• State Resolved
  • Locked Locked
  • Replies 28 replies
  • Subscribers 29 subscribers
  • Views 6787 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.

DRV8313 supply current glitch

Alan MacIntyre
Prodigy 220 points
Other Parts Discussed in Thread: DRV8313EVM, DRV8313

I'm having an odd problem with the DRV8313. The design has two separate motor channels. The breadboard uses two DRV8313EVM boards modified to interface with an external microcontroller. It works fine. However, the first run of PCBs short the 24V motor supply briefly on power up. The PCBs use the DRV8313RHH (markings: DRV8313 TI 5AI CQ10 G4), rather than the DRV8313PWP. An LT4363-1 (current limit 5A) ramps the motor supply voltage in about 3ms. The PCBs glitch the supply (draw excessive current for about 0.5ms) as the voltage ramps between 1-3V. Sometimes the LT4363-1 current limit trips, sometimes not. Heating up the PCB tends to reduce the short current or interval. With the motor supply voltage at 24V, the DRV8313s on the PCB function normally. The states of the reset and sleep inputs don't affect the glitch. To eliminate other possible sources for the glitch, I wired up a DRV8313RHH with minimal connections. (inputs and OUT pins floating) The hand-wired DRV8313RHH glitched the supply like the PCB. Any ideas about what could be causing the glitch?

  • 0
    TI__Guru** 114905 points
    Hi Alan,

    I am not aware of any reason for the glitch.

    Are there any differences between the connections between the RHH and PWP boards? Is the motor connected as the power supply ramps? If so, can you disconnect the motor to eliminate that current path on power up.

    Can you provide a schematic of the PCB and hand-wired boards?
  • 0 in reply to Rick Duncan
    Prodigy 220 points

    Hi Rick,

     

    Thanks for the quick response. There are no differences between the RHH and PWP boards. I modified the EVM boards to match the schematic. The motor was not connected when I was testing the glitch problem. (However, the glitch is still present with the motor connected.) I will need permission from my client to post the schematic page. The hand-wired board is the same as the data sheet functional block diagram, section 7.2, with the following changes: no 10uF cap, no connection other than the 0.47uF cap on V3P3, INx, ENx, RESET, SLEEP, FAULT float, and PGNDx, COMPx connect to ground. This is the same as the PCB, except the 10uF is added, INx, ENx are held low by the microcontroller during the ramp, and RESET and SLEEP are either held high or low.

     

    Alan

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    To confirm, did you remove the 100uF bulk cap C1 on the EVMs?

    We may have to investigate further on our end. Can you provide a scope capture of the VM ramp so we can duplicate it?
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    Yes. I did remove the 100uF bulk cap on the EVMs. I have several scope captures that might be helpful. How do I send you a zip file?

    Alan
  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points

    Select the "Use rich formatting" to the right of the Reply button. Once selected, select the paper clip icon to attach the file.

  • 0 in reply to Rick Duncan
    Prodigy 220 points

    Here's the file.

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Thanks for the info. We will investigate and get back to you.
  • 0 in reply to Rick Duncan
    TI__Guru** 114905 points
    Hi Alan,

    There are a few items to try.

    Have you checked for shorts or low impedance paths between pin pairs like 19/20 and 24/25? The pin pitch of the RHH package is tighter which could create shorts.

    Do you have any routes under the device that could create shorts?

    As an experiment, is it possible to add a 5 to 10 Ohm resistor in series between the V3P3 pin and the 4.7uF cap? If you are able to do this, please power up again and look for a difference in current. It could be the current is due to the inrush of the cap.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    I measured 9Kohms between VM and GND on the (3) first run boards, 5Kohms on the breadboard, and 1Mohm on the hand-wired board. 65K between V3P3 and GND on the first run and hand-wired boards, 50K on the breadboard. Other than the thermal pad, there are no traces directly under the DRV8313s. I added a 10ohm resistor in series with the 0.47uF cap on the hand-wired board. The board still glitched like before.

    A few more possible clues. All of the PCBs and the hand-wired board with the DRV8313RHH have about the same current glitch. They differ a bit on the amount of current. All of the PCBs function as expected (drive motors) after the glitch occurs and VM reaches 24V. One board trips the 5A current limit every time. Heating the board gets it just under the trip limit. The other two boards trip sometimes. Using a 180ohm resistor in series with the 24V supply brings the breadboard VM voltage to 24V with no problem. On the PCBs and hand-wired board, the 180ohm brings VM to 1.45V and 125mA. The breadboard VM idle current is about 4mA. The VM idle current of the PCBs (after the glitch occurs) is also about 4mA, about 1mA on the hand-wired board.

    Alan
  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Sorry for the delay. Have you been able to get permission to send the schematic? That plus the layout may help.
  • 0 in reply to Rick Duncan
    Prodigy 220 points

    Hi Rick,

     

    I don't have permission from my client to send the product schematic or PCB layout. However, here is the schematic of the hand-wired DRV8313RHH that has the same current glitch as the PCBs.

     

    Alan

    DRV8313.pdf

     

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    I don't see anything in the schematic of concern. There is one difference which should have no effect, but would like to rule out.

    As an experiment, can you cut the COMPO output connection to GND and report the result?

    Thanks.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    Disconnecting COMPO from ground didn’t make a difference. The hand-wired board still glitched. Figure 13 of the spec sheet shows COMPO connected to ground when unused.

    Alan
  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    The designer would like you to try the following experiments if you don't mind.
    1) Please capture the current information on the DRV8313EVM like you did on your boards. When you do this, please confirm whether the bulk cap is installed or uninstalled.
    2) Connect each floating input directly to ground and look for the disappearance of the glitch. It is up to you how you want to do this. You can connect them all at once or one at a time. If you do see a difference, please identify the input that causes the glitch.
    3) Assuming you are still seeing the glitch, disconnect the V3P3 cap and try again. For this one, you can probably stop the ramp at 8V as long as the power supply ramp rate is the same as the 24V ramp.

    Thank you for your patience on this.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    I'm a little confused about the experiments you are requesting. The DRV8313EVM boards (with PWPs) don't glitch. Only the boards with RHHs. (PCBs and hand-wired board) I removed the bulk caps from the EVM boards. The PCBs have 10uF caps. The hand-wired board just has a 0.1uF cap. The microcontroller holds the INx, ENx pins low during the ramp. It holds RESET and SLEEP high. I have also tried it with RESET and SLEEP held low. No difference. I disconnected the V3P3 0.47uF cap on the hand-wired board. It still glitched. I'm using the breadboard LT4363-1 to generate the ramp when testing the hand-wired board. It tripped at 4V and 2.5A after 0.5ms of ramp. This is typical.

    Alan
  • 0 in reply to Alan MacIntyre
    Prodigy 220 points

    Hi Rick,

     

    Maybe this is what you were asking for. The attached scope capture (of the breadboard EVMs) shows what I would expect the ramp voltage and current to look like.

     

    Alan

    DRV8313EVM ramp.pdf

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Sorry for the confusion. Did you hold all inputs pins low on the hand-wired board? Like the COMPO experiment, we don't expect to see any difference but we are trying to isolate this. After that the only remaining items are the charge pump capacitors. The charge pump is disabled during power up, so it should not be the cause. Adding a series resistor with charge pump caps should isolate those paths also.

    The waveform you captured for the EVMs (PWP) is perfect. Sometimes the current in the working board can provide clues, especially if it is be close to tripping. In this case, it does not look that way.
  • 0 in reply to Rick Duncan
    Prodigy 220 points

    Hi Rick,

     

    Okay. I connected INx, ENx, RESET and SLEEP to GND on the hand-wired board. It still glitched like before. I then added 100ohm resistors in series with the VCP and CP caps. It still glitched, but was just short of 2.5A (see attached) so the LT4363-1 didn't trip. Could be a clue or a coincidence.

     

    Alan

    DRV8313RHH hw glitch.pdf

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Thanks for gathering the information. At this point, we are still puzzled and have ordered some RHH devices to try to duplicate your finding. These should be available early next week. I will update you once I get the devices in the lab.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    Some additional information. I ordered a test socket and wired it up like the hand-wired DRV8313. I also ordered a few more DRV8313RHHs. These are marked 5C1 C507 G4. (The first chips are marked 5A1 CQ10 G4) I confirmed the one remaining unsoldered 5A1 chip glitched (2.5A) in the socket like the other chips installed in PCBs and hand-wired board. I then checked a new 5C1 chip in the socket. It didn't glitch. I tried a second. It did glitch, but only reached 0.75A. And, it only glitched once while connected. There's about 0.14V of residual bias on the 24V bus with the ramp off that must have an effect. The third new 5C1 chip glitched once at 0.5A and the fourth once at 0.75A.

    Alan
  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points

    Hi Alan,

    Thanks for the additional information. I expect to receive samples this afternoon and will attempt to duplicate your finding here. Will update you once I have tested the devices.

  • 0 in reply to Rick Duncan
    TI__Guru** 114905 points
    Hi Alan,

    I was lucky enough to receive samples with the same code as your first material (5A1 CQ10 G4). Unfortunately I have tried 6 samples and found nothing more than a 5mA current spike during power up. The board I have uses a socket.

    I have a few more to try later today. Will update again later.
  • 0 in reply to Rick Duncan
    Prodigy 220 points

    Hi Rick,

     

    Well, that's a bit disappointing, but it could be good news. (I've done something wrong.) Have you tried powering the chip in series with a 180ohm resistor? I attached a picture of my test socket along with several scope captures. Maybe a difference in the wiring?

     

    Alan

    DRV8313RHH socket 160419.pdf

  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Thanks for sending the picture. That plus my schematic has provided a clue.

    Would you please float pin 18 on the socket and report your findings? I am working in parallel to confirm the clue on this end.

    Thanks.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    No glitch with pin 18 floating.

    Alan
  • 0 in reply to Alan MacIntyre
    TI__Guru** 114905 points
    Hi Alan,

    Pin 18 should not have been labeled GND in the datasheet. It should have been labeled "Reserved -- DoNotConnect" When powering up there is an internal FET that turns on briefly as you noted causing the glitch.

    The datasheet will be updated to reflect this change.

    I am sorry for the trouble we put you through.
  • 0 in reply to Rick Duncan
    Prodigy 220 points
    Hi Rick,

    Thanks for all your help resolving this problem.

    Alan