EK-TM4C1294XL: Discovered RA2 Launch Pads read 247-250 ohm VDD to ground and behave badly at times.

BTW: These EK-TM4C1294XL launch pads still execute and flash applications even with VDD 247-250 Ohms. Highly suspect production fallouts in both these launch pads and perhaps more are reeking havoc on the community! A diode check of GPIO ports to ground also indicates a low drop reading 0.770v versus 0.889 or high as 1.2v a pristine gate behavior.

Who would ever think to do an ohmic test of VDD to ground or GPIO diode test on a brand new launch pad that flashes and seems to run applications ???

Hi Bob,

Bob Crosby said:

I cannot make out much from the resistance readings you gave

Unpowered launch pad DMM ohms check and diode check both detect low resistance short exist yet the board powers up. Hard to imagine DMM ohms check 1vdc would cause rail protection diodes to conduct when +3v3 is typical supply. There is notable board warp around the MCU via's for VDD decoupling caps already extremely close to each other but even 8 diopter magnification can't tell if shorts exist between them.

These problems are nothing new most have been here out of the box with one of two boards reported this forum last year. For one the ADC will randomly lock up if a channel input is left floating without a connection and the application samples the channel. Never bother to check JP2 current draw since the application was running and LED's blinking made it seem all was ok. Did report this forum a 100mv noise condition on 3v3 LDO back some time ago.

The 7.5k-9.7k ohm VDD launch pads don't seem to lock up nearly as soon or often. if static electricity has somehow shorted protection diodes the current draw JP2 should be roughly 0.013ma higher VDD draw. Been very careful to discharge my self prior to handling launch pads, especially in the basement lab we have foil covered insulation near work benches to discharge body static.

cb1_mobile said:

are you not measuring "everything" coming into contact w/3V3

+3v3 JP2 is removed so no it is only the MCU and decoupling capacitors + internal peripherals. Reversing DMM leads still measures same very low 240-250 ohm value on two boards with odd issues. Yet explain good launch pads (plural) reading 7.8k- 9.8k+ ohms have no issues with random lock up or the PWM voids. DMM on diode check lead reversal measures 0.4v 0.2v drop compared to 1.9v drop good launch pad.

A curve tracer (bench repair checked CMOS device)  would show a vertical closed oval indicating capacitance with very low resistance versus a nice check mark.

cb1_mobile said:

As (always) BP faults the vendor - still proclaims "void" (even when - & especially when - "Alias" & improper measurement technique) rise quickly/powerfully as "top" suspects...

New launch pad does not have voids in PWM seems to indicate 240 ohm VDD was causing some interval HW failure with PWM or ADC peripherals.

Bob Crosby said:

Comparing the different launch pads at 300mV might be a better indication of a resistive shorts as that voltage is low enough not to turn on the transistors.

Some what confused since that would violate ohms law which requires DMM injects 1 volt into working circuit applied at very low micro amps, not milliamps.

Another test: DMM diode check injects milliamps into VDD rail, also reveals very low voltage drop (0.2v) on bad lauch pads and (1.6v-1.9v) on good launch pads. DMM diode check can illuminate most LED so there is milliamp current present in the drop, often under 30ma.

It is possible a bad capacitor exists in the VDD rail but suspect a PCB copper whisker short between traces is to blame.

FYI: Surprisingly VREFA+ input to MCU traverses out to the X11 header is not indicated in the schematic and VDD C16 parallel to C41 is not indicated either.

Hi BP101,

Sorry, I seem to be having an unusually difficult time explaining my point. First, I am not saying the boards are bad or good. My point is that measuring the "resistance" of the power rails of a 3.3V semiconductor by applying 1V and measuring current can be very misleading as the 1V is very near the threshold voltage and is starting to power on the active circuits. If on a board you measured 250 OHMs with a 1V forcing voltage, that implies the meter measured 4mA of current. If instead you used 300mV forcing voltage and measured 1.2mA of current, I would agree you have a 250 OHM resistive short. It is like measuring the forward "resistance" of a diode. If I measure a diode which has a nominal 700mV voltage drop with a 1V forcing voltage, I would think I had a low resistance. If I measured the same diode with a 300mV forcing voltage, I would measure a very high resistance. I have run into the same issue with in-circuit board testes that tried to identify shorts between traces and used too high a forcing voltage. Since the threshold voltage on devices does vary, some parts would pass, while others would "fail". Both sets of parts were actually good and well within range of the specifications.

Now back to your issue. I am more impressed by the results of your diode measurement. Measuring a 200mV drop is much less than I would expect. You suggest a bad capacitor or a copper whisker. That could be, but I would not expect a 250 Ohm copper whisker to affect operation. It would simply add 13mA to your operating current. A bad capacitor may affect operation. Particularly if the issue is related to filtering noise on the power rail. Another possible source would be an over-stressed ESD structure. When too much current is passed through the ESD structure, sometimes the power dissipated causes melting of the internal copper and you create an internal resistive short. The problem is the ESD structure is destroyed and can no longer help protect against over voltage transients. That may explain the different behavior in your system where you are dealing with the fly-back voltages from the current switching. Both an over-stressed ESD structure and a bad capacitor can be caused by over-voltages from the fly-back of switching large currents with inductive loads.

Bob Crosby said:

Both an over-stressed ESD structure and a bad capacitor can be caused by over-voltages from the fly-back of switching large currents with inductive loads.

One of these launch pads has never been subjected to a motor inverter and had issues with random USB host disconnects brand new out of the box. At that time logged forum complaint +3v3 LDO might be to noisy when powered via computer USB and simply figured disconnects must be application bugs and set it to the side for months unused.

Also like to note motor inverter launch pad +3v3 LDO is powered by LM5007 bucking +24v to +5v. The LM5007 filters most all +/-4vac PWM transients down to +/-300mv levels on the launch pad connected to motor inverter drivers FAN7382. Oddly PWM flyback pulses don't occur on +3v3 LDO and rather on the (Cf) decoupled VREFA+. Flyback pulses are coming into INA240 analog current monitors via low side inverter shunts and bleeding into ADC reference VREFA+ but they are millivolt in size. Yet they are even higher in amplitude +/-4vac on the +24vdc LA battery power.

BP101 said:

Should it not require at least 2KV or more to short 1 or several rail protection diodes?

Vendor's "Bob" (& this reporter) provided insight on just this issue - just yesterday.    Having past worked at a similar, semi-giant - we note that those "protection diodes" are NOT designed to survive, "unlimited & extensive" current flow.   Note too that these diodes conduct when the input voltage exceeds a "diode drop" - either above VDD - or below GND.   Repeated "exercise" of these diodes usually proves problematic - and the destructive, internal metalization often cascades - "wreaking" havoc.

Now some here know that you've made an intense study of Power FETs - and are knowledgeable of the FET's "body diodes"    Those FET body diodes are very substantial structures - unlike the diodes (squeezed) into the far more complex MCU.    MCU protections are to be used "sparingly" for best, long-term device performance, even survival...

As the board in question provides an easy means to measure & monitor current - that measure avoids any possibility of  "unwanted MCU (and other device) junction turn-ons."     (caused by a (likely) improper "resistance measurement" (incorrect injection voltage level) of the MCU's power rails.    The simple fact that something "Can be done" does not extend to, "Should be done."    You know that my firm employs ARM MCUs from multiple vendors - "NOT ONE PROVIDES the VDD to GND" resistance measures you seek!    Such should be telling - is it not?

MCU's current measurement - repeated under identical set-up & operating conditions - provides the "Gold Standard" for such measures.     And explains why the VDD line to the MCU is easily accessed/interrupted - & provides quick, simple, effective MCU current monitoring.

One final - not related point - you were an early adopter of vendor's improved current monitor, "INA-240."   (and were gracious enough to "share" your find.)    This device was specifically intended to "reduce the normal/customary, negative effects imposed upon such monitors - by "high level common mode swings."    You've placed that device(s) in the Low Side FETs' return path (earlier BP post reveals) - where the common mode swings are minimal!     Far better placement - and implementation of the INA-240 - occurs when the device is placed w/in the motor's Phase line(s) - where it IS exposed to (and minimizes) the negative effects of such, "severe common mode swings."

cb1_mobile said:

"NOT ONE PROVIDES the VDD to GND" resistance measures you seek!    Such should be telling - is it not?

Are you implying the diode check or even the rail resistance checks are not valid method that I have done for over 3 decades on hundreds of bad silicon devices. Find it really odd that a new Launch Pad out of the box never connected headers to anything is having these very same VDD low drop/resistance readings.

I sand by either they both were static zapped severe enough to cause rail diodes to short out or the PCB is defective and or 200mv VDD drop are one in 100's of MCU runs. A simple lab test would measure a new EK-TM4C1294XL JP2 removed test VDD diode drop on JP2 pin 2 then zap it good with the ESD gun and take a second reading after. Do this several times and up the ESD gun voltage each time then re-test VDD rail drop. The idea is the diode protects VDD input from +4v up to a 2KV and +/-400mv PWM spikes on VREFA+ is not going to short a protection diode out. If any thing might occur is the substrate gets weakened in the area of concentrated electrons punching through but not on a clean stand alone VDD. That is unless it is more prone to such failure when directly connected to VREFA+ as R41=0R0 and stupidly runs out to X11 like a dang antenna wire waiting for an ESD zap.

Again the question for TI-FE is does a severe enough static zap cause internal MCU destruction to then measure VDD at a 200mv drop or not and does it still POR after doing so?

cb1_mobile said:

Far better placement - and implementation of the INA-240 - occurs when the device is placed w/in the motor's Phase line(s) - where it IS exposed to (and minimizes) the negative effects of such, "severe common mode swings."

Accordingly the INA240 front end claims it filters most all FET delta dv/dt by design and according to the forum FE it is exposed to even more CMM swing in line versus low side were CMM is virtually Zero. FE has yet to explain how delta dv/dt is passing into the 240 output but it is very low level -100mv.  

Hi BP101,

I suspect you know this, but just in case some new users might be confused, the ESD protection diodes do not protect from all voltages up to 2KV. The ESD specification of 2KV is for a specific test with a limited energy.  The details can be found in this JEDEC specification referenced in the datasheet.  JESD22-A114F

As for measuring the "resistance" of the power pins, I simply caution against coming up with pass/fail criteria that are not supported in the data sheet. It is certainly a valid method for measuring pin leakage when the forcing voltage used is low enough, but as you approach the threshold voltage of the transistors, you can get widely varying results. If in this forum we were to say boards that a measure less than 500 Ohms identifies a bad board without defining how the measurement is to be taken, a new user might measure 100mA of current at 3..3V and return a perfectly good board calculating the "resistance" is only 330 Ohms. (I know you understand these things, this is clarification for others who may be following this thread.)

I don't mean to disagree that you have some damaged boards. I think it likely. I am concerned how they came to be damaged. I am watching this forum closely to see if others are having the same problem to identify if we have a manufacturing problem. The one "out of the box" failure does concern me.

I do suggest that you not discount that flyback from switching large loads, particularly if they are inductive, can cause electrical over-stress on the parts if the parts are not adequately protected.

BP101 said:

...diode check or even the rail resistance checks are not valid method(s) that I have done for over 3 decades

Indeed - as vendor's Bob & I have noted - that type "rail" measure proves FAR MORE CHALLENGING TODAY due to the lowered voltages required by vastly smaller IC line & spacing geometries - demanded by the far greater IC complexity!      Somehow - you've missed this clear, well publicized fact - have you not?    (lock-step - 3 decade old tactics (may) require (some) update! ... These ain't your father's MCUs...)

Again - no MCU vendor that firm/I have noted lists - nor advises/advocates for - your rail resistance measures!    Not one!     (likely due to the strict, restrictive requirements imposed upon such measures - and the fact that a (superior) current measurement technique proves far safer, more meaningful & repeatable.)    

As you report a "rash" of board "irregularities" - would it not prove wise for you to invest in a proper, anti-static desk pad, wrist band - and insure that such ESD system is proper.   (ground, especially)    You past mentioned a lesser means of ESD "protection" - yet floor, desk, wrist-band - and proper set-up is noted as the "gold standard" for ESD protection.    (that properly terminated wrist-band - and its disciplined use - is critical...)    Note too that firm/I have (easily) purchased well beyond 200 MCU Eval boards (multiple vendors) have YET to encounter a single one - "DOA!"

As to the usage (placement) of INA-240 - it is clearly intended to monitor the phase currents (where large common mode swings reside).    Little benefit results from your positioning this device in the Low Side FET return.   (once again - you opt for construction "ease" vs. "best practice.")     I'd bet that you scoped the INA's output w/"scope ground lead intact" (full length) - which is (almost) certain to add undesired "pick-up."

Hi Bob,

Bob Crosby said:

I do suggest that you not discount that flyback from switching large loads, particularly if they are inductive, can cause electrical over-stress on the parts if the parts are not adequately protected

Find this an odd point as it is not really considered high voltage when and where it enters the TM4C comparators and VREFA+ is +/-300mv tops. The same PWM flyback surges are called FET switch node spikes in high frequency buck power supplies per Fairchild forum. PWM is only 12.5khz and spikes are randomly occurring and the INA240 is filtering out all the lower PWM delta dv/dt the INA282 let slip through.

Bob Crosby said:

As for measuring the "resistance" of the power pins, I simply caution against coming up with pass/fail criteria that are not supported in the data sheet

I don't recall finding any type of electronic troubleshooting techniques in the data sheet as that is outside the scope of the document. All I can say is a curve tracer does the same basic test and it is an industry wide standard to examine all controller pins for sings of damage. I have replaced thousands of bad CMOS devices that have failed the diode check or resistance curve on one or several pins including VDD. Google B&K curve tracer, a wonderful instrument for detecting bad silicon junctions etc...

Bob Crosby said:

suspect you know this, but just in case some new users might be confused, the ESD protection diodes do not protect from all voltages up to 2KV.

Good to know and was thinking why put TVS diodes on the external I/O if they already exist but have added anyway in critical places such as the analog/comparator shared inputs. Three analog comparators -INN are hard wired to 3 analog channel inputs respectively. Have to wonder if some kind of strange current flow in the MUX was occurring and adding to the VDD issue. Well it is only two launch pads and the replacement with inverter is working great so far no low drop or resistance VDD.