LMV358 (Uno R3) – possible partial damage from 12V PSU?

[jk-72]

Hi, I'm new to this forum and have a bit of a puzzling problem. I just started playing around with the Uno R3, powered from 12V, only to find out that the 5V regulator U1 (AMS1117) burns like hell. I measured the temperature of 74°C on its plastic housing, but there was no reason to draw any high currents from this regulator. An attempt to replicate this 'hot U1' scenario at work (for further investigation) failed because the U1 was cold there.

A separate measurement of the currents drawn from the U1 regulator and USBVCC (I cut some tracks on the PCB to totally separate these 5V sources) revealed that some current is drawn from USBVCC even with 12V PSU on and when 5V USB supply is supposed to be disconnected by T1 MOSFET (the situation at work where U1 regulator was cold). I actually measured 47mA from USBVCC and only 12mA from U1 (should have been 0mA from USBVCC and 59mA from U1).

The reason for drawing current from 5V USB was the MOSFET switch T1 not being opened as it should. When an external PSU is connected, the gate of T1 (U5A pin 1 output) should be close to +5V, but only some +3.7V was measured, low enough for T1 to be closed (VGS = -1.3V). Comparator U5A input levels were correct (+5.83V pin 3, and +3.3V pin 2). I cut the track between U5A pin 1 and the gate of T1 to measure the disconnected U5A output – still only 3.7V, so definitely a faulty U5A.

The slightly different 5V levels between USBVCC and U1 output have explained the measured currents: 4.896V from U1, 4.906V from USB 5V. The 10mV difference between these two 5V lines was large enough for such a high current (47mA) drawn from USBVCC when T1 was closed due to a fault on U5A's faulty output.

The U1 regulator was so hot in my home workshop because of the opposite situation. My USB voltage level, supplied from the USB sharing switch, measured only 4.74V, while the U1 regulator gave 4.896V, a difference of over 150mV. A huge current had to be pumped from the U1 regulator to the USB 5V (I hope my USB sharing switch was not harmed).

The only cause for the damage to the U5A output, using the recommended +12V PSU, seems to me to be a persistent slight overvoltage on the CMP input. 12V PSU gives some 11.66V on VIN (behind diode D1), and 5.83V on CMP input. But the maximum permitted input voltage of the LMV358IDGKR is only about 5.7V. Can this really be an explanation?

At work, we have seven ATE test units, each with an Arduino Uno R3, running for long hours every day, powered by quality 12V external PSUs. They all work fine, no problems. After this strange experience with a damaged U5A, I checked the voltages on the U5A outputs on all 7 Arduinos (when powered from 12V) – all had damaged U5, same fault, voltage on pin 1 measured between +3.7V and +3.8V, but U5A input levels were correct. All seven units have worked for several years with no problems and still work well, even with the partially damaged U5 devices. The problem is that this damage (around the 3.7V output of the comparator) may never be discovered because the Arduinos still have the 'correct' value of the +5V line so they work OK.

My question is – is it possible that Arduino’s recommended input voltage 12V can cause the degradation of the U5 comparator over time, resulting in the output voltage dropping from 5V to just 3.7V, and therefore preventing the USBVCC to be disconnected (via the T1 switch) when an external PSU is used? I’m going to replace all those damaged LMV358IDGKR op-amps and will fit an extra 4V7 zener protection diodes between CMP input and GND. All units still work OK, but it’s not very healthy to have two 5V lines with different voltage levels connected.

Has anyone who uses 12V for Uno R3 experienced this particular failure before? I only discovered this by accident when the super-hot regulator seemed strange to me, so I started looking for an explanation. Just wondering – anyone running the Uno R3 from 12V, could you try checking the voltage level at pin 1 of op-amp U5A?

Thanks for any feedback.

[jremington]

The Uno R3 linear regulator can be easily overloaded, leading to thermal shutdown and possible damage, by drawing excessive current from the 5V pin. I recommend that you not exceed 9V on Vin.

What is connected to that pin, or, what has been connected to the failing unit in the past? Motors are particularly suspect because they inject high voltage spikes into the power leads.

[Wawa]

Good detective work, and I see where you're coming from.
From your observations, I also think the LMV358 is the source of the problem. Is this a clone Uno R3, or original. I know that older high voltage LM358 opamps don't have rail2rail outputs, with about 3.7volt output swing on a 5volt supply. Maybe they used one of these as replacement during production, not realising it will give problems.
5.83volt on the +input should be fine, because of the opamp's internal ~0.7volt clamping diodes (datasheet page16) and the 10k:10k divider that limits input current. Max input current is however not in the datasheet, and this situation should have been avoided with a Schottky diode between +input and VCC of the opamp.
Leo..

[jk-72]

Many thanks for the replies. The boards are genuine green Arduinos with SGM8542 op-amps and black Elegoo with U5 marked 'M5P' (I don't actually know if this is LMV358IDGKR or not), some photos attached.

The thing is, when measuring the brand new unit to verify whether this low 3.7V op-amp output is also there (new board powered from 9V just in case, and no USB connected), the output of U5A is measured +5V as it should be, T1 MOSFET fully open. So all the affected units that are currently in use and only show 3.7V on pin 1 must have originally also had a +5V output.

There is no big load from Arduino's 5V line, it just powers a few relays, the total current should never exceed 200mA. The reason the U1 regulator burned so hot at my home was just because it must have pumped an excessive current into the USBVCC that had a lower 5V level.

Using a 9V PSU instead of 12V is definitely the best way to prevent these failures from happening again, but since 12V was used at the workplace and since 12V is still within the recommended Arduino input voltage range, it would be good to know if these U5A damages (which often don’t affect the functionality at all) could be caused only by that upper input voltage limit of 12V PSU, or if the cause is somewhere else.

[Wawa]

@jk-72

There is no big load from Arduino's 5V line, it just powers a few relays, the total current should never exceed 200mA.

That 12volt rating is in the danger zone, and only ok without anything else powered from the Uno.
200mA draw pushes the thermal limitation of the 5volt regulator when on a 12volt supply.
200mA + 50mA for the Uno is 250mA * 7volt across the regulator = 1.75watt.
The 5volt regulator can't dissipate more than 1watt without a proper heatsink.

I don't see how this could affect the output of the opamp though.
If you can, try to replace the opamp (needs a hot air rework tool) and let us know.
I still think that 5.83volt input could be a cause, which is a design flaw.

The best way to power an Uno R3 is with a 5volt cellphone charger, connected to the USB socket. That bypasses the 5volt regulator.
Leo..

[jk-72]

Thanks Leo. That 200mA current consumption from Arduino's 5V is just an exaggerated estimate including the power of the board itself, in reality it should be much less. I'm inclined to agree with you that the 5.83V op-amp input voltage could be the cause, but only because I don't see anything else there. And if that is a case, then I think it would be really useful to know whether anyone else who regularly runs the Uno R3 from 12V also has U5A damaged in the same way, with the output voltage on pin 1 low enough to keep T1 switch closed even though it should be open.

Based on experience with seven ATE units in my workplace (7 Arduinos), this damage probably won’t affect any external functionality at all, so it can easily go undetected forever, and maybe it is more widespread 'problem' than expected (?).

Next week I'll replace all faulty op-amps (already bought LMV358IDGKR as per the Arduino datasheet), will fit 4V7 zeners between CMP input and GND, maybe add some shottky between CMP and +5V too, and let here know. Thanks.

[Wawa]

@jk-72

will fit 4V7 zeners between CMP input and GND,

Forget about zeners.
A Schottky clamping diode between +input and opamp supply is much better.
The diode will always keep input at or below "VCC +0.3volt", whatever VCC might be.
A zener is useless when the 5volt regulator gets too hot and shuts down (VCC= 0volt).
Leo..

[TomGeorge]

Hi, @jk-72
Welcome to the forum.

@jk-72

Thanks Leo. That 200mA current consumption from Arduino's 5V is just an exaggerated estimate including the power of the board itself, in reality it should be much less.

Can you please post a link to specs/data of your relays?

Can you please post a copy of your circuit, in CAD or a picture of a hand drawn circuit in jpg, png?
Hand drawn and photographed is perfectly acceptable.
Please include ALL hardware, power supplies, component names and pin labels.

How are you driving the relays, via BJT or MOSFET?
You cannot drive relays directly of the UNO output pins, look up the spec for the UNO board.

UNO R3 | Arduino Documentation

Look at the tech specs.

Tom...

[jk-72]

''...I know that older high voltage LM358 opamps don't have rail2rail outputs, with about 3.7volt output swing on a 5volt supply. Maybe they used one of these as replacement during production, not realising it will give problems...''

---

Spot on, that's it. No input overvoltage, no high currents, no damaged op-amp to blame, just a silly manufacturer's fault to fit the wrong part. Tracking the device marking revealed the culprit. LMV358IDGKR is labeled R5B, R5Q or R5R, but the troubled comparator shows M5P or M5L, which is actually the LM358DGKR – a high-voltage device that does not have a rail to rail output indeed.

I wrongly blamed the op amp for being damaged, just because the brand-new Arduino I was testing worked as it should. Not realizing that different op-amp had been fitted there. The new Arduino was the genuine green one, equipped with SGM8542, which is not LMV358IDGKR, but it is a rail to rail device. However, all of our Arduinos that are used in the ATE units are Elegoo black boards and they all have the wrong high voltage op-amp. I wonder if it was just a short period of time that the wrong parts were installed there or if they still use them.

Today I also tested new set of LMV358IDGKR op-amps I bought (device marking R5R), and all good, so I'll just replace the high voltage op-amps from the Elegoo boards with these (to keep T1 switch busy working as it should) and job's done. I will probably also install an extra protection schottky diode there between the CMP input and +5V, just in case.

Thank you all for your helpful tips, really appreciated.
Jarda

[raschemmel]

You can test any op amp in circuit by shorting the V- input (2 or 6 ussually) to the output (1 or 7 depending on the pinout. This is easily done by soldering a 30 guage wire from the inverting input to the output. What this does is change the configuration to a VOLTAGE FOLLOWER, which by definition means the output follows the voltage on the NON-INVERTING input. I have used this method to troubleshoot op amps in circuit many times and it always works (unless of course the output pin is shorted to GND or some other signal). The device you mentioned is a RAIL-TO-RAIL op amp configured as an open loop comparator so the output should be one of the two rails depending on the input. Since 5.38 is > 3.3V the output should be the positive rail,
assuming of course the power pin actually has 5V on it.

[Wawa]

@jk-72

Elegoo black boards and they all have the wrong high voltage op-amp.

Did you file a complaint? (I would).
Leo..

[jk-72]

No harm done, plus these Elegoo boards were purchased over 4 years ago, so no point in trying to claim them this time. But I will keep an eye to check the U5 op-amp markings and its output voltages on each new Arduino.