5VDC to 3.3VDC converter AMS1117

Discussion in 'General Electronics Chat' started by Nissim, Oct 6, 2012.

  1. Nissim

    Thread Starter New Member

    Oct 6, 2012
    7
    0
    Hello
    I have a new PCB design . currently on the board I have AMS1117 converter and
    ARM7 Chip AT91SAM7x512 AU . I feed the converter input 5vdc and I get output of converter 4.1VDC !!. When I disconnect the output pin of the converter from the PCB ,I get 3v3 as expected .
    I double checked pinout of AMS1117 on datasheet and its correctly connected.
    I use external 5VDC stable power source
    I have done the following :
    I checked to shorts on 3.3vdc and 5 vdc nets
    I tried 3 different conveters
    I added 1K resistor between 3.3vdc and gnd but I still get 4.2vdc

    I would appreciate if someone could assist me

    many thanks
     
  2. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    Put a scope on the output of the 1117 when you get a wrong voltage and see if it's oscillating. The DMM will be reading the peak values.

    The 1117 is an industry standard quasi low dropout regulator which needs input and output caps to be stable. 99.9% of the time the problem is wrong type caps, no caps, or wrong value caps.

    You need an input cap too if it is located more than 1 cm from the previous filter cap.
     
    Last edited: Oct 6, 2012
  3. Nissim

    Thread Starter New Member

    Oct 6, 2012
    7
    0
    Thank you very much for your helpful reply. I added 10Uf and a few 1.1uf in parallel . I still getting 4VDC . I suspect I got the wrong LM1117 .
     
  4. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I don't think that is true if what you said before is correct:

    If it was a 4V regulator, it could never put out 3.3V.


    It is possible the layout may have a ground loop so that the voltage rises as the output load current increases.
     
  5. Nissim

    Thread Starter New Member

    Oct 6, 2012
    7
    0
    Thank you for your reply

    I fitted today LM3940 3.3 ( fixed 3v3 regulator)
    On output 3.3Vdc net I have 10UF and 4 caps of 0.1uf caps all caps are in parallel .
    I use oscilloscope and I get clean 5VDC on input and clean 3.98VDC on output
    Currently on my PCB I have at91sam7x512 , LM3940 and caps on input 5vdc and output 3.3vdc net

    With regards to ground loop - My 5vdc is not drawing excessive current
    I cant imagine that GND net with current load ATMEL consumes is have such power drop to alter the output voltage from 3.3vdc to around 4VDC

    Any further ideas to try ?
     
    Last edited: Oct 7, 2012
  6. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    A ground loop is when the physical layout is such that a relatively large current flows along a trace and causes a voltage drop which creates an error in the regulated output voltage.
     
  7. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I believe that there is some component or circuitry which is powered from 5V that is connected to the regulator's output and is forcing current into that line which drives the voltage up. A linear regulator can only source current (not sink) and when some other circuitry forces current into the output line, the voltage rises as you are seeing.
     
  8. Nissim

    Thread Starter New Member

    Oct 6, 2012
    7
    0
    Thank you very much indeed for your prompt reply

    I am not sure if you meant regulator INPUT "5V that is connected to the regulator's output "
    I will connect amp meter in serial to output 3.3vdc and will measure the current
     
  9. bountyhunter

    Well-Known Member

    Sep 7, 2009
    2,498
    507
    I understand you have the 5V connected to the input of the 3.3V regulator.

    But I believe there is something connected on the output that is pulling it up.

    It is probably not an intended connection, it may be a mistake of some kind in the layout or a blown device which has 5V also connected to it.

    I guess this because:

    Your regulator seems to work when it's output is disconnected from the board, but is pulled up to 4V by something when it is connected to the board.

    I would look carefully at every component or connection that goes to the regulator 3.3V output line. Also, anything connected to 5V which is connected to CPU since it could be pulling up the 3.3V line internally.

    I believe the problem is there. This type of problem is very hard to find because it is not something that is supposed to be there.
     
  10. Nissim

    Thread Starter New Member

    Oct 6, 2012
    7
    0
    I have checked all the above and I still can not get 3.3vdc output.
    I read in the datasheet that to ensure fixed voltage output the minimum load current required is 10ma .
    Since the IC cannot sink current, I am wondering if its a good idea to put a resistor ( say 10KR) between Vout to GND .
     
  11. colinb

    Active Member

    Jun 15, 2011
    351
    35
    Do you have the AMS1117-3.3 (the fixed output version)? Or the AMS1117 adjustable version? Are you sure you have the right one?

    You said that you put a 1 Kohm resistor from Vout to GND... but there is no minimum load requirement for the fixed-output AMS1117-3.3 -- this minimum load is part of the quiescent current of the device for fixed-output versions.

    I'm not clear on when the regulator puts out 3.3 V and when it puts out 4 V. It sounded like the regulator puts out 3.3 V when no load is connected and 4 V when you connect the load... is that right?

    The situation sounds really odd. And you see the same effect with the LM3940?

    I would look at the output capacitor. If the ESR is too low, the LM3940 will become unstable. It needs to be more than 0.1 ohm. If you use ceramic caps you will not meet this -- you can add a 1 ohm resistor in series with the output capacitor if you want to use ceramic caps. Or use just an aluminum electrolytic cap. The AMS1117 seems to be less picky about ESR but it's not strictly specified.
     
Loading...