THVD1505: receive issue in mass production

User,

Can you describe the issue in more detail? What kind of issue is being seen? And when does it happen? Are these boards being placed under any kind of functionality or validation test prior to the issue occurring?

Regards,

Eric Hackett 

Hi Eric,

Thanks for your quickly reply.

There are two RS485 interfaces in our product, sent and receive the data (random character string ) between them with Baud rate 115200 in our test. For the failure, one RS485 reception fail when test in production. Then we use serial interface test again, we found that the products can receive well when sending by single character but can’t when sending character string, there will be missing string, even messy information.

For example:  When send the character string “a2 bc 53 f8 ce 91 1f c4 89 0c 5a 7d a3 88 8c 95 22 66 e3 16 cb 78 9a a3 a8 7e 83 ae 5b a0 51 11 f0 eb ba 45 98 f8 e3”, but the receive information is “& &*X*X**&*X**”, normal is receive information is same as send.

And the issue followed with the chip when do cross-validation (A-B-A test). 

Hi Eric,

Can you share me your e-mail address, I‘ll pass the detail information to you if can, thank you.

Hi,

In the setup, are both RS485 THVD1505? Is there any pull up/down and termination resistor on bus? 

Hi,

Yes, boths RS485 THVD1505, but the fail location is random. attached Circuit schematic for referrence. thanks.

Is it possible to decrease the pull up/down resistor from 4.7kOhm to 2kOhm. With current setup (bias on both sides), the bus voltage is about 5V*120Ohm/(4.7kOhm+120Ohm+4.7kOhm) = 63mV, which is less than the receiver threshold (100mV). You can refer to this blog for the detailed discussion.

(+) How much loading can an auto-polarity RS-485 bus support? - Analog - Technical articles - TI E2E support forums

Thanks for your reply, we use this design for long time, but no similar abnormal was found before, are there any other suspicious points about this issue? thank you.

Based on the calculation, the bus voltage is typically around 63mV. The device threshold can be considered as a gaussian distribution. If the majority of the parts has the threshold <60mV, you may see a few outlier after large samples. 

BTW, TI has a website to check the part's marking.

Packaging Part Marking Lookup | Texas Instruments (ti.com)

Thanks for your reply, I'll contact you or your team if any other reques.