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.

TLC59283: Constant current output stability

Part Number: TLC59283

Problem with TLC59283, 16 Channel Constant Current LED Driver with Pre-Charge FET.

Schematic diagram: similar to Typical application Circuit in the datasheet. 3x TLC59283 for 6 LED matrices 5x7, multiplexed LEDS.
So 5 columns that go to all te matrices.

V-LED is 3.5V.
Voltage constant-current outputs: 2V if not active (no current through LED), 1.5V when active (current through the LED).
R-IREF 5k6, so output current 9 - 10 mA.


When one TLC59283 has many constant-current outputs active, for example 2x7 (2 columns, 7 rows), there is no problem.
When 2 or more TLC59283’s have many constant-current outputs active, and they do that in the same columns, then constant-current outputs that column is active seems to oscillate.

For example IIIIII or HHHHHH or LLLLLL on displays: each TLC59283 has 2x7 outputs active (2 columns, 7 rows).

Oscillation frequency approx. 19 to 25 MHz, seems to vary with V-LED, or probably the Voltage on the constant-current outputs, because that depends on V-LED.



Question: is this a known phenomenon, is there something we can/should do to prevent this ?




  • Hi Pieter,

        Can you share any waveforms about it?

  • 2744.schematic.pdfSchematic diagram

    One column has all LEDS active, 14 LEDS per TLC59283, x3 = 42 LEDS in total

    Channel 1: (multplexed) anode voltage at column that has all LEDS active.

    Channel 2: (multiplexed) anode voltage at other colomn. Notice oscillation interfering.

    Channel 1: (multplexed) anode voltage at column that has all LEDS active.

    Channel 2: U-DISPL (V-LED). Notice oscillation interfering (less than on other column)

    Channel 1: (multplexed) anode voltage at column that has all LEDS active.

    Channel 2: voltage on output of TLC59283, 2 columns LEDS off, 2 columns few LEDS active, 1 column all LEDS active

    (multplexed) anode voltage at column that has all LEDS active: other time base, to observe HF approx 20 MHz.

    REMARK: oscillation of current outputs (?) does not occur when only one LTC59283 has all 14 LEDS active.

    Regards,

    Pieter

  • Hi Pieter,

        Does the chip is far from your LED matrix in your application as shown in the picture? The parasitic inductance of long wire will affect the stability of the constant current loop. Try to reduce the length of your connecting wire.

  • Hi Charles,

    I assume you mean the coloured cable.

    Thought the same thing, tried wih a cable of approx 6 cm, but that did not stop oscillation. No substantial change of oscillation frequency or amplitude. For me this indicates that inductance of the cable does not play a role.

    I have been experimenting with capacitors on the FET sources: if I put a 1nF capacitor from FET source to GND, this does not help.

    If I put a 10nF capacitor from FET source to ground, the oscillation seems to disappear...

    But  somehow this does not feel good...  It gives some information, but it may have effect on 'ghosting' and perhaps EMC/EMI radiation. Have not tried that yet.

    You may have noticed that we use a schottky diode and 10 Ohm resistor instead of a zener diode and resistor.  In the forum I read that the voltage of the zener should be less then the LED voltage, which indicates that a current flows through the zener when that column is activated. Could it be that this may heve this effect on the constant current outputs ? If so, what resistor should be preferred , or how much current should flow through the zener ?

    I tried other manufacturer 100uF decoupling capacitors C1 and C2, lower ESR, that seems to improve, reduces the interference at U-DISP.  But the interference measured at LEDS and constant current outputs is not reduced by these lower esr capactitors:  the fact that I measure more interferende on the LEDS and constant current outputs than on U-DISPL seems to indicate that the interference does not come from the cable, but originates ar the contant current drivers.

    I noticed that the levels at the constant current outputs drop rather low during oscillation....

    Regards,

    Pieter

  • Hi Pieter,

        The wire I mentioned is the total length from VLED to the corresponding channel. What's the total length of your wire? The length of your wire is the major reason for oscillation.

        For Zener, the clamp voltage of zener: 0.6~0.7 VLED (rules of thumb). The value of resistor depends on the effect of upside ghosting and the discharge time.

  • Hi Charles,

    Thank you for your information.

    Length of VLED, that is from C1 / C2 to the FETS is up to approx 6 cm (4.5 inches).

    There are 5 pcb-traces (one for each colums) going from left to right over the pcb to all 6 matrices, that lengt is approx 23 cmm (9 inches).
    Each of the FETS are connected to one of these pcb-traces, somewhat out of the middle of the pcb, additional length up to approx 5 cm (2 inches).

    We can not change the size of the LED display pcb, so the pcb-traces can not be shorter. Unless there is another way, it lookes we have to use more FETS: for example 5 separate FETS for eacht display driver, resulting to shorter pcb-traces to the LED matrix columns.

    Is there any information available regarding effect of this parasitic inductance on the LED driver ? (I Can not find info in datasheets or other TI technical information.)

    Regards,

    Pieter

  • Hi Pieter,

        It's a pity that we don't have any existing materials concerned about this. Typically, 10 cm will start to lead to oscillation in other device. You can add a resistor in series with your connecting wire and check whether it helps or not.