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.

Is 450 volt cap across isolation barrier on HVMotor kit and F28069ISo control card adequate?

Other Parts Discussed in Thread: MOTORWARE, TMDSHVMTRINSPIN, TMDSCNCD28027F

I note the cap C71 on the F28069ISo (rev 3) control card appears (from the BOM) to be a 0.1uF 450 Volt Ceramic.

A similar capacitor appears at C17 on the ISO-USB-to-JTAG-II Macro on the HV board.

This seems to me an unsuitable and arguable unsafe type and voltage rating of capacitor to be placed between a 240 volt grid zone and a USB low voltage interface.

Am I missing something - or being unduly conservative? Or is the design "unsafe".

 Richard.


  • Richard,

    C71 is a no-pop on the current Rev0.2 (69MISO) and Rev0.3 (69ISO) controlCARDs. You are correct, it would be unsafe.  We discovered that oversight ourselves....and only lost 1 motherboard in the process. :)

     

  • Thanks Chris. I am finding the rev levels a little confusing. In the current "Motorware" documentation there are two rev levels for the F2806xMISO control board called "R0_2" and "R0_3". The card I received a couple of days ago as part of my TMDSHVMTRINSPIN is labelled: "F2806y ISO Control card Release 0.2". According to the BOMs in Motorware documentation - the R0_2 C71 ls a 10 volt cap - which seems clearly wrong. On the R0_3 BOM it is listed as a 450 volt cap which I think is still inadequate. As you say C71 is not populated on my actual board - so my board does not seem to align exactly with either BOM in the motorware documentation.. I assume my board otherwise aligns with the R0_2 BOM - though I haven't checked this .

    I would point out though that my actual HV Motor Control board DOES have a C17 part populated and this is probably a 450 volt ceramic cap as listed in the BOM. I really don't think this is adequate or safe for 250 volt AC grid operation. What do you think?

    I presume the caps across the isolation barrier were originally designed in to limit common mode dv/dt across the digital isolators. Removing the cap therefore would appear to leave the design open to false triggering of the digital isolators. Has TI verified this is not an issue?

    Thanks Chris

    Richard.

  • Richard,

    let's separate the issue

    1. C17 on the USB-ISO connector of the TMDSHVMTR r1p1 kit

    I agree, on first inspection this looks like an issue if you are using the USB-JTAG on the HVKIT.  Let me send this to the owner. The good news is that you aren't using this connection, you are going to use the emulation on the controlCARD.

    2. C71 on TMDSCNCD28069MISO controlCARD.  Yes, the 69MISO version of this card is revison 0.2   The standard 69ISO version is at 0.3    I can't really explain why we didn't revise both at the same time...I believe it was because the 0.3 version brings out the USB HOST connector and the 69M silicon on the 69MISO doesn't support USB HOST so we didn't want to confuse anyone?  It's been awhile since that decision was made.

    Yes, C71 should not be populated on either. I agree the BOMs on each should be updated for C71 = NOPOP. I will send this to the owner as well.

     

  • Thanks Chris. I appreciate your detailed response. That does explain many things.

    There are still a couple of questions in my mind:

    1) If I want to use the F28062F control card - then I do need to use the ISO USB on the HV board. Right? Then the "safety of C17" issue comes into play. One motivation for using the F28062F would be the added cost of the "M" parts in cost sensitive applications.

    2) I also wonder about the strategy of "de-populating" C71 on the "M" control cards. Presumably it was designed in for a reason. What are we sacrificing by removing it.

    I am asking these questions so I can make a decision on whether to "de-populate C17 on the HV board for testing - and whether I can use that as a reference design for a low cost controller.

    Regards and thanks Chris.

    Richard.

  • 1) there is no 62F controlCARD. 62F (and all the other F and M devices) are a subset of 69M. For all InstaSPIN-FOC & -MOTION controlCARD development you have two choices TMDSCNCD28069MISO or TMDSCNCD28027F.  If you want to design your own board you can use any device you like.  62F is popular due to having the lowest cost in the 6x series.

    2) C71 was put on by a designer who made a mistake....the first version even had a 10V cap if I recall!  The solution was to simply no-pop the cap that crossed the boundary instead of re-designing the PCB to remove the footprint.  It is certainly safe.  We have been using them for 2.5+ years in all sorts of high voltage/current applications w/o issue.

    I can't adequately answer your question regarding removing it on the HVKit.  I haven't got a solid answer yet. If it was me, I would use TMDSCNCD28069MISO for testing, without a doubt.

    For your own hardware I would not put on isolated JTAG. Just bring out a JTAG header (which you can no-pop once in  production). This will allow you to use a low cost emulator that plugs into your board.  The ISO emulators are like $299, and if you are REALLY cheap I've even seen someone just cut a 69MISO board and wire out the JTAG signals. Presto, "free" isolated emulator.

     

  • OK. Thanks Chris. Yes. I was confusing 28027F and 28062F. Not sure how I got that wrong. Brain parity error I think...:-) But you correctly assumed I meant 28027F.

    And you suggestions make perfect sense - leaving isolation off the "cost sensitive" design.

    If you can get an answer on the wisdom of  removing C17 from the HV board - it would be helpful to me. It appears to be analogous to de-populating C71 on the 69MISO card.  I would quite like to run the eval with the 28027F - to see exactly how the "F" (FOC) behaves without the "M"  (MOTION) algorithms. This is probably possible with the "M" parts - but there may be subtle differences between the "FOC" in the M parts and the FOC in the F parts.

    You have been very helpful and I appreciate it...

    Richard.





  • "I would quite like to run the eval with the 28027F - to see exactly how the "F" (FOC) behaves without the "M"  (MOTION) algorithms. This is probably possible with the "M" parts - but there may be subtle differences between the "FOC" in the M parts and the FOC in the F parts."

    Richard,

    What you want to do is absolutely possible by using the 69M controlCARD.  There is no subtle difference. There are function calls to software that is part of FOC, and completely different function calls to software that is part of MOTION.  The MOTION function calls check a certain part ID of the device, and if enabled (if they are truly an M device) the functions are allowed to run.  So to be clear, the F and M libraries are both on the F and M devices, but only on the M devices are the M libraries allowed to be called.

    And if you look at the documenation and the way the software projects are structured you will notice that the FOC and MOTION projects are kept completely separate.  AND the main source code for the FOC projects is the exact same used to build a project  for an F2806xF and F2802xF device.

    The only software difference between the 6xF and 2xF is the revision of the ROM library. It is a very small bug fix that the 2x has (v1p7).  This fix is done for the 6x devices in the main project source code by the function

    #ifdef FAST_ROM_V1p6
    void softwareUpdate1p6(CTRL_Handle handle)

    You will note that it is fixing a scaling issue with how the Ls value is given to the estimator when loaded from user.h

    So using the 69M card to test InstaSPIN-FOC projects with the intent of using F28027F silicon is perfectly reasonable.  You will probably want to set the clock MHz to 60 in the user.h file to more accurately replicated the 27F. 

    On the other hand, we have used the isolated emulation built into the HVKIT for years without any damage, so I believe it must be safe and you and I are probably overlooking something. In that case using the HVKIT directly with the F28027F controlCARD is a valid path as well.

     

     

  • Richard,

    I apologize for getting to this late.

    The main reason a capacitor is placed across an isolation boundary is to provide some high-frequency noise suppression between the two grounds and reduce EMI (to allow the emulator to stay better connected under noise).  The negative of this approach is that some amount of current can now flow through the isolation boundary.  In many applications, no capacitance will be fine (and in some cases none will be a requirement) .

    As you are likely gathering from this thread we have had some mistakes with regards to this in the past.  Some of this were do to flaws in our manufacturing flow and other involve us comprehending the problem better.  I will say that any new 'isolated' cCARDs that are designed will have a 2.2nF-10nF 1500kV capacitor across the isolation boundary. 

    I will admit that I am not an expert on this.

    ----

    A 450V isolation capacitor will very likely perform adequately under the conditions that this board will work under.  As Chris mentions, we also have not heard of any issues.  However, for end-use industrial applications this should get bumped up.

    ----

    The R0.2 vs R0.3 controlCARD confusion comes because the original designer of the F2806x controlCARD did not revise the silkscreen to show R0.3 when they updated the PCB to R0.3.  The key way to know that you're getting R0.3 is that it should be labeled TEX065 on one of the edges of the physical board.

    ----

    Hopefully this clarifies things somewhat.


    Thank you,
    Brett

  • Many thanks Chris and Brett for the great support.

    It looks like my 69Miso card is at rev0.2. It has TEX052 on it.

    I consider this issue closed now... and I'll answer "yes" to the "Did this answer your question".

    Thanks again

    Richard.