-
HELP!!! What is wrong with this schematic AMS1117 is burning all the time
Posted by HackerMy on 03 Apr, 2021 14:43 -
What is wrong with this schematic?!?!?! It is working with no problems for days without the capacitor[C32], but when i install the Capacitor C32, AMS1117 starts heating instantly and burn in a matter of a seconds(1-2sec). It is burning even with nothing attached to it(no sensors, no LCD display, no outputs...). I tried with Electrolytic 470uF and with Tantal 330uF[in datasheet the recommend Tantal 22uF, but i have only Tantal 330uF] - there is no difference - the AMS1117 is still burning.
Thank You!
The easier to read schematic:
-
1117 regulators can oscillate or behave in a weird way if the ESR of the capacitor on the output is outside some safe range, typically 0.1 ohm ... 1 ohm
Some particular 1117 regulators are "optimized" to support ceramic capacitors (which have esr well below 0.1 ohm) but if you want your design to be safe if user replaces IC with 1117 regulator from other brand, it's best to pick a capacitor with esr within that range.
Your 470uF 16v may have esr much lower than 0.1 ohm, and that could cause problems. You also don't really need that much capacitance... Consider sticking to something like 47uF ...150 uF 10v..16v electrolytic (maybe also 25v or higher, with lower capacitance values, it's all about the esr of the chosen capacitor series)... you really don't need much more with linear regulators.
Most datasheets will suggest 10uF but I suspect they think of 10uF in their circuits as tantalum capacitors as those had esr around 0.2-0.4 ohm . 10uF electrolytics may have too high esr (over 1 ohm), hence why I suggest going with 47..150 uF, a range where the esr of most series is below 1 ohm and the value is not a super high amount.... and maybe you reuse the same part somewhere else in your design saving you money.
You should also add some capacitance on the input ... a 0.01uF ceramic and maybe 10..47uF 25v electrolytic on the input should be plenty. It would probably work with a 0.01uF and a 1uF...10uF ceramic rated for at least 25v or more, in parallel.
Also, consider replacing the linear regulator altogether ... there's quite a drop from 12v down to 5v , so that difference gets dissipated into the chip and circuit board as heat. With even 50mA, you're dissipating (12-5)x0.05 = 0.35 watts
For example, have a look at AP62200 : https://www.digikey.com/en/products/detail/diodes-incorporated/AP62200Z6-7/10491593
You only need a standard inductor and a bunch of ceramic resistors and capacitors. Datasheet contains example for 12v->5v circuit.
There's also AP63205 which is fixed 5v output, so saves you a bunch of resistors : https://www.digikey.com/en/products/detail/diodes-incorporated/AP63205WU-7/9858424
-
Thank You for the replay.
The funny thing is that i am using the same idea with others project, and the capacitors there are 470uF/16V or 25V without problems. The only differance between the working ones and this one is: with the last one i am using the RESET-pin as a Output!?!?!?! And this is really the only difference. I do not get it... -
First of all - do not make schematics like that one with each component separate, it is extremely hard to read, connect the components with wires and only label the wires that connect to some other circuits not showed on that particular schematic sheet.
Now, in addition to the previous post, you have to provide a minimum load to the regulator output - look in the datasheet for minimum load specification. You can connect simply a resistor across the output , which should be sized to draw more than minimum current. Say, a 500 Ohm resistor will do in this case. -
Scatttergrams are fine for your PCB work but a nightmare to follow for others. Rein in your private thinking guys.
-
I redraw the schematic...hope it is easier to read now... [ Specified attachment is not available ]
-
I'd encourage you to stop using the 1117 regulators altogether.
There's other linear regulators which aren't so finicky about capacitors work even with ceramic capacitors well
Here's some examples:
AP7383: https://www.diodes.com/assets/Datasheets/AP7383.pdf
up to 30v input, max 150mA output (in fixed voltages versions):
Link to 5v fixed version : https://www.digikey.com/en/products/detail/diodes-incorporated/AP7383-50Y-13/9765672
Also AP7370 (max 18v in, 5v fixed output, up to 300mA): https://www.digikey.com/en/products/detail/diodes-incorporated/AP7370-50Y-13/10235829
MIC5225 : https://ww1.microchip.com/downloads/en/DeviceDoc/mic5225.pdf
up to 16v input, max 150mA output, fixed and adjustable versions, very low voltage drop (~ 310mV at 150mA), stable with ceramic capacitors.
Link to adjustable version : https://www.digikey.com/en/products/detail/microchip-technology/MIC5225YM5-TR/1834698
Link to 5v fixed version : https://www.digikey.com/en/products/detail/microchip-technology/MIC5225-5-0YM5-TR/1834697
MCP1703 : https://ww1.microchip.com/downloads/en/DeviceDoc/22049f.pdf
up to 16v in, fixed versions incl. 5v, up to 250mA
link to 5v fixed :
sot223: https://www.digikey.com/en/products/detail/microchip-technology/MCP1703T-5002E-DB/1635602
sot23-3: https://www.digikey.com/en/products/detail/microchip-technology/MCP1703AT-5002E-CB/3131480
and if you need more current there's always the classic LM317 and 78xx regulators, but some of these may not be stable with ceramic capacitors. -
I would also consider as a possible cause of the fault the complete lack of input capacitance: it could be that the inrush current required to charge the output capacitance caused the regulator to be exposed to an overvoltage transient, depending on the length of your power supply wires, or maybe it is just triggering weird oscillations. You typically want at least some capacitance at the input of the regulator.
-
oscillation frequency dear. Good thing is to read datasheet!
-
AMS1117 heats because it auto-oscillates. That happens when the capacitor of the output has a too low ESR.
Put a 1 ohm or so resistor in series with C32/470uF (also it is way too big, datasheet recommends a 10uF tantal capacitor, not 470uF), or use another capacitor with higher ESR. https://datasheet.lcsc.com/szlcsc/2001081204_Shikues-AMS1117-1-2_C475600.pdf
That capacitor C32 is NOT there to act like an energy tank (its purpose is NOT the same as the capacitors found after a rectifying bridge). C32 is for the stability of the 1117 only. If its ESR is too low or too high, the stabilizer will oscillate. That is why an Al electrolytic capacitor (not low ESR) is preferable to a MLCC that might have very low ESR. -
I catch an issue with AMS1117, it was burned out after short circuit. So I replaced it with ST LD1117A, works ok
-
Seeing as your input voltage is high relative to output you should just go with a l78m05 in dpak, it's the same as a trustworthy 7805(except 500-800ma max current) just in a smaller package and has none of the weird capacitor issues the 1117 has. LDO regulators are only useful if the input voltage is close to the output voltage, if there's a big difference just use the old trustworthy 780x series.
-
Seeing as your input voltage is high relative to output you should just go with a l78m05 in dpak, it's the same as a trustworthy 7805(except 500-800ma max current) just in a smaller package
They have the same small DPAK SMT and large TO-220 TH packages. -
Could it be that the amps flowing through the regulator is too damn high (whatever is connected to out rst and other 2 jumpers)
The limit to asm1117 is I think (read datasheet before trusting me) is 800mA and 78l05 won't help either with it's 100mA limit. 7805 has 1amp limit. But do also know that the limit is not for continuous load.
But do also know that tantalums are sometimes marked with darker line on positive. (blew up 2 of mine before I figured it out :p) -
Could it be that the amps flowing through the regulator is too damn high (whatever is connected to out rst and other 2 jumpers)
The limit to asm1117 is I think (read datasheet before trusting me) is 800mA and 78l05 won't help either with it's 100mA limit. 7805 has 1amp limit. But do also know that the limit is not for continuous load.
But do also know that tantalums are sometimes marked with darker line on positive. (blew up 2 of mine before I figured it out :p)
Ceramic capacitors have very low ESR. Tantalum capacitors are somewhere between ceramic capacitors and electrolytic capacitors. Almost all solid (polymer) capacitors have ESR lower than tantalum capacitors.
So the 100nF ceramic capacitor will not help the stability of 1117 regulator, but can help filter at some frequencies.
As I said in my post and as others have said, you either need to
1. use a capacitor series with higher ESR (typically ESR scales with capacitance and volume, hence why I suggested sticking within 47...150uF as less than 47uF could have more than 1 ohm ESR and more than 100-150 can be too low) OR
2. artificially increase the ESR by adding a resistor in series with the capacitor on the output ... if I'm forced to use 1117 regulators with ceramic capacitors I'd use a 0.2 ..1 ohm resistor
But in the end as I suggested in another post above, if you don't need that much current, you can use smarter / better linear regulators which can work with ceramic capacitors which are much cheaper than an electrolytic capacitor or tantalum capacitor and use less space and you can do lower height designs (you may have to lay electrolytic capacitors flat or use multiple lower height one in some designs) and can be better for some circuits (ex less chance of being ripped off the board or leak electrolyte damaging the circuit)
The 1117 are super cheap, for example let's say 10 cents each... and the other ldos may be 25-50 cents each, but include the price of capacitor and the space on board saved and you're nearly at the same price.
Another tip : I could be a good idea to add a resistor in front of the linear regulator, or at least a resistor footprint - you can add a 0 ohm resistor or a jumper link if you don't want to implement this.
The resistor can drop some voltage across it, giving the linear regulator less voltage to work with and therefore heat less.
For example, the schematic has 12v in and 5v out and let's say you want at least 8v to reach the regulator at 0.20 A of current ... You have the formula V = I x R so R = (12v - 8v ) / 0.20 = 4/0.20 = 20 ohm .... so let's say you go with either 18 or 22 ohm resistor
The power dissipated in the resistor will be P = I x I x R = 0.20 x 0.20 x 18 = 0.72 watts so you'd would probably be fine with a 1w rated resistor.
There's 1w rated surface mount resistors in 1206 footprint.. and bigger footprint ones are super cheap : https://www.digikey.com/short/4fwvv30f
And now, your regulator only needs to drop 8v-5v = 3v at 0.20a so it would only dissipate 0.6 watts.
-
Thank You for the replay.
The funny thing is that i am using the same idea with others project, and the capacitors there are 470uF/16V or 25V without problems. The only differance between the working ones and this one is: with the last one i am using the RESET-pin as a Output!?!?!?! And this is really the only difference. I do not get it...
The RESET function can be disabled by programming the right fuse on the chip, which lets you use it as a weak I/O pin. It's weaker than the other I/O pins because that pin is designed to tolerate 12V for high-voltage programming. Once you disable RESET, you can no longer use the ISP method of programming. The high-voltage programming method still works, but you may not be equipped to do that.
If you have a known-good binary, this isn't an obstacle. Just program it normally, THEN change fuses.
If you're doing software development, it's another story... Using a larger AVR with more pins might be advisable until the final stages. -
Thank You for the advices....The 47uF capacitor solved the problem with burning AMS1117. I added also 100nF on the input(12V) just in case. I prepared 3 more PCBs, and the interesting thing is that AMS1117 has different Vout - the first one has 4.96V, the second 5.65V, the thirth has 5.25V, and the forth one is 7.65V(but i assumed that this one is defective, and i replaced it with new one). Is it possible that those AMS1117 are Chinese cheap clones...How is this even possible unless AMS1117 are fake....
Ceramic capacitors have very low ESR. Tantalum capacitors are somewhere between ceramic capacitors and electrolytic capacitors. Almost all solid (polymer) capacitors have ESR lower than tantalum capacitors.
So the 100nF ceramic capacitor will not help the stability of 1117 regulator, but can help filter at some frequencies.
As I said in my post and as others have said, you either need to
1. use a capacitor series with higher ESR (typically ESR scales with capacitance and volume, hence why I suggested sticking within 47...150uF as less than 47uF could have more than 1 ohm ESR and more than 100-150 can be too low) OR
2. artificially increase the ESR by adding a resistor in series with the capacitor on the output ... if I'm forced to use 1117 regulators with ceramic capacitors I'd use a 0.2 ..1 ohm resistor
But in the end as I suggested in another post above, if you don't need that much current, you can use smarter / better linear regulators which can work with ceramic capacitors which are much cheaper than an electrolytic capacitor or tantalum capacitor and use less space and you can do lower height designs (you may have to lay electrolytic capacitors flat or use multiple lower height one in some designs) and can be better for some circuits (ex less chance of being ripped off the board or leak electrolyte damaging the circuit)
The 1117 are super cheap, for example let's say 10 cents each... and the other ldos may be 25-50 cents each, but include the price of capacitor and the space on board saved and you're nearly at the same price.
Another tip : I could be a good idea to add a resistor in front of the linear regulator, or at least a resistor footprint - you can add a 0 ohm resistor or a jumper link if you don't want to implement this.
The resistor can drop some voltage across it, giving the linear regulator less voltage to work with and therefore heat less.
For example, the schematic has 12v in and 5v out and let's say you want at least 8v to reach the regulator at 0.20 A of current ... You have the formula V = I x R so R = (12v - 9v ) / 0.25 = 4/0.20 = 20 ohm .... so let's say you go with either 18 or 22 ohm resistor
The power dissipated in the resistor will be P = I x I x R = 0.20 x 0.20 x 18 = 0.72 watts so you'd would probably be fine with a 1w rated resistor.
There's 1w rated surface mount resistors in 1206 footprint.. and bigger footprint ones are super cheap : https://www.digikey.com/short/4fwvv30f
And now, your regulator only needs to drop 8v-5v = 3v at 0.20a so it would only dissipate 0.6 watts.
I am considering also to put a diode on the 12V before the regulator...like that(i think) will drop the voltage a little bit, and will serve also as a reverse polarity protection. The consumption of the entire PCB is less than 100mA [by calculations shouldn't be more than 65mA]. I still do not understand how 470uF capacitor working on the others PCBs i have already made before-it is at the front of my eyes right now working for months without any problem. I use 12V/5A laptop power supply - it should be pretty stable... -
Thank You for the replay.
The funny thing is that i am using the same idea with others project, and the capacitors there are 470uF/16V or 25V without problems. The only differance between the working ones and this one is: with the last one i am using the RESET-pin as a Output!?!?!?! And this is really the only difference. I do not get it...
The RESET function can be disabled by programming the right fuse on the chip, which lets you use it as a weak I/O pin. It's weaker than the other I/O pins because that pin is designed to tolerate 12V for high-voltage programming. Once you disable RESET, you can no longer use the ISP method of programming. The high-voltage programming method still works, but you may not be equipped to do that.
If you have a known-good binary, this isn't an obstacle. Just program it normally, THEN change fuses.
If you're doing software development, it's another story... Using a larger AVR with more pins might be advisable until the final stages.
Disabling and enabling the RESET fuse is not an issue. I have all the equipment to do that and much more.
Thank You! -
I prepared 3 more PCBs, and the interesting thing is that AMS1117 has different Vout - the first one has 4.96V, the second 5.65V, the thirth has 5.25V, and the forth one is 7.65V(but i assumed that this one is defective, and i replaced it with new one). Is it possible that those AMS1117 are Chinese cheap clones...How is this even possible unless AMS1117 are fake....
If they measure apparently different voltages, then most probabbly they still oscillate. I don't think fake AMS1117 exist, they are too cheap to worth the effort of faking them.
It is possible they apparently don't work because the datasheet specifications were not met:- at least 2...10mA of load at all times (add a LED if the load is a microcontroller that can consume less, e.g. when in standby) <--- see minimal load current in the 1117's datasheet
- 10uF tantal (or electrolytic, or any other type that does not have low ESR - something between 0.2 ... 2 ohms should work). Again, the output capacitor does not have to be BIG, bigger is not better in this case, that capacitor does NOT act like the filtering capacitor after a rectifying bridge, and it doesn't have much to do with the load.
The output capacitor here is not a filtering capacitor, therefore bigger does not mean better.
If the above doesn't make any sense, just put at the output a 10uF tantal, and a 10mA LED to met the minimum required current, and all the 1117 should stabilize rock solid 5V output.
Resist the temptations of adding a big "filtering" capacitor, because that's not a filtering capacitor. -
I don't think fake AMS1117 exist, they are too cheap to worth the effort of faking them.
-
1117 is so old design that it's reverse engineered and made by lots of companies. Some may try to reduce the die size (and therefore make more dies out of a wafer and make more profit) by removing some things like various protections, or reducing the size of the transistor(s), going cheap on the internal reference voltage, even making the ground tab thinner on to220 care save pennies.
They can easily wipe the text on the top of a chip and put AMS on it, just because it sells better or because lots of people search for them because they fail often in Vizio TVs and other brand TVs (which use same or similar power supplies, often same OEMs)QuoteThank You for the advices....The 47uF capacitor solved the problem with burning AMS1117. I added also 100nF on the input(12V) just in case. I prepared 3 more PCBs, and the interesting thing is that AMS1117 has different Vout - the first one has 4.96V, the second 5.65V, the thirth has 5.25V, and the forth one is 7.65V(but i assumed that this one is defective, and i replaced it with new one). Is it possible that those AMS1117 are Chinese cheap clones...How is this even possible unless AMS1117 are fake....
I'd suggest adding a bigger electrolytic capacitor on the input, something like 47..100uF 16v or better should be fine. The long cable of a dc/wallwart adapter can have inductance (a very tiny amount) and that can cause issues if your product's power consumption varies.
And yeah, a minimum load would help.. for example a basic on/off led ... get a red led and a 1000 ohm resistor in series with it, and that should consume around 5mA and that should be good enough as a minimum load.
Or just pick a better linear regulator. -
the interesting thing is that AMS1117 has different Vout - the first one has 4.96V, the second 5.65V, the thirth has 5.25V, and the forth one is 7.65V(but i assumed that this one is defective, and i replaced it with new one). Is it possible that those AMS1117 are Chinese cheap clones...How is this even possible unless AMS1117 are fake....
Only 4.96v is within spec, the rest are not. Dunno if they oscillate or are counterfeit but there are plenty of counterfeit vreg ICs in the wild. -
Here's a test of some low quality copies:
https://goughlui.com/2021/03/27/note-linear-regulator-woes-when-is-an-ams1117-not-an-ams1117/ -
It depends on the part number. There are adjustable versions.OP should pay attention to the part number.the interesting thing is that AMS1117 has different Vout - the first one has 4.96V, the second 5.65V, the thirth has 5.25V, and the forth one is 7.65V(but i assumed that this one is defective, and i replaced it with new one). Is it possible that those AMS1117 are Chinese cheap clones...How is this even possible unless AMS1117 are fake....
Only 4.96v is within spec, the rest are not. Dunno if they oscillate or are counterfeit but there are plenty of counterfeit vreg ICs in the wild. -
It depends on the part number. There are adjustable versions.OP should pay attention to the part number.
He uses fixed 5V version. If you accidentally use adjustable version instead, you will get around 1.25V output voltage.