Voltage regulator NOT regulating?

hello,has anyone every had a issue with the voltage regulator not regulating at the manufacturer rated specs? i am using a ams1117 3.3v voltage reg,but it keeps coming out at 4.53 volts on the meter and only 5v 1amp is being supplied the chip goes from room temp to being able to fry a egg on in a matter of less than a minute,im sure this might be a issue with the traces on the board but also why would it produce more voltage than it says on the data sheet if its within the specs defined my the manufacturer? i also changed out the voltage reg to make sure it wasn’t a bad reg and same problem.

also check voltage supply it was at 5v 1 amp from the power point on the board but the reg is giving me slack could someone point me in a few area’s i may not have thought of,or is beyond my expertise?

thanks in advance

Maybe it acts as a diode when the GND is open?

How much current is going though it? How hot does it get? Where did you buy it?

If you bought it on Ali / Ebay / etc, it could even turn into a bowl of petunia’s if you look cross eyed for a few moments. Maybe it’s just broken. Have you tried replacing it?

Another possibility is instability resulting in oscillation because of bad quality or wrong value output capacitor or loading. AMD1117 is quite an old design and though often adequate, it’s not a very good regulator. Have you measured the output with an oscilloscope? Is there any ripple on the output?

1 Like

not sure about acting as a diode i have it being powered by 5v 1 amp type c usb, and to answer questions in sequence, amazon,strongly agree,this is the 2nd voltage reg ive replaced when i first soldered it on a few days ago the readings were correct then i accidentally powered the board with a 5v 2.5 amp power brick desoldered it added new one now the voltage is 4.53 v which is very weird i could understand less but if its more its kinda hard to think of a solution i wish i could just slap a resistor in there but its powering a esp8266 so it needs its 3.3v so it might be fried also now (yay!)

and sadly i dont own a oscilloscope im bad at pretending to be a engineer when i measure with a multmeter the output is pretty consistent though,i dont think the problem is with board but if i get a little restless i might just hand solder some wire in and splice a cord to power the regulator and see if that might offer some clarity

1 Like
  1. Do you have capacitors on it’s output?
  2. What type and size?
  3. How long is the wiring to those capacitors?
  4. Does the output voltage change if you put a capacitor (100uF elco or so) between output and GND (This may be an indication of instability, but it’s a guess).

1,yes smoothing caps both on input and out put
2. smd 10uf non talatum?
3. less that a centemeter
4.i might have to desolder one to check that but i will def try in a bit

im going to slap this here just to leave no stone un-turned

please try to ignore all the other crap but the is the powering circ to my board can post more if needed

1 Like

2). Datasheet suggests 22uF minimum.
2a). Ceramic capacitors can have a too low ESR and also cause instabilities.
4). I did not mean replace, but add more capacitance.
5). Another quite possible cause is a too small output current. If your uC goes to sleep and draws less then 10mA the output voltage of the AMS1117 will rise.

2 Likes

2.i am going to de-solder and do that immediately
2a. didnt know that but i will forever keep that in mind now
4.i caught that
5.,the mcu im using is pretty hungry especially on boot i dont think it would/should have enough time to rise like that but its good to remember this in the future for deep sleep to keep in mind the head but hopefully i can dedicate more time to learn and incorporate the better non linear regulators that dont dissipate with head

im going to try the 22uf first and come back and tell you how it goes be back in about 30 mins

Hi @Do_not_sleep
Is this a top view or a bottom view of the reg?

1 Like

Blue = Back (See also the Layers Manager).

2). You don’t have to desolder. You can stack SMT capacitors (and resistors) on top of each other.

2). Also, for quick test you can use a radial Electrolytic capacitor and then just hold it in the right place. If this changes the apparent DC voltage there is something weird happening.

1 Like

its on the back layer so its reversed

if anyone has any questions on the board please let me know i can explain anything to help clarify

I have no incentive to go looking at a low res screenshot or compare pinouts. Besides that, my mind just shuts off when people start calling themself dumb.

When I saw your AMS1117 with 10 thermal via’s, I assumed they were connected to a copper plane that doubles as a heatsink, but the copper plane was turned off. But I do not see the edge of the copper plane. The antenna of the ESP8266 does not properly when it’s too close to a copper zone.

I’m being very polite by using the term “GOSH!”

This exclamation is what happens with extreme brain fade.
Edited my post and put the whole matter down to Monday-itis.

Careful!! Those letters can be re-arranged to spell “SHOG”!! :grinning:

Generally the first thing to do is to check the voltage directly on the pins of the chip. Ground pin must be grounded; etc. I am not familiar with this particular supplier (of what is probably an LT1117 cross) but the default output current is 0 mA on this device. So it appears that no minimum load is needed.

BTW I have not used LT1117. But I have used LM317 a few notable times. The LM317 is a great device when its limits are observed. Very stable. I have used it as a combination voltage reference and regulator in a production battery charger. I have used it as a post regulator which regulates and filters the output of a production 48V switching power supply with low output noise.

These ICs were around before larger value ceramic capacitors were commonly used. Tantalum and electrolytic capacitors have higher equivalent series resistance (ESR) than ceramic capacitors, and this device might not work properly with only a 22 uF ceramic capacitor on the output for example. I would choose a “low ESR” or “low impedance” electrolytic or tantalum or polymer capacitor; probably 47 uF or more and rated for 6V3 or 10V0. Then an additional 1 uF ceramic may be beneficial.

Beware that you are flirting with the classic electronic malfunction. That malfunction is when regulators oscillate and oscillators don’t.

2 Likes

i get it,i get it.esp07 has a u.fl for external antenna thats the plan.thanks for the help

I’ve often uses 1117-3v3 regulators for ESP boards and never had any issue. Used all kinds of 10uF capacitors: ceramic, tantalum, electrolyte. All worked absolutely fine.

The wiring in your screenshot looks okay-ish. Component placement could be better, but there’s no obvious error.

Can’t really help you here, maybe you have a bad batch. Or some bad solder connection. if possible post a good quality picture of your board.

2 Likes

But not usable here.
It is stable as all classic voltage regulators with voltage-drop around 2…2.5V.
Voltage reguilators with voltage drop of 1V3 (1117 serie) can be more demanding when it comes to stability, but there is typically no problem with them.
The problems begins when you came to regulators with pnp or PMOS as main transistor. They allow to get voltage drop of 0.1… 0.3V but at the cost of stability problems.

1 Like

LM317 can be usable if the current is low enough.
With an allowable voltage drop of 5-3.3 = 1.7V you can get about 200mA out of it when it’s cold, and 500mA when it’s hot.

image

I also checked AMS1117 and LM1117 datasheets.
AMS1117 does not mention the pass element, but LM1117 has a block diagram and to my surprise it shows indeed an NPN transistor. So it’s my bad, this thing is probably a lot more stable then I first thought it was.

But I did search the internet a bit yesterday and there are several reports of various instabilities with this thing. Of these often used part numbers there are often clones and fakes and if you buy stuff from shady websites such as amazon you can only guess as to what you actually get delivered.

1 Like

Might be possible. Probably a good idea to test them first standalone with an adjustable load (like maybe a potentiometer) and see if they work according to expectations.

I bought mine from a large supplier (think mouser) and as i said they work absolutely fine. If you get them from Amazon or Ebay, they could be selling fake ones or rejects.

The ESP boards themselves also have caps and their own voltage regulator from 3.3V to about 1.1V I think, so a small ripple due to bad caps shouldn’t matter that much.

1 Like

This is how electronics recycling is done in Guiyu, China.