• State Resolved
  • Locked Locked
  • Replies 6 replies
  • Subscribers 33 subscribers
  • Views 356 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.

Original question:

UA9638C-EP: Tj question

UA9638C-EP: Power and Thermals

Nino Abundes
Prodigy 60 points
Part Number: UA9638C-EP
Other Parts Discussed in Thread: AM26C31

Tool/software:

Hello,

I am confused what the actual thermal capabilities of the UA9638C-EP are.

The device claims it is rated up to ambient temperatures of 85C. The electrical characteristics table specifies that the supply current with no load and no switching is a max of 65mA. At 5V, 65mA is 325mW. However the dissipation ratings table specifies that the power rating at 85C is 199mW. This would imply that it is not possible to run the device at Ta=85C... What in the datasheet is wrong here? Does it actually draw 65mA without a load? Seems like there are other line drivers that draw much less unloaded.

I read the answer below but it doesn't address this issue:

https://e2e.ti.com/support/interface-group/interface/f/interface-forum/925068/ua9638c-ep-tj-question?tisearch=e2e-sitesearch&keymatch=UA9638C-EP# 

Thanks,

Nino

  • +1
    TI__Guru 73591 points

    Hi Nino,

    This is kind of mis leading. The current value spec'd in the datasheet is assuming the device it driving a load of 100 ohms (this is part of the RS422 spec). The spec requires that the 100 ohm load must have at least 2V drop across it. If we do the math for current: V=IxR where Vdrop_100ohms=2V and R=100 then I is 20mA minimum. So some of the power will go across the resistor and the remaining power will go across the device. If Vcc=5V and the resistor gets a 2V drop then the voltage drop across the device is 3V. Using the power equation : P=VxI then we can say that the P_device's_driver = 60mW. Since the device has 2 drivers then the total power would be 120mW (P_device's_driverx2). 

    The walk through above uses 20mA minimum but since RS422 spec requires this as a minimum and silicon can have a lot of variations the VoD (voltage drop across the 100 ohm resistor) is going to be larger than 2V to ensure that is margin to meet the required spec. So using 20mA is actually the most conservative value we can use. Since the max supply current is 65mA we can assume the max current per channel is roughly half of this (a little less since the quiescent current for this device to be on is likely a 1mA or 2mA). 

    So if we assume 30mA is the current being driven through one driver and keep the voltage drop the same (in real life it would increase but to add some margin we will hold it constant), then we would have 3Vx30mA = 90mW per channel or 180mW across the device. This would mean that we can support ~200mW across the device at 85C because the 100 ohm resistor is supposed to handle the remaining power. 

    Ultimately, I think the test condition stating no load is a typo. 65mA without any load would be quite extreme for a device that takes TTL inputs to drive an unloaded output. 

    -Bobby

  • 0 in reply to BOBBY
    Prodigy 60 points

    Hello Bobby,

    I totally buy your answer, but is there any way to verify this? Does TI have any test data?

    Thanks,

    Nino

  • 0 in reply to Nino Abundes
    TI__Guru 73591 points

    Hi Nino,

    This device looks like it was made in 2010, back when we didn't have a shared server to put all of our characterization data into. (Essentially no one would know where that data is now).

    Only way I can think of to verify this is to simply put the device on a break out board and test it without a load. I would need to try to work with my admin to see if I can get a breakout board from digikey. This may end up taking a week or two. If you're willing to wait that long, I'll try to locate a board to run the test.

    EDIT: It looks like the commercial version of this device comes in the PDIP package so I can use that to evaluate more quickly then trying to go through our Admin to place a digikey order. I verified the two devices do have the same dies (the EP part has mounting/materials and goes through a stricter final test) so the results should be the same as testing the EP part.... Icc in the commercial part also states the 65mA spec. If you're okay with it, I can perform the test on the PDIP package which would arrive earlier. (I don't have access to getting samples for the EP part so I wouldn't be able to test this one in the SOIC package anyways)

    -Bobby

  • 0 in reply to BOBBY
    Prodigy 60 points

    Bobby,

    We were mainly interested in it because of it being an enhanced plastic part, but it seems like there is limited data on this part perhaps because it was a national part. I would find the data useful, but I am also considering just switching to an original TI part (even though it is older) in the AM26C31. It just has a lot more data in the datasheet, so maybe it is a better choice if support is ever needed?

    Thanks,

    Nino

  • +1 in reply to Nino Abundes
    TI__Guru 73591 points

    Hi Nino,

    Nino Abundes said:
    It just has a lot more data in the datasheet, so maybe it is a better choice if support is ever needed?

    I would agree with this. We recently changed the die for the AM26C31 device to our latest process node technology (using 300mm wafers, TI has been decommissioning older technology devices using 150/200mm wafers) so the longevity for this device is pretty safe for the long term. 

    The UA device is in an older process technology..... I haven't heard of any EOL plans for any of our EP parts though. 

    I can collect the Icc data for 3 parts (would be the commercial which would perform the same as the EP since they share the same die) at room temperature with and without driving a load if you are leaning towards the EP part more. 

    -Bobby

  • 0 in reply to BOBBY
    Prodigy 60 points

    It seems like EP may not buy us much if the AM26C31 is safe. Perhaps I will just design the AM26C31 in...

    Thanks,

    Nino