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.

UCC2813-5 schemetic review

Other Parts Discussed in Thread: UCC2813-5, TLV431, UCC2813-1, UCC2813-4

Dear all

I will develop the product using UCC2813-5.
(Input : AC 50V or DC 50V +/- 30%
 OutPut : 5V / 12A (60W)
A circuit diagram based on the circuit proposed by TI Webench has been written and I have checking the schemetic.
As I am not a power expert, Power experts will review the circuit diagrams and review any fixes or modifications.
Thank you and BR

PJ

  • Hello PJ,

    Thank you for considering the UCC2813-5 and providing the schematic diagram. It will take me some time to review it and I may not be able to fully respond until early next week.

    One thing I can point out right now: a 50Vac input will have a peak voltage of 70.7V (+41.4% of nominal) whereas the DC input spec is only 50V +/-30%, so the nominal AC input already exceeds the maximum DC specification of 65V. And the AC input has no tolerance limits specified.

    Please adjust these input specs so that both have an allowable tolerance and so that they do not conflict with each other.

    Regards,
    Ulrich

  • Dear Ulrich

    Could you inform to me the result of circuit review data ?

    Unforturnally, the Power board is not working. so I have some question about UCC2813-5

    1. VCC : Test data is 3.8V( UVLO=3.6V)

    2. RC : 20Hz~21Hz

    3. VREF is PWM signal

    If possible, Could you inform to em the improve point in ASAP?

    Thanks and BR

    PJ

  • Hello PJ,

    I’m sorry that I lost track of this issue. I reviewed your circuit today and have the following comments:

    1. I don’t know the exactly specifications of your design, but I’ll assume 12A at 5V is the maximum current needed (not 12.5A or 13A for margin). I’ll assume that the bulk voltage is based on 50Vac +/-30% with a minimum valley voltage of 35V (which is -30% of 50Vdc).

    2. I assume that your posting for item 1 “VCC: = 3.8V” is an average DMM reading;
    that item 2 “RC: 20Hz ~21Hz” should be 20us~21us (about 50kHz);
    and that item 3 “VREF is PWM” means that VREF is toggling on and off.
    Please let me know if any of my assumptions are incorrect.

    3. The UCC2813-5 controller has VCC turn-on at 4.1V (typ) and turn-off at 3.6V (typ) which leaves only 0.5V delta-V in which to start-up. For total output capacitance of 5640uF and 12A maximum available current, Vout should increase to 5V in 2.35ms, IF there is no external load.
    If there is a load during start-up, then Vout rise time will be longer.
    Also, D10 and TC32 attached to the TLV431 form a soft-start circuit which automatically makes the start-up time much longer. So the fastest time Vout can get to 5V is 2.35ms without soft-start and without external load.

    The VDD bias cap TC13 is 15uF and must supply all of the bias current until (Vout+Vfs) *(14T/8T) –Vfa > 3.6V. If Vfs = 0.5V and Vfa = 0.8V, then Vout must be > 2.1V to keep VCC above UVLO. Vout can reach this in 0.987ms, no-load, no soft-start. VCC bias current = 0.5mA + (70nC*50kHz) = 4mA during start-up. To drop only 0.4V in 1ms at 4mA, TC13 must be >10uF so 15uF should work. However the TC32 soft-start circuit makes Vout rise much more slowly. The UCC2813-x devices already have ~4ms soft-start built-in (see Section 8.3.13 of the datasheet).

     I suggest to remove TC32 and increase TC13 to 68uF to see if the power supply will start up with no load. If it does, then you will need to decide if you want to keep the TC32 soft-start or not. Soft-start helps to avoid output overshoot. If you have too much overshoot, then you’ll need to add just enough soft-start to limit the overshoot (maybe it can be less than 33uF). And you need to increase VCC cap TC13 to allow for the longest start-up time, based on the soft-start cap value and any external load during start-up.

    4. The two 100K resistors from the bulk rail to VCC are not low enough to provide worst-case charge-up current for TC13. The UCC2813-x specification for VCC current before start is 100uA (typ) and 230uA (max). At 35Vdc input, the 200K will only allow (35-4)/200K = 155uA, so if some devices need more than 155uA before start, then VCC will never charge to 4.1V.

     I recommend reducing the (100K + 100K) to (47K + 47K) to guarantee there will always be enough start-up current for any device. Worst-case dissipation < 92V^2/94K = 90mW, or <45mW in each resistor, so 2012-size can still be used.

    I hope these suggestions and recommendations will help you get the converter running.

    Regards,
    Ulrich
  • Dear Ulrich

    Thanks for your quickly supports.

    I have test by your advice, but UCC2813 is not working.
    UCC2813 VCC voltage is under 4V yet.
    I will upload test data ppt file, Please check to file and inform to me the improvement in ASAP.

    Thanks and BR
    PJUCC2813D-5_test_data.pptx

  • Hi PJ,

    From your test data Q2 gate voltage (# 5), I can see that the Q2 MOSFET (FDB2614) cannot turn on at <2V. I'm sorry that I did not recognize this problem before. But the UCC2813-5 tries to start at 4.1V and OUT level will only be about 2V which cannot drive a normal MOSFET. This MOSFET needs gate voltage >3V, typically 4~5V to turn on properly.
    Screenshot #4 shows no switching on the drain, because the MOSFET cannot turn on.

    There are 2 basic options to decide here:
    1. Change the controller to a higher voltage device such as UCC2813-1 or UCC2813-4 to drive a regular MOSFET, but the transformer turns ratio will have to be changed, too, or
    2. Change the MOSFET to a logic-level MOSFET (such as FQB34N20L) which can turn on at 1~2Vgs. This option does not need to change the controller and transformer. Note: Rds(on) of this Fet is 75mR, compared to 22mR of the FDB2614. Be watchful of the temperature rise.

    I recommend to choose option 2 for the short term, to get the design running and debugged in most of its operation. Then later you can decide to keep it that way, or change controller and transformer turns-ratio to use a higher gate-voltage MOSFET with lower Rds(on).

    On another issue: Why did you change the RC resistor from 10K to 33K? The transformer inductance is designed based on the switching frequency and reducing Fsw by a factor of ~3 might lead to transformer saturation. I recommend to change R1 back to 10K. This value makes the oscillator run at ~100kHz, but 50% duty-cycle limit flip-flop in the UCC2813-5 controller makes the final switching frequency = ~50kHz. (33K at R1 would make Fsw = about 16.7kHz.)

    Screenshot #3) RC(pin4) does not look right. I think there is aliasing due to low sample rate of the oscilloscope. To see high frequency switching at slow oscilloscope sweep speeds, you need to increase the sample rate. The same goes for shots #5 and #6.
    For example, in screen shots 2-5, the time sweep is 100ms/div and your sample rate is 100kS/s. This means there are 10k samples per division, which = 10kS/100ms = 1S/10us. Shot #3 looks like R1 = 10K, so that Fosc = 100kHz, or 10us period. at 1 sample every 10us, the RC waveform would be sampled at nearly the same point every cycle, so it would look like the very low 6.9Hz signal you show in the test data. A higher sample rate ( and a faster sweep speed) should show the true switching frequency.

    Regards,
    Ulrich
  • Dear Ulrich

    Thanks for your cooperation.


    I have reply for your comment.

     And I have some question.
    Q1 : UCC2813 VCC level is 3.8V, Is that correct? ( If Rstartup vaule is changed from 10K to 47K, VCC level is not changed)
    Q2 : If Out level is 4V, MOSFET is good working. But Out level is always about 2V. Why reason ?

    Our schudles is very urgent, So Inform to me the improvement in ASAP.

    Refer to attahed file for test data and
     

    Thanks and Br

    PJ

    =============================================================================================UCC2813D-5_test_data_hpj.pptx
     
    From your test data Q2 gate voltage (# 5), I can see that the Q2 MOSFET (FDB2614) cannot turn on at <2V. I'm sorry that I did not recognize this problem before. But the UCC2813-5 tries to start at 4.1V and OUT level will only be about 2V which cannot drive a normal MOSFET. This MOSFET needs gate voltage >3V, typically 4~5V to turn on properly.
    Screenshot #4 shows no switching on the drain, because the MOSFET cannot turn on.
    ==> That`s right. MOSFET is dead, so I have changed new part,
    But UCC2813 VCC is not charge up over 5V, so Out port level is low(4~5V)

    There are 2 basic options to decide here:
    1. Change the controller to a higher voltage device such as UCC2813-1 or UCC2813-4 to drive a regular MOSFET, but the transformer turns ratio will have to be changed, too, or
    2. Change the MOSFET to a logic-level MOSFET (such as FQB34N20L) which can turn on at 1~2Vgs. This option does not need to change the controller and transformer. Note: Rds(on) of this Fet is 75mR, compared to 22mR of the FDB2614. Be watchful of the temperature rise.

    I recommend to choose option 2 for the short term, to get the design running and debugged in most of its operation. Then later you can decide to keep it that way, or change controller and transformer turns-ratio to use a higher gate-voltage MOSFET with lower Rds(on).

    On another issue: Why did you change the RC resistor from 10K to 33K? The transformer inductance is designed based on the switching frequency and reducing Fsw by a factor of ~3 might lead to transformer saturation. I recommend to change R1 back to 10K. This value makes the oscillator run at ~100kHz, but 50% duty-cycle limit flip-flop in the UCC2813-5 controller makes the final switching frequency = ~50kHz. (33K at R1 would make Fsw = about 16.7kHz.)
    ==> I have changed R value(33K -> 10K)

    Screenshot #3) RC(pin4) does not look right. I think there is aliasing due to low sample rate of the oscilloscope. To see high frequency switching at slow oscilloscope sweep speeds, you need to increase the sample rate. The same goes for shots #5 and #6.
    For example, in screen shots 2-5, the time sweep is 100ms/div and your sample rate is 100kS/s. This means there are 10k samples per division, which = 10kS/100ms = 1S/10us. Shot #3 looks like R1 = 10K, so that Fosc = 100kHz, or 10us period. at 1 sample every 10us, the RC waveform would be sampled at nearly the same point every cycle, so it would look like the very low 6.9Hz signal you show in the test data. A higher sample rate ( and a faster sweep speed) should show the true switching frequency.

    Regards,
    Ulrich