiring checks and careful compare, they were confirmed 100% identical circuits.

The difference was finally spotted by some sharp eyes : the transformer primaries had Phase switch on alternate ends.

The unit with phase furthest removed from the secondary winding, was the good one, and the phase switch-arc-spike on the bad one, was coupling thru to the secondary with enough energy to damage the controller.

Sure enough, swap the wires fixed it, and I think a snubber may have been added too, to both after that, as added insurance.

 

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

barnacle wrote:
... cause them to emit 10.7v, 12.5v, 4.8v and indeed any voltage other than the expected 3.3v.

 

Like quite a few other old linear regulators, according to the datasheet the AZ1117I got a minimum load current of up to 5 mA (max), if the voltage drop is >1.5V (like in the given case).

 

If the load current is too small, the output voltage will typically rise to a higher voltage depending on the individual device and the actual load.

A well-known example showing such a behaviour is the LM317.

 

If the application doesn’t have a defined minimum load, simply add a suitable resistor at the output – or use a different LDO.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

s'AVR wrote:
– or use a different LDO.
The reign of the bipolar LDO is ending other than for ultra low noise (CMOS can be low noise though usually doesn't extend much past 6V in)

 

"Dare to be naïve." - Buckminster Fuller