Due to the U.S. Thanksgiving holiday, please expect delayed responses during the week of 11/22.

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.

UCC28720 Design not working!

Other Parts Discussed in Thread: UCC28720, UCC28720EVM-212, UCC28700, UCC28704, UCC24636, UCC28730, UCC24650

Hello, I've re-created the UCC28720-EVM on my own circuit card, emulating everything "almost" exactly - placement and layout is very very close.  I am even using the Wurth transformer called out in the design.  Only difference is the transistor Q1, which I chose ST Micro ST13003, because the transistor in the original design shows to be EOL.

I am targeting a 5W USB charger.


I get no output voltage, and indeed no VS either.  VDD is a nice triagngle wave between 13V and 20V, which I'm assuming is the device trying to charge and turn on through the HV port?  But, because there is no VS voltage the part isn't coming on?

Please let me know what else to try and measure - a I said I did everything like the EVM sample circuit.

Thanks!

  • Scott,

    Can you post your schematic and waveforms as noted below to help us assist you?

    Do you have one of the TI EVM's in your lab to compare the waveforms and behaviour to your board?

    What input voltage and output load settings are you using for initial debug?


    When you first apply input voltage, the VDD pin will charge from the HV pin. When VDD reaches ~21 V, the IC will try to start. It should deliver an initial 3 switching pulses at the DRV pin, and then uses the VS pin during these switching cycles to measure Vin & Vout. If the input voltage is too low, the IC will not start up, there will be no more DRV pulses. The VDD level will be allowed to slowly decay to the reset level, then it will re-charge back to 21 V, and the sequence will repeat until Vin is determined to be high enough.

    If you followed the EVM exactly, the start level (peak of AC, peak-detected onto the bulk cap) will be ~ Vstart = Ivsl(run) * R5 * Npa = 225 uA *82.5 k * 4.91 = 91 V pk, or ~64 V ac.

    If you try to start below this level, the IC will not start.

    - If you are above this Vin level, check that there is a set of 3 switching pulses at the DRV pin when VDD reaches 21 V.

    - If so, check that the main power switch is turning on, and the collector voltage is collapsing close to zero during the on-time.

    - If so, then check the transformer aux winding voltage (anode of D4), it should swing negative during the DRV pulses, and then very slightly positive during the flyback interval.

    - If there is switching at the anode of D4, then check the VS pin during the 3 DRV pulses - during DRV on-time, VS should be clamped slightly negative (~300 mV), and should then swing slightly positive during the flyback interval.


    If the circuit is behaving correctly as per the list above, it should start. If you do not see all the waveforms as listed above, then there must be an open circuit, or a short, or some other fault with the board.

    If you can check all the points above, and then post waveforms and schematic, we should be able to help you with further debug.


    Thanks,
    Bernard
  • Scott,

    Now that I checked the datasheet for the ST13003 BJT, I see a problem. The hfe of that device is very low, only 8-20, whereas the origianl EVM uses 2STL2580 with hfe of 60 min, 100 typical.

    Since the UCC28720 base drive current is limited to 19 mA typ (at min Ipeak level initially), if the hfe is too low, there will not be enough collector current to saturate the BJT, the collector voltage will be too high, and so the voltage across the primary winding will be much lower than the full bulk cap voltage - hence it will appear that the input voltage is too low. Besides this, operating in this mode will be very dissipative for the BJT.

    I would recommend changing to a BJT with a much higher hfe(min).

    On the ST website, the 2STL2580 device is still active, so not sure why you say that it is EOL.


    I hope this helps expaln your issue, if so please click the "verify answer" button.


    Thanks,
    Bernard
  • Bernard,

    Thank you.  The ST website shows the part as active, but Digi-Key and Mouser both say it's EOL and not abailable - no stock.  Could you perhaps suggest a good alternative that is available?

    I'll mark answered as soon as I can verify that the transistor is indeed the issue.

    Thanks,

    Scott

  • Scott

    I will check with the original designer of the EVM for a recommended alternative.

    Let me know if the issue is indeed due to the choice of NPN transistor, or due to some other fault on the board.

    Thanks,
    Bernard
  • Bernard,

    I found a transistor, Diodes Incorporated ZTX458, with hfe of 100.  Now, what happens is the fuseable resistor R9 fails open.  It happens relatively slowly - 20 or 30 seconds?  Enough for me to see that saw-tooth wave on VDD but not long enough for me to capture it.

    Though the ST website shows the 2STL2580 to be an active part, all of the distributors show it as no stock and EOL.

    Also, I do not have a TI EVM in my lab to compare waveforms.

    My test input voltage is 120VAC ( U.S. Mains), and my current output load is 10K ohms (the load resistor in the design).

    My schematic is attached.

    Scott

    watercurrent_5wa Charger Schematic.pdf

  • Bernard,

    I'm finally gotten things to settle enough to get a waveform.  The attached shows VDD on channel 1 (yellow) and DRV on channel 2 (blue).  It's not enough to get the device turned on. 

    Interestingly, the following picture is the same signals with a different - much much older scope that I don't particularly trust because it's not triggering correctly, but it shows some startup pulses.  The ONLY difference is the scope itself - I just moved the probed from the old to the new to get the above shot.

    Thanks for your help!

    Scott

  • Scott

    The first sope must have lower bandwidth and sample-rate, so it is not catpuring the short and infrequent DRV pulses. The second plot is clearly much higher bandwidth and smaple-rate, since there is also a lot more HF content on the VDD waveform.

    I would suggest that you trigger the scope on DRV with normal trigger and zoom in - is there just a sinlge DRV pulse or a set of there? What are the on-times/pulse-widths?

    I would then suggest that you check the voltages at the aux winding and VS pins per the suggestions above on an earlier post - there may be a short or open or winding polarity error.

    Thanks,

    Bernard

  • Bernard,

    Thanks for the help.  There are 3 pulses, most of the time, though there are occasions where either there are fewer pulses or they're spread out and outside my ability to capture them.

    The pulses are approx. 300nS wide:

    There is no measurable voltage on across the primary or auxiliary winding.  

    The transistor collector doesn't pull low, and there isn't any appreciable voltage on VSS:

    The transistor collector does not appear to pull low:

    The transformer does not measure open across any of their windings.

    Here is a picture of the layout - all traces are on the same side.  If it's different than the schematic, I cannot see where.

  • Scott,

    Ok, there’s a lot going on here.

    First of all, I would recommend maybe ordering the UCC28720EVM-212 to have as a reference? I don’t have one to hand in my lab, but I have ordered one so I can capture waveforms for you to compare with – it likely won’t arrive until next week though.

    Secondly, what is your spec? I assume you are running with low-line 120 V ac only? Or does your design need to operate over universal AC range 90-264 V?

    If you need to operate at high-line, you need to make sure the BJT is sufficiently rated for Vce max with Vbe = 0 V. This is the Vces rating, which is 800 V for the 2STL2580, and 700 V for ST13003. However, the ZTX458 does not spec Vces, just Vceo. ZTX458 also does not specify any switching characteristics (tr, ts, tf), so it looks like it may not be so suitable.

    Regarding the choice of BJT, I had a look at the other EVM’s, TI Designs and reference designs on the UCC28720 product folder. The 2STL2580 device is commonly used, but many also use variants of the ST13003 (MJE13003, PHE13003A, etc). So clearly, the ST13003 must be usable, although the hfe will limit the collector current and therefore the power level that can be achieved.

    I also know that unlike MOSFETs, BJT’s are notorious for using different pin-assignments for base, collector, emitter – you cannot assume that all devices in 3-pin packages have the same pinout. The 2STL2580 and ST13003 do share the same pinout, 1=E, 2=C, 3=B – I see that your layout numbers the pins the other way round, but it will work for both as long as the device is rotated 180 degrees. However, the ZTX458 is a different pinout, 1=E, 2=B, 3=C. Maybe this is why you initially had issues with the fuse blowing? You didn’t mention how you resolved the fuse-blowing issue, presumably this was the cause?

    While the ZTX458 has higher hfe of 100, this is only at low collector current levels. With Ic = 100 mA, hfe can be as low as 15 – so pretty similar to ST13003; and for the reasons noted above, it's probably not really suitable. However, I think it should be possible to get the power supply working at least at low power with the ST13003 BJT, since many designs already use the ST13003 or similar.


    Regarding your first set of scope plots (the low vs. high sample-rate plots), I missed some nuances of these plots initially – the plot showing VDD & DRV shows the DRV pulses being ~60 V in amplitude! This can’t really be the DRV/base level, maybe it’s actually the collector of the BJT?

    The other nuance is that the behaviour when VDD reaches the 21-V start level is different in some cases – for the first event shown, some switching activity occurs, then VDD decays slowly since some fault gets detected. But for the second and third events, something different happens, the width of the blue switching event traces looks thicker, and VDD drops by a small step before going into the low power fault mode. And then the fourth event looks the same as the first. Maybe #1/#4 are just the single pulse events that you observed when you zoomed in, and the #2/#3 are the triple-pulse events? Did you observe any startup events where more than 3 pulses are generated?


    Looking at the later zoomed-in plots, I assume yellow is VDD. Where is the blue trace measured? – I see that it’s now much lower amplitude. However, since the DRV pin is externally connected through a diode (BJT b-e junction) and 2-ohm resistor to GND, the DRV pin should be much lower amplitude, ~1 V, and ramping as the emitter current builds up. If the DRV is ~6-V amplitude as seen in these latest plots, it looks like this is maybe the compliance level of the internal current source that delivers the DRV turn-on current, indicating that the DRV is driving into a high impedance. Maybe a reverse-biased PN junction? This would be the case if the ST13003 is inserted backwards, C would be correct, but B/E would be swapped.

    The narrow 300-ns pulse-width is the minimum on-time due to leading-edge blanking (290 ns typ in the datasheet), this would indicate that the CS pin is high and terminating the pulse at the minimum possible time – again maybe due to the BJT pinout being incorrect?


    NEXT STEPS:
    ----------------
    I would recommend that you go back to the ST13003 (despite the low-ish hfe, it should work at low power).

    Verify that it is being inserted in the board correctly – use a handheld DVM to verify the base-emitter junction is fwd-biased with no input voltage applied to the circuit.

    Then probe the collector, VDD, CS & DRV pins during startup, to see that the BJT is actually switching on, and capture the first few pulses. DRV should only get to about 1-2 V amplitude, collector should fall to a low level, CS should be a ramp to ~190 mV.

    If the switching waveforms all look good, and the power supply still won’t start, next check the transformer aux winding and VS pin as noted previously.


    Let’s see if the above steps can get to root cause of the issues.

    Thanks,
    Bernard
  • Bernard Keogh said:
    Scott,

    Ok, there’s a lot going on here.

    First of all, I would recommend maybe ordering the UCC28720EVM-212 to have as a reference? I don’t have one to hand in my lab, but I have ordered one so I can capture waveforms for you to compare with – it likely won’t arrive until next week though.

    I don't have one - I'll see if I can get my hands on one but like you I won't be able to get it before Monday afternoon at the earliest.

    Bernard Keogh said:

    Secondly, what is your spec? I assume you are running with low-line 120 V ac only? Or does your design need to operate over universal AC range 90-264 V?

    120Vac for now, I'd like it to be universal though

    Bernard Keogh said:

    Looking at the later zoomed-in plots, I assume yellow is VDD. Where is the blue trace measured? – I see that it’s now much lower amplitude. However, since the DRV pin is externally connected through a diode (BJT b-e junction) and 2-ohm resistor to GND, the DRV pin should be much lower amplitude, ~1 V, and ramping as the emitter current builds up. If the DRV is ~6-V amplitude as seen in these latest plots, it looks like this is maybe the compliance level of the internal current source that delivers the DRV turn-on current, indicating that the DRV is driving into a high impedance. Maybe a reverse-biased PN junction? This would be the case if the ST13003 is inserted backwards, C would be correct, but B/E would be swapped.

    The narrow 300-ns pulse-width is the minimum on-time due to leading-edge blanking (290 ns typ in the datasheet), this would indicate that the CS pin is high and terminating the pulse at the minimum possible time – again maybe due to the BJT pinout being incorrect?

    Yes, Yellow is VDD, but between D4 and R4 - was convenient place to tack a TP.  Blue is DRV, measured at the base of Q1 (just left a let a little longer to clip onto.

    Bernard Keogh said:

    NEXT STEPS:
    ----------------
    I would recommend that you go back to the ST13003 (despite the low-ish hfe, it should work at low power).

    Verify that it is being inserted in the board correctly – use a handheld DVM to verify the base-emitter junction is fwd-biased with no input voltage applied to the circuit.

    Then probe the collector, VDD, CS & DRV pins during startup, to see that the BJT is actually switching on, and capture the first few pulses. DRV should only get to about 1-2 V amplitude, collector should fall to a low level, CS should be a ramp to ~190 mV.

    You were right - I did have the pinout backwards - which I did find with the ZTX458 - somewhat spectacularly :D

    So, have replaced with ST13003, installed correctly, and verified several times.  It still hasn't started up, but here are the waveforms of the measures above:

    VDD (yellow) + DRV:

    Q1.E (yellow) + DRV:

    CS pin + DRV:

    Q1 Collector + VDD:

    So, what I see are a couple things - first there's only one pulse of DRV not 3, and it looks to me like CS might be sourcing current?  It raises high (5v) at the first pulse, until right before it tries again?

    Could I have damaged the controller by Q1 being installed backwards?  I'll try changing it out and see if that makes a difference?

    Scott

  • Scott

    It's worth checking if the Rcs shunt resistor has been blown open by the previous events that took out the fuse. That would explain the waveforms you are seeing, very short min on-time since the open CS resistor will make it appear that a huge current is flowing.

    Thanks,

    Bernard

  • Bernard,

    That was indeed a problem!  The Rcs and the R8 were both bad.

    Now, I do get an output voltage, but only ~500mV.

    So I tested all the output components, and they are fine.

    Without any C, I get the switching waveform:

    Measured across the 10K-Ohm load resistor shown in the schematic.

    When I put the caps back in, I get 0.5V and no appreciable ripple even :/

    The caps are

    APSC100ELL681MJB5S

    Which was also a cross from the original, but I think they're pretty darn similar in almost all respects.

    This part should be exceedingly easy :/

    Scott

  • Scott,

    At this stage, is it possible for you to move on to a fresh untested board, and start from scratch with the BJT inserted correctly before first power up? This might just be quicker at this stage.

    If not, then when you said that the CS resistor was open, do you mean R12? If so, then this may have damaged the internals of the CS pin of control IC UCC28720, and it may be wise to put in a fresh one.


    At this stage, we would need to see plots of the switching waveforms over a few cycles, ideally Q1.coll, CS pin, T1.1, T1.9, maybe also Vout & Vdd. We need to see the peak current level, switching freq, etc. And then over a longer timescale to see if the controller is hitting a fault and running in hiccup mode.

    I don't think it makes sense to run without any output cap at all, the flyback won't operate correctly and it doesn't tell us anything meaningful. However, you can try running with a smaller output cap temporarily to see if things start up.

    In your most recent scope plot, what are the blue and yellow channels?

    When you say that Vout was only ~0.5 V, was this measured using a DC voltmeter? That may just be the average of the output voltage. What does it look like on the scope?


    Thanks,
    Bernard
  • Bernard,

    That was it - I rebuilt a fresh board and all seems to be working fine!

    Thanks so much for all your help!

    Now, I wonder if this design could be adapted for a 12W application?  I know there are other devices/reference designs that are 12W.  My particular problem is with the height - and in particular the magnetics!

    Scott

  • Scott,

    Just out of curiosity, why did you choose the UCC28720 device in particular? This is designed for driving a BJT, rather than one of the other devices in the family, designed to drive FETs. The BJT designs are usually driven by the need for very low cost, since the high-voltage BJT's are lower cost than MOSFET's. But the power level is limited.


    You can search the available TI designs and reference designs here:
    www.ti.com/.../refdesignsearchresults.tsp


    Click on the button “Search power designs by parameters”, then you can filter by input voltage, output voltage, AC input, isolated output, etc. The scripts can take a while to execute and update the page after each selection, it can look like it’s hung, you need to patient.

    Using this, I searched for AC input, isolated designs, 90-264 V input, 5 V output, and found a few designs that may be of interest:

    1. http://www.ti.com/tool/PMP8286
    5 V out @ 2.1 A. This uses UCC28700, another PSR controller, with MOSFET instead of BJT, and is maybe not necessarily low profile, but is very small and volumetrically efficient. This uses diode rectificiation.

    There are also many designs for both 2-A output and 3-A output:

    2. www.ti.com/.../PMP11600
    5 V @ 2 A using UCC28704 + UCC24636 SR driver

    3. www.ti.com/.../PMP11612
    5 V @ 2 A using UCC28704 + diode rectifier

    4. www.ti.com/.../PMP11295
    5 V @ 3 A USB-C using UCC28730 + UCC24650 wqke up (for better transient response) + UCC24636 SR driver

    There are many more 2-A and 3-A designs listed using a variety of different controllers in a variety of different board form factors – take a look to see which ones look most suitable. What is your height target or limit?

    Thanks,
    Bernard