• State Resolved
  • Locked Locked
  • Replies 7 replies
  • Answers 1 answer
  • Subscribers 35 subscribers
  • Views 253 views
  • Users 0 members are here
Support feedback
Options
Options
Related

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.

THVD8000: THVD8000 on low-voltage DC and ESD protection

Théo PRUVOST
Prodigy 10 points

Part Number: THVD8000
Other Parts Discussed in Thread: BQ24074, TPS27081A, TPS22919

Hi all,

I have some concerns about my design around the THVD8000, as I can't totally related to existing designs.

Prototype1-Power Stage.pdf 

Prototype1-Power Stage_page-0001.jpg

The board is enclosed inside a box made of brass and is battery powered, with an external charger. To be able to charge the battery, the body of the box is put to GND while an isolated brass pin is powered with 5VDC (could be re-designed to 7VDC if losses are too high). Charging the battery and powering the system while so can draw up to 1.2A on the power line. In case it matters, the power pin is a T-shaped cylinder of 13mm and 7mm diameter with a total length of about 22mm. In this configuration, adding a separate line for data would be very costly and complicated, that's where the THVD8000 comes into play. It allows the internal MCU to communicate with the charger MCU, mostly to monitor the battery charge, but also for backup data exchange.

I've dimensionned the coils and capacitors for 1MHz carrier frequency, and 2 nodes (device + charger). The calculations gave 120uH for the coils (which, given the sizes of such components for low DCR and 1.2A, I tried to keep the value low) and 33nF for the capacitors, that I figured I could increase safely to 47nF. The 120 Ohm termination resistor is paralleled with a 100nF capacitor as per the schematic of the EVM, but I could be wrong doing so. I've placed jumpers to be able to try 5MHz and also 500kHz, although I know that later frequency would need to replace the coils, it's just for testing. Do you think that part is somewhat okay? Do I need extra filtering on Vin to have a smooth VDC suppy or are the coils enough?

I've placed the ESD diode directly on the powerline. I figured it would thus protect both the THVD8000 and the load. The main use cases are a human manipulating the body of the box and suddenly tying BODY_GND to a different potential, and/or a short by conductive material between the body and the power pin. Here, I should note that I'm not totally confident with the fact the the body of the box isn't the "true ground". I expect that while charging, the coils can drop the voltage by up to 200mV (maybe more if I can't source good coils). I don't know if it's relevant. Maybe I should design the charger so that BODY_GND is actually earthed? Any advice on that topic would be appreciated.

Concerning reverse current protection, I've looked at several possibilities. I'd rather not use a Schottky diode, because I'm already quite concerned about the losses in the coils at maximum charge. If the charger delivers a standard 5VDC, it must go over 4 coils and still provide 4.35V to the input of the BQ24074. If I change the design to supply 7VDC, then a Schottky could be a solution. But anyway, I've checked how I could use a TPS22919 or a TPS27081A to act as a reverse current protection instead, but I'm not sure how to properly wire such an IC. Maybe using a pull-down resistor on the EN line and a diode between VIN and EN? Looking at the internals of the BQ24074, I've left such solution aside, as I think the MOSFET inside is already doing this job between IN and OUT. Am I wrong to think that?

Note: I don't need reverse polarity protection on the battery. This part is supposed to be assembled in the factory and never removed.

Any comment would be greatly appreciated.

Some additionnal (not so relevant) details: the charger provides the "master" VDC on the power line. Once filtered out of the data, Vin reaches the BQ24074 for power path management and battery charging. Between the BQ24074 and the battery, a BQ27411 fuel gauge is placed to monitor battery status and report to MCU. The regulated output of the BQ24074 is fed to an LDO to ensure the VCC rail is at a fixed 3.3V. The BQ24074 provides a PGOOD signal that wakes up the MCU when plugged with the charger. The MCU then enables the THVD8000 to start communication. When on battery only, the THVD8000 is switched off to save power.

Thanks !

Théo

  • 0
    TI__Guru 54042 points

    Hi Théo,

    1. I wouldn't use the EVM as the "golden standard" for this device - not that the EVM is bad (there are some choices - like the 100nF is actually real bad to put there- I think it should be pF not nF - I don't agree with but overall it is okay) however - there is so much variability with how this is actually implemented in real life - there is no real golden standard. That being said - the EVM is actually designed for 2 nodes at 1MHz so it actually may work pretty well in your case. 

    2. 120uH is correct for 2 nodes at 1MHz - but that is effective 120uH - so you need to account for derating a tolerance - so real inductance is = 120uH + tolerance + derating. You can use that for 5MHz as well - as our equations give minimum inductance - and higher speeds give smaller inductances - the only real concern would be SRF of the inductor (if even listed) - but generally the spike to 1MHz to 5MHz isn't that big. 

    3. 31.8nF is what I am getting as the minimum capacitance - so anything bigger is safe. In DC power systems 1uF is pretty common because it works for all modulation frequencies - in AC systems is where you would really want to get as close to the 31.8nF as possible since AC signals will pass through the capacitive coupling. 

    4. Since your "DC power" is real low (typically we see 24V and 36V more commonly - 5V is fine though and it wouldn't be the first just less common) having the TVS diode on the shared bus is okay - typically we do have it at the local node that is not the shared bus (so between IC and Caps) - so that is probably not a big deal since the DC voltage is low - however best practice is at least one locally for these types of applications. 

     

    5. So C14 should be at most 100pF but more realistically I would not put it there at all - filter caps to ground are more effective (on A and B to ground 50pF to 100pF or cap between A and B <= 100pF)

    6. For the reverse polarity protection  - all your considerations seem very legitimate - I would suggest creating a new thread that just goes and talks about the Power parts specifically (The TPS/TLV devices ) - that way you have the best person answering your questions because honestly I am a THVD8000 expert not so much on those power parts though - and I just know you will get a better response if the main title includes the TPS or TLV part number. In similar applications I have seen the diode is the standard way that problem is solved. You may also want to consult another thread with the BQ device because that device is actually probably the most complicated in this application - nothing big stands out why you'd need to contact them - but they may have seen things similar to this and they could give some extra information. 

    Please let me know if my above response didn't answer all your THVD8000 questions - I would also suggest creating at least one more thread on the power parts to get the best support possible for those questions - you may also want to look into ideal diode IC's - they are pretty complicated but they essentially act as ideal diodes when setup (they are NFETs with a charge pump at their core) 

    Best,

    Parker Dodson

  • 0 in reply to Parker Dodson
    Prodigy 10 points

    Hi Parker,

    thanks a lot for the many useful inputs. You're right, I forgot to account for the tolerance and derating. I've selected a 150uH component instead, but the SRF would not allow usage at 5MHz. Anyway, I expect I will need to change the coils to test other frequencies, 1MHz is fine for now, the concern is about how the frequency can go through the brass material of the pin and body and the conductive components doing the junction (screw, wire between body and PCB, contact pins of the charger, ...). I guess I can only do real worl tests here to find out.

    Speaking of the behavior of the current inside the body, I was wondering if I could actually only use the "Vin" pin to convey the data, so the body would be connected to the "real" ground. As the body can be manipulated by a human, I'll be more at ease if it was earthed. Also, as the device is removed from the charger, do you think the coils need flyback diodes to let them discharge as the connection is cut (at BODY_GND and BODY_VIN)? Would these diodes interfere with the data signal? Maybe it's where C14 could be useful?

    I was about to move the TVS diodes behind the capacitors, but I'm afraid they would no longer protect the rest of the circuit in case of long external events. The device itself will be exposed to complicated conditions when in use (not so much during charge, that why it's made out of solid brass!). Do you think I should simply duplicate? One between Vin and GND and one behing the capacitors between A and B? I guess that's why it's better to have larger capacitors, so they can let larger ESD pass, but all in all, I'd rather not and catch the ESD before.

    Kind regards,

  • 0 in reply to Parker Dodson
    Prodigy 10 points

    Actually, I did not see that the question was already answered, and obviously there was something I did not grasp properly about the usage of the coil on GND line. Here is a revised version :

    In this design, I kept the 120R termination resistor. Is it possible? If not in pseudo-differential and I need to remove it, will I need to place a line driver then?

  • 0 in reply to Théo PRUVOST
    TI__Mastermind 22652 points

    Hi Théo,

    Parker is out of office on US holiday until Tuesday 2/17. You can expect a response next week. Thanks for your patience.

    -Ethan

  • +1 in reply to Théo PRUVOST
    TI__Guru 54042 points

    Hi Théo,

    Thank you for your patience while I was out of office. 

    Your revised Pseudo-Differential setup looks fine. The downside of doing it pseudo-differentially is that you lose a lot of the EMI/EMC benefits of differential bus and you will essentially half your VOD. So essentially your max length of bus could be compromised - that being said max length at 1MHz is generally hundreds of meters when everything is proper - so it won't necessarily cause issues in application - but it is something real world testing should be done on to get an idea of how your system with your selected cabling works etc.... 

    For the diodes I would put one inside the local node and one on shared bus - but like I said you probably can get away with just one- however I would keep at least a pad for a local one. 

    Finally - I don't think you need flyback caps in this application - I have never really seen them used with these devices before and it could impact signal integrity. Differential caps can help with transients - but generally filter caps (A or B to GND) are better at getting rid of noise that is both differential and common mode - where differential caps really only target differential noise. 

    Please let me know if you have any other questions!

    Best,

    Parker Dodson

  • 0 in reply to Parker Dodson
    Prodigy 10 points

    Hi Parker,

    thank you much for your time (and no pb for the vacations, I hope it was a good one :) ).

    I think I have now enough input, thanks to you, to produce a prototype. I've added a few solder jumpers to be able to re-route between pseudo-diff and real-diff. I have little concern about EMI, if the pseudo-differential setup works, this means that the entire body of the device is to real GND, enclosing the entire PCB and Vin pin in a faraday cage. The connection to the charger will be very short. Two contact pins and a pair of short wires to connect the pins to the charger's PCB. I think the distance between the THVD8000 of the device to the THVD8000 of the charger will be less than 20cm.

    I'll place a footprint for an additional TSV diode behind the caps, then, just in case. When I receive the prototype, I'll do some tests of connection/disconnection to check whether the coils discharge is a problem. Hopefully it isn't, so I don't risk perturbating the signal with extra caps/diodes.

    Kind regards,

    Théo

  • 0 in reply to Théo PRUVOST
    TI__Guru 54042 points

    Hi Théo,

    Sounds great! Please don't hesitate to reach back out if you have any additional questions and we will see what we can do!

    Best,

    Parker Dodson