• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 62 subscribers
  • Views 1019 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

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.

UCC28701 based high-voltage constant-current supply: chip appears dead

Lindsay Wilson
Prodigy 105 points
Other Parts Discussed in Thread: UCC28701, TIDA-00182

Hello,

I am attempting to design a high-voltage constant-current supply based on the UCC28701 controller. The final application will be a simmer supply for a laser flashlamp. Here are the supply specifications:

24VDC input

300V open-circuit output

300mA regulated output current

I have followed the design guidelines given in the various application notes, mostly in TIDA-00182 (www.ti.com/.../tida-00182). Here is the schematic I'm currently working with, and a photo of the setup.

imajeenyus.com/.../full_schematic.pdf

http://imajeenyus.com/temp/simmer/IMG_9766.JPG

The transformer is wound on a gapped ETD34 core and has about 1:13 winding ratio. Note that I am taking the voltage sense feedback directly from the secondary winding, since there is no requirement here for an isolated supply. It saves having a separate auxiliary winding.

Before you recoil in fright at the photo of my test setup, yes, I am aware that "dead bugging" switchmode supplies is far from ideal and causes lots of problems with stray inductance. However, that issue is not really relevant at the moment.

Essentially, the UCC28701 refuses to do anything at all, period. There is no gate drive output, not even a few pulses on startup. I would've thought that, even if the feedback components were not quite right (which, having double-checked the calculations, I'm pretty sure they are), it would still do something. The actual power side of the circuit works perfectly - if I drive the MOSFET "manually" from a signal generator + gate driver chip, I can get the specified voltage and current at the output. The feedback waveform looks excellent, with very little ringing.

I have left the NTC input open, which I think is correct, so it's pulled up and doesn't shut down the chip.

I am really at a loss what to do next, and would appreciate any advice. Is there something I'm completely missing about the operation of this chip?

Many thanks,

Lindsay Wilson

  • 0
    TI__Expert 7395 points
    Lindsay,
    Thanks for your interest in the UCC28701.

    Have you run through your design using our design calculator here?: www.ti.com/.../toolssoftware

    Use this first to verify you are not breaking any limits of the IC based on your design.

    Regards
    John
  • 0 in reply to John Stevens
    Prodigy 105 points
    Hi,

    Many thanks for the reply. I've checked through it and, as far as I can make it fit my application (it's mainly oriented towards an offline step-down supply), the values I'm using are correct.

    Now, something rather odd is happening. I decided to have another look at any waveforms that might be present during startup, so I connected scope probes to both the gate drive output, and the current sense input. Lo and behold, the silly thing decided to work (apparently) perfectly well. Nice 300V regulated output, although I haven't checked current limiting yet. It's actually sufficient just to have a scope probe on the current sense input. Note - once it IS running, the scope probe can be removed and it still runs.

    Thinking a small filter capacitor might help (since presumably that's all the scope probe is doing), I added a 100pF cap (C4) from the CS input to ground, and tried running without any probes connected, but unfortunately that didn't do the trick.

    I've some scope traces and also the PCB layout here:

    www.flickr.com/.../72157666741600765

    At startup, the output rises to the desired 300V within a few ms. The drain voltage and current waveforms look good, although the maximum drain voltage is getting a little high for the 100V MOSFET I'm using.

    So it now looks like it's some sort of layout issue. I know it's probably very hard to say anything general, but do you have any ideas what I could improve in this respect? For example, am I asking for trouble trying to do this with large through-hole components and should instead switch to SMD?

    Thanks again for the reply.

    Lindsay
  • 0 in reply to John Stevens
    Prodigy 105 points
    Hello,

    Thank you very much for the reply. I've had a check over the design calculator and, as far as I
    can tell (since it's more oriented towards an off-line supply), everything looks to be within
    spec.

    However, something odd is now happening. I decided to try and measure any startup waveforms, so
    connected scope probes to both the DRV output and CS input. Lo and behold, the chip decided to
    work. It produces the required 300V regulated output, although I haven't tested the current
    limiting yet. I eventually determined that the simple presence of a scope probe on the CS input
    was enough to get the chip started during power-on, but it would continue running without
    the probe connected thereafter.

    I tried adding a small 100pF capacitor (C4) between CS and ground, to filter any noise, but
    this unfortunately didn't help.

    I have put some of the scope traces, and the PCB layout, in an album here -

    www.flickr.com/.../72157666741600765

    It appears that the problem is something to do with circuit layout/noise suppression. At the
    moment, I have all of the controller circuit on the one PCB, and only the transformer and
    output rectifier/capacitor are connected with flying leads (which I've tried to keep as short
    as possible).

    Obviously this sort of layout is just for the prototyping stage, but do you think that maybe
    attempting to do it this way (with through-hole components, for example) is just not going to
    work, and I'd be better trying an all-SMD board with better supply decoupling etc? It's hard
    to know in cases like this, but I'd appreciate any tips. At least I now know the circuit does
    work, albeit a little irregularly!

    Best regards,

    Lindsay
  • 0 in reply to Lindsay Wilson
    TI__Expert 7395 points
    Lindsay,
    Yes, with the probe making it work on CS it sounds like you need to play with your filtering for startup. Be sure filter cap and current sense resistor are as close to CS pin as possible. You may also be articulacy hitting current limit because of this.
    In general, yes- through-hole PCB with flying leads sounds like a bad idea. Any high di/dt or Dv/dt mode or path on your proto needs to remain as short and fat of a connection as possible- xfmr and rectifier are examples of those so be cautious. Please review layout recommendations at the bottom of the Datasheet. Be sure supply bypass cap ISO as close to IC as possible.
    If I were you I would build an smd board even if just proto.
  • 0 in reply to John Stevens
    Prodigy 105 points

    Hi John,

    Many thanks for the reply - I've been having another look at the PCB layouts for the various application notes, and I'll start work on laying out a SMD version. I checked the through-hole prototype again (with scope probe attached!) and it does limit current successfully, so at least I know values are in the ballpark range.

    Best regards,

    Lindsay

  • 0 in reply to Lindsay Wilson
    TI__Expert 7395 points
    Lindsay,
    Thanks for the follow-up. Please "verify answer" and I'll consider this thread closed for now. Feel free to post a new thread if you run into any concerns with your SMD proto. Would be happy to help.
    Regards
    John