tl; dr: LED driver board's vias were faulty and would gain tens of ohms when board heated up. Via for ground to resistor divider of voltage feedback would heat up from nearby inverter components, and ground voltage would jump several tenths of a volt, triggering overvoltage and disabling LED strings until all were off. Fix is to run extra ground wire to that point.
I saw this broken Acer S271HL bid monitor on craigslist for $30 and figured I'd give repairing it a shot. First I found that there's very little information about problems, repair, or available parts for this model. The datasheet for the MP3394 LED controller was of great help in understanding operation.
The symptoms: turned on fine, but after a few minutes or less one or more backlight segments would go dark, leaving the monitor with only part of the bottom edge illuminated, as if with spotlights on a wall at night. Eventually or immediately the last light would go dark. Flashlight showed image, so it was just the backlight failing. Turning off and on would sometimes get the lights for a minute or so more, though they lasted less time and sometimes it just blinked on for a moment.
Inside the case, voltages were good: input 19V, 5V and 3.3V on the signal board, 19V on the LED driver board, and when the LEDs were working, 39V going to the LED common. Inverter components checked out: diode, inductor mosfet. LED return voltages were decent: 0.57V, 0.82V, 1.11V, 1.11V.
I checked the LED strings by using a 36V supply through a 1K resistor to power each string separately, and measuring the voltage drop across the string. Each gave a 31.2V drop, very close to each other, and each lit with similar brightness (dim since we're running very little current through). Through the 1K resistor this comes to 4.8mA, so their voltage drop will be a little less than normal. This uses an M270HGE-L10 panel, specified at 38.4V typical string voltage at 100mA. Each string has 12 LEDs. I didn't want to use much current and damage them, and this test gave good evidence that they were working, since they gave such similar readings. Given that sometimes the strings would go individually dark, not all at once, if it were due a problem with the LEDS then all of them would have to be bad, either due to a defect, problem with common lines, or damage done to all due to problem. So I focused on the driver board.
The LED driver does many safety checks and shuts things down if any fail, in order to avoid damage/fire. One thing it checks for is an LED string shorting, or going open. It regulates the current to each string and it must detect these conditions to avoid putting excessive current. The strings are common anode, and the ground returns from each go to open-drain drivers. It regulates the overall LED voltage based on the string with the largest voltage drop, because that one will be closest to ground and otherwise it wouldn't be able to bring about enough of a drop. I'll call this the dominant string. If the voltage drop becomes too great, it will need to raise the LED drive voltage too high. There is an over-voltage protection pin (OVP) to detect this. It is driven by a simple voltage divider, delivering around 1.23V at the highest allowable LED driver voltage (configured to 46.07V in this case).
This is where the core problem was. The divider has two resistors, 300k from the LED drive voltage side, and 8.22k to ground. The ground side of the resistor connects to ground through a via to the ground plane on the underside of the board. It's near the switcher MOSFET, diode, and inductor. The board heats up in operation, and the via begins to increase in resistance, and make sudden jumps to much greater resistance. Running a hair dryer on the board would trigger the fault in a couple of seconds. Looking at the OVP line, it would be fine for a while, then start to rise, and eventually jump up and down by several tenths of a volt. All the while, the actual switcher voltage was pretty stable, not jumping around like this. This made the driver think that the dominant string had too great a voltage drop (and thus was open), so the driver disabled this string and went to the next, and so on, until all were off.
For my repair, I soldered a ground wire to this OVP divider resistor to bypass the crappy via. I also added wires for two of the LED drive lines, which also went through two vias each and were showing tens of ohms when the board was warm. I further added one for the LED high side voltage to the capacitors, and one more from the capacitor ground to the ground of the LED drive chip, since all the LED loads return through it. I added these in the order of LED drive lines, which didn't fix it, then the one to chip ground and LED high side, which still didn't fix it, and finally to the OVP divider, which did.
I didn't run lines for all the vias, as there were no other signals through vias that seemed they would be much affected. The ground/positive ones have dozens, so one failing won't be a big deal. The enable and PWM dim control lines do go through several vias, but they also go through resistors so extra resistance won't be an issue. Though, if they start failing by opening, it could be an issue. They are father from sources of heat as well.
The previous owner and tried to repair it, replacing the caps, so unfortunately I don't know for sure what the original electrolytics were. I replaced a couple with some better units, though I lack any good low-ESR caps to properly recap it with. The unit was only a year old so the original caps were probably fine.
This was a very time-intensive repair. I spent over a dozen hours trying to make any sense before I found the via issue, and even then spent several hours before I found the OVP issue. I learned a lot about the LED driver and got to home my o-scope skills. At least I can share my findings to hopefully save someone else from this ordeal.
Had I needed to replace the LED controller chip, I found that part number MP3394ES was the proper one (SOIC16) and got more hits on suppliers than MP3394 alone.
I'm going to have to open it up again to add a circuit (attiny85-based most likely) to allow further reduction of backlight brightness, as it's too damn bright even on 0. Datasheet shows that at 1kHz PWM rate it uses, it can go down to 3%, while currently Acer only has it go down to about 20%.
The LED connector pinout is 1, 2, comm, comm, 3, 4. Common anode (positive). String 1 (black wire) is the one on the right-most edge when facing the front of the monitor, 4 on the left-most edge.
Thanks everyone for making this forum a great resource, and I hope this is helpful to anyone repairing an Acer monitor with a similar problem. What a maddening problem with the PCB.
Pictures show voltages when working. LED connector voltages on 1, 2, 5, and 6 will vary; one should be around 0.35V, and the others greater by up to a volt or so.
As the first shows, the backlight is along the bottom strip of the panel only, with the light guide spreading it evenly across. Each string of LEDs lights one quarter of the edge.
I saw this broken Acer S271HL bid monitor on craigslist for $30 and figured I'd give repairing it a shot. First I found that there's very little information about problems, repair, or available parts for this model. The datasheet for the MP3394 LED controller was of great help in understanding operation.
The symptoms: turned on fine, but after a few minutes or less one or more backlight segments would go dark, leaving the monitor with only part of the bottom edge illuminated, as if with spotlights on a wall at night. Eventually or immediately the last light would go dark. Flashlight showed image, so it was just the backlight failing. Turning off and on would sometimes get the lights for a minute or so more, though they lasted less time and sometimes it just blinked on for a moment.
Inside the case, voltages were good: input 19V, 5V and 3.3V on the signal board, 19V on the LED driver board, and when the LEDs were working, 39V going to the LED common. Inverter components checked out: diode, inductor mosfet. LED return voltages were decent: 0.57V, 0.82V, 1.11V, 1.11V.
I checked the LED strings by using a 36V supply through a 1K resistor to power each string separately, and measuring the voltage drop across the string. Each gave a 31.2V drop, very close to each other, and each lit with similar brightness (dim since we're running very little current through). Through the 1K resistor this comes to 4.8mA, so their voltage drop will be a little less than normal. This uses an M270HGE-L10 panel, specified at 38.4V typical string voltage at 100mA. Each string has 12 LEDs. I didn't want to use much current and damage them, and this test gave good evidence that they were working, since they gave such similar readings. Given that sometimes the strings would go individually dark, not all at once, if it were due a problem with the LEDS then all of them would have to be bad, either due to a defect, problem with common lines, or damage done to all due to problem. So I focused on the driver board.
The LED driver does many safety checks and shuts things down if any fail, in order to avoid damage/fire. One thing it checks for is an LED string shorting, or going open. It regulates the current to each string and it must detect these conditions to avoid putting excessive current. The strings are common anode, and the ground returns from each go to open-drain drivers. It regulates the overall LED voltage based on the string with the largest voltage drop, because that one will be closest to ground and otherwise it wouldn't be able to bring about enough of a drop. I'll call this the dominant string. If the voltage drop becomes too great, it will need to raise the LED drive voltage too high. There is an over-voltage protection pin (OVP) to detect this. It is driven by a simple voltage divider, delivering around 1.23V at the highest allowable LED driver voltage (configured to 46.07V in this case).
This is where the core problem was. The divider has two resistors, 300k from the LED drive voltage side, and 8.22k to ground. The ground side of the resistor connects to ground through a via to the ground plane on the underside of the board. It's near the switcher MOSFET, diode, and inductor. The board heats up in operation, and the via begins to increase in resistance, and make sudden jumps to much greater resistance. Running a hair dryer on the board would trigger the fault in a couple of seconds. Looking at the OVP line, it would be fine for a while, then start to rise, and eventually jump up and down by several tenths of a volt. All the while, the actual switcher voltage was pretty stable, not jumping around like this. This made the driver think that the dominant string had too great a voltage drop (and thus was open), so the driver disabled this string and went to the next, and so on, until all were off.
For my repair, I soldered a ground wire to this OVP divider resistor to bypass the crappy via. I also added wires for two of the LED drive lines, which also went through two vias each and were showing tens of ohms when the board was warm. I further added one for the LED high side voltage to the capacitors, and one more from the capacitor ground to the ground of the LED drive chip, since all the LED loads return through it. I added these in the order of LED drive lines, which didn't fix it, then the one to chip ground and LED high side, which still didn't fix it, and finally to the OVP divider, which did.
I didn't run lines for all the vias, as there were no other signals through vias that seemed they would be much affected. The ground/positive ones have dozens, so one failing won't be a big deal. The enable and PWM dim control lines do go through several vias, but they also go through resistors so extra resistance won't be an issue. Though, if they start failing by opening, it could be an issue. They are father from sources of heat as well.
The previous owner and tried to repair it, replacing the caps, so unfortunately I don't know for sure what the original electrolytics were. I replaced a couple with some better units, though I lack any good low-ESR caps to properly recap it with. The unit was only a year old so the original caps were probably fine.
This was a very time-intensive repair. I spent over a dozen hours trying to make any sense before I found the via issue, and even then spent several hours before I found the OVP issue. I learned a lot about the LED driver and got to home my o-scope skills. At least I can share my findings to hopefully save someone else from this ordeal.
Had I needed to replace the LED controller chip, I found that part number MP3394ES was the proper one (SOIC16) and got more hits on suppliers than MP3394 alone.
I'm going to have to open it up again to add a circuit (attiny85-based most likely) to allow further reduction of backlight brightness, as it's too damn bright even on 0. Datasheet shows that at 1kHz PWM rate it uses, it can go down to 3%, while currently Acer only has it go down to about 20%.
The LED connector pinout is 1, 2, comm, comm, 3, 4. Common anode (positive). String 1 (black wire) is the one on the right-most edge when facing the front of the monitor, 4 on the left-most edge.
Thanks everyone for making this forum a great resource, and I hope this is helpful to anyone repairing an Acer monitor with a similar problem. What a maddening problem with the PCB.
Pictures show voltages when working. LED connector voltages on 1, 2, 5, and 6 will vary; one should be around 0.35V, and the others greater by up to a volt or so.
As the first shows, the backlight is along the bottom strip of the panel only, with the light guide spreading it evenly across. Each string of LEDs lights one quarter of the edge.
Comment