This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

MAX3232E charge pump ripple

Other Parts Discussed in Thread: MAX3232E, TRS202E

Our MAX3232E charge pump design shows a fairly high ripple at V+ anv V-, about 740mV centering around ±5.5V in magnitude. This 740mV seems quite high for a power supply... what ripple voltage should we expect from this IC? This specification does not seem to appear in the datasheet.

The capacitor selection follows the 5V guidelines stated in the TI datasheet, as I understand it. We selected higher capacitor values, but maintained the presrcibed ratio of about 7:1. Our design uses C1 = 0.15µF and C2, C3, C4 = 1µF, X7R ceramic. The Vcc capacitor also has a value of 0.15µF as recommended under the "Power-Supply Decoupling" section. Adding a 3kΩ load to the V+ or V- measurement only increases the frequency of the ripple.

Please help.

  • Jim.

    Keeping the ratio is good thinking. However, this 7:1 C1 ratio can still give a large V+ voltage increase in just one cycle of boost. With your larger storage capacitors, a bigger ratio can be used. Try cutting the C1 value in half.

    The charge pump runs as needed to keep about 5.6V on both V+ and V-
    VCC charges C1. (to Vcc)
    VCC + C1 charges V+ (about VCC + 0.6V)
    V+ charges C2
    C2 charges V-
  • The 1µF capacitors rated 25V lose 30% of their capacitance when biased at 5V... I will replace C1 wth 0.1µF instead of 0.15µF. Still, I wonder what value of ripple we should see from this charge pump design?
  • Okay, I will perhaps try a C1 value around 0.1µF to 0.047µF. The datasheet guidelines seem a bit loose, based upon your recommendations. I would expect ripple to resemble the fractional ratio of capacitances times about 5.5V
  • Jim,

    You are very close. 7:1 caps could get a boost increase of VCC / ( 7 + 1 ). This is 688mV for VCC=5.5V
    It is normal for fairly large ripple using the recommend capacitors.
  • Ronald,
    Thank you for this information and for your patience. When I introduce the C1 value of 0.047µF with a 3kΩ load on V+, the ripple voltage magnitude stays about the same, about 700mV. Does the charge pump trigger have a significant hysteresis, perhaps 500mV? That might explain why this value seems relatively constant under load. I have scope images and ppt file of this information, if any way to post or share them.
    Jim
  • Jim,

    No hysteresis for regulation threshold.
    I do recommend placing load at DOUT pins instead V+ and V- directly.
    I remember there being an issue in the past, when someone wanted to use the V+ or V- to run a small external circuit.

    You may attach files after selecting "Use rich formatting"
  • A ppt file of the notes and waveforms attached. When I changed the C1 value to 0.047µF, I did not change the Vcc capacitors from its previous value of 0.15µF. Would this make a difference in the ripple? I would guess not, but do not know for sure. If 700mV represents the best we can expect, perhaps this conversation need not continue for long. Strange that the capacitor change did not affect the ripple of the loaded V+ and V-. Understood that we should check at the data terminals. 

    RS232_charge_pump.pdf

  • Ron,

    An experiment with both flying capacitors, C1 and C2, at 47nF yielded asymmetric results. This time, the V- output showed much smaller ripple as expected, under about 200mV, but not V+. Please see the attachment. I prefer to see cause-and-effect in designs, because this validation makes a correct approach more understandable and attainable. Why does the V+ behavior differ? I manually metered the capacitor values, and they appear reasonably close to the target values.

    The charge pump has an aspect of operation that I don't get. Do I need to make C1 even smaller than 47nF?

    Jim

    RS232_charge_pump_47nF_fly.pdf

  • Jim,

    I tried to find a 5V example of V+ and V- waveform, but I could only find 3.3V waveforms; see bottom of post.

    This list is the order that energy moves.

    The charge pump runs as needed to keep about 5.6V on both V+ and V-

    VCC charges C1. (to Vcc)

    VCC + C1 charges V+ (about VCC + 0.6V)

    V+ charges C2

    C2 charges V-

    Therefore C2 won't affect V+ in a significant way. It simply mirrors V+ to V- when V+ is at the minimum of the ripple.

    Running the charge pump as need does save some power, but the side effect is increased ripple voltage. The ripple voltage does not hurt the data transmission.

    Perhaps pin to pin compatible TRS202E is a better fit. It run charge pump continuously for less ripple voltage

    Here is no DOUT load V+ and V- waveforms.

  • Ron,

    After checking all of the capacitor values again... I managed to capture an oscilloscope trace of V+ under load that seems to make sense. See the lower left image within the attachment. The larger ripple apparently resulted from multiple cycles of the charge pump, which must have a very fast cycle. The LLH image shows that a given cycle resulted in a delta V+ of about 200mV as expected for each cycle. The timing of the supervisor that turns on the charge pump, along with the associated threshold, apparently combine to yield well-disguised results. The larger vertical delta V events look like they consist of 3 or more charge transfer cycles.

    This does not result in very good regulation, certainly not optimum, although it might prove good enough to service RS232 communications with shielded cable. Do you see any drawback to using both 47nF (C1 & C2) flying capacitors? This departs significantly from the datasheet recommendations, although this may seems possibly inconsequential.The charging ramps for both V polarities go nearly vertical.

    I will check out the pin compatible replacement that you suggest, perhaps it regulates with much smaller ripple.

    Regards,

    Jim

    MAX3232E_charge_pump_47nF_fly.pdf

  • Jim,

    You can observe the switch cycles better on the C1- pins (or other flying caps pins)

    Lower C1, or C2 caps are OK if the maximum load on DOUT can be provided.

    Data transmission on long cable (capacitance) and receiver load.