Fake/weird LM2679 ?

Discussion in 'General Electronics Chat' started by akis02, Nov 12, 2013.

  1. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    I bought a bunch of "LM2679" on ebay. I tried two out of the bag and here's what I have seen:

    1) The IC maintains 1.21V on the feedback pin (pin 6), and maintains the selected voltage at the output, but the moment you try to pull some current (say 100mA) the output voltage collapses.

    2) I noticed that the Current Limiting resistor on pin 5 is having an effect. The docs say this resistor sets the current limit to 37,125 / Radj, therefore for a 6A current we would need something like a 6K resistor. I have noticed that if I reduce this resistor I can get more output out of the IC, so there is something happening there.

    I do not think these ICs are defective, I mean, I cannot believe that they were rejects which someone inside NS took out and sold on the black market. Surely NS have procedures to ensure this does not happen ?

    Maybe these are re-labelled chips, so we are looking for another "LMxxxx" product, in a 7-pin TO-220 package, and with pin 6 being the 1.21V pin for sure.

    Can anyone help?
     
  2. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    Please post the whole circuit diagram, this is a step down switching regulator running at 260khz, so each component is quite important. Also please post a picture of your actual physical setup, it will have an effect too.
    Also, do you have access to an oscilloscope?
     
  3. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,232
    Could it be that your inductor is saturating? It needs to be big enough to support the current.
     
  4. #12

    Expert

    Nov 30, 2010
    16,248
    6,745
    Counterfeiters are not careful to keep their counterfeits in the same brand name as the chip they are re-labeling.

    But you're probably just having a misunderstanding. We work on those here.
    On the surface, you seem to make sense. Now we're looking for something peculiar to this circuit.
    Layout is critical and the board is going to look like a capacitor farm.
    Have you checked the TI Simple Switcher website? They do complete design specifications and sometimes you can even buy a circuit board. That simplifies things a LOT!
     
    Last edited: Nov 12, 2013
  5. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    I have built the circuit twice, using two ICs out of the bag of 20.

    The first circuit was made with point-to-point connections (all soldered). The second circuit was built using a general purpose prototype board.

    I understand that the components and placement are important, but here I am observing fundamental issues, not simply a drop in efficiency or some increased ripple at the output. I cannot even pull 50mA out of the circuit before the output drops.

    However the reference pin maintains the proper 1.21V when there is very little current drawn (1mA) but that soon drops when I try to draw more power.

    Even with inefficient components and layout, I would expect to be able to draw at least a few 100s mA with no problems.

    The circuit is very simple just as described in the docs, nothing added or removed. The coil is hefty enough, 22uH at 3-4A.
     
  6. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    Yes I have a scope. Not sure where to connect it and what I would expect to see. I checked the switch output (on the diode) and I did sort of see the 260KHz but there are multiple curves there, not a steady 260KHz.

    My coil is 22uH/4.6A surely man enough to give me at least a few 10s of mA.

    Do you think that the layout is so critical that you would not even be able to pull 20mA out? I can pull about 15mA before it drops.

    Edit: the seller has just informed me that these have come out of a sealed manufacturer's box containing 5000 pieces. He is adamant that they are genuine. What can I be doing wrong?
     
  7. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    The most important thing is the proximity of the input and output caps to the IC.
    Also, could you post the values of the rest of the components? I.e. all the input, output, bootstrapa and softstart caps, diode, feedback resistors.
     
  8. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    Input cap is 100mF/63V, ESR=123mOhms. Later added 470uF/63V, ESR=15-20mOhms (no difference).

    Output cap is 100uF/35V, ESR=150mOhms, later added 470uF/35V, ESR=30mOhms (no difference).

    Coil is 22uH/4.6A.

    Softstart is 10nF, boost is 10nF.

    Feedback potentiometer is 1K to ground (over which the 1.21V is being developed) and a 10K on the top leg, so we get 10 * 1.21 + 1 * 1.21 = 13.3V. I do get this voltage, and the voltage is stable when I vary the input voltage from 15V to 30V, so something is working right.

    Diode (my main suspect) is SB1100 and SB5100 (no difference between the two). Maybe the diodes are to blame? I have a box of other diodes, not sure what parameter is so crucial to this.
     
  9. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    100mF as in 100000uF? Etiher way, it seems you are missing a ceramic cap right at the supply pins of the IC. The bulk capacitance on the input is nice, but it is too slow to acutally matter until very high loads. You need very low ESR and low inductance cap to cover the fast transients, which boils down to the recommended 470nF ceramic you can see in the datasheet´s diagram.

    Not sure about the diode, as the datasheet doesn´t show the reverse recovery time, but a slow-ish diode should only drop efficiency, not disable the smps completely. Being a schottky I think it should not be the problem.
     
  10. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    I am attaching pictures of the switch on the scope. Here are screenouts of the switch pin under loads, input voltage is 21V, output voltage is 13.3V, loads are 30R, 100R and 470R and as you can see the straight line is what we get at the output (more or less)
     
  11. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    Sorry that was a typo. It was 100uF. And yes there is also a 100nF in parallel with it as well as a 100nF at the output.

    I also replaced the 22uH/4.6A choke with a RM12/N87 core coil at 27uH with and without an air gap, and it made no difference.

    I am coming to suspect the current limiting circuitry. It should be simply a resistor from the pin to the ground, and 6K should give us in excess of 6A, which is vastly more than the 15mA it manages before the output drops significantly.

    I tried lower resistors for the current limiting and they seemed to have made a difference, I even went as low as 10R, and it improved things but still the output dropped.
     
  12. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    As for the waveform, you should basically see a stable square wave at the output of the IC.
     
  13. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    hm that is weird, could you put some small resistor in series with the inductor and measure the voltage across it to get current? Also the voltage on the soft start pin might give some hints.
    Since the frequency is nowhere near the claimed 260khz, try changing the softstart cap for some different value to see if it has any effect. Also what is your supply voltage?
     
  14. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    Supply is 21V, output is 13.3V. Bearing in mind that the output voltage is stable at 13.3V with input voltage from 15V to 30V. So something is working right.

    I have changed the coil from this:

    http://uk.farnell.com/panasonic/elc16b220l/choke-22uh-4-6a-0r031/dp/1749170?ref=lookahead

    to RM12/N87/27uH (wound by hand).

    But it made no difference to the output really. The RM12 is huge compared to the previous coil, definitely not saturated.

    Here is a much better scope readings. The frequency is 260KHz.
     
  15. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    That looks rather weird, there should be no ringing before the rising edge, I´d expect some after an edge but definitely not like this.
    Try probing as many points as you can, with the ground lead as short as possible to minimize inductance and false ringing due to the probe, similarly to this setup http://www.impedanceanalyzer.com.au/media/1/20130531-passive_probe_with_pcb_.jpg

    Check that the supply at the chip´s pins is stable, maybe try using the other channel to find some correllation between the ringing and other parts of the circuit such as the softstart and bootstrap. Maybe try putting in a good old 1N4148 which is good for up to 100mA and is really fast, and see if it gets any better.
    And the inductor curent, you may find that it strarts crapping out when the output goes form discontinous mode to continuous - the inductor current doesn´t drop to zero before the next cycle.

    Actually the datasheet says 200mA continous for the 1n4148
     
    Last edited: Nov 12, 2013
  16. #12

    Expert

    Nov 30, 2010
    16,248
    6,745
    Page 7 of the datasheet: That prior ringing shape is characteristic of discontinuous mode, not some strange and mysterious inductor in a time warp.
     
  17. ronv

    AAC Fanatic!

    Nov 12, 2008
    3,281
    1,232
    Yep. Kind of everything looks ok. You see the full input voltage at the switch output, then you see the inductor turn the diode on. The only thing a little strange about those pictures is it looks like the switch is only on for a very short period of time yet inductor current was built up. My guess (and that's all it is) would be noise on either the FB line or current sense line.
     
  18. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    I have added caps on the feedback line, made no difference, I am suspecting the current limit circuit is being triggered somehow.
     
  19. kubeek

    AAC Fanatic!

    Sep 20, 2005
    4,669
    804
    So, could you measure the inductor current to rule out the current limit working as it should?
     
  20. akis02

    Thread Starter Member

    Apr 30, 2011
    57
    3
    Problem half fixed. It was a bad solder connection on the current limit resistor. The reason this eluded me for so long, and I wasted everyone's time as well, is because I had constructed the very same circuit using point-to-point soldering and I had the precise same problem, so I thought I was seeing the same problem!

    The square wave now looks very clean, as it should be, and the duty cycle changes according to the load.

    I say "half-fixed" because on a 13.3V/7R load the voltage sometimes, but not always, does drop to 10V, so here I suspect noise being the culprit. This is built on a prototype board without ground planes, and this may well be the new issue.
     
Loading...