Hello,

Can you please tell us more about what specifically is failing? It may be helpful if you could capture the input signals and corresponding output signals on an oscilloscope to show the abnormal behavior.

Regards,

Max

Jeevaaraam,

Can you elaborate the test setup, like the input signal, supply control logic, output, etc? Like Max mentioned, it would be helpful to capture some waveform in the failing condition. These two devices do have similar electrical characteristics. 

Regards,

Hao

IC waveform characteristics (New vs Old)

To identify the characteristics of old IC and New IC, it is tested the Tx to Rx loop in the Table test setup as shown in Figure 1 at ambient temperature,

Where Processor module-1 (50-13-1D) is having IC part (AM26LS32AMDREP) and Processor module-2 (17-19-B2) is having IC part (AM26C32MDREP). The waveforms are captured at the TTL output of the ‘Rx Line Driver IC’ – U35/5 and measurement are listed in the Table 1.

Table 1 : Test results - Waveform characteristics

It is observed that there is change in the output (TTL) waveform characteristics of IC part (AM26LS32AMDREP) and IC part (AM26C32MDREP).

Description of failure:

My product should work for the temperature range from -40 deg C to +55 deg C.

The signal flow is

Board-1_Tx --> Mother board --> Board-2_Rx

As I have mentioned earlier, the IC “AM26C32MDREP” is working fine at room temperature (25 deg C) and high temperature (+55 deg C). During low temperature starting from approximately -30 deg C, one of the Rx channel is failing in following errors,

1.                       i.        No data error in receiver FIFO à 1024 bytes of data transmitted – Tested for multiple iterations.
2.                      ii.        CRC Error à 10 bytes of data transmitted.
3.                     iii.        Header Error à 10 bytes of data transmitted.

If I have replaced the IC part “AM26C32MDREP” by “AM26LS32AMDREP”, the same board is working fine for entire working temperature range (-40 deg C to +55 deg C à 10 cycles)

Thank you for the information. It would help me to understand the setup by mapping the ports from block level to IC level. Can you clarify if the signal from module 1 TX to module 2 RX is TTL or RS-485? I assume AM26LS32AMDRE or AM26C32AMDRE are used at RX side of each module, but I would like to get your confirmation. Thanks!

BTW, it might be helpful to compare the receiver's input waveform at different temperature. I'm not sure how much loss it could be on the trace.

Hao

<<I assume AM26LS32AMDRE or AM26C32AMDRE are used at RX side of each module>>

JS_06-Feb-2020: Yes, you are correct.

 The Tx and Rx paths are as follows,

1.1.1         TX Path

1.1.2         RX Path

Thanks for your confirmation. In this case, it looks more likely the issue is at the RS422 side rather than the TTL side of AM26C32AMDREP. What's the distance from TX to RX? What is the media, like cable? What's the data rate? Again, it would be helpful to compare the waveform in different conditions to find the cause. One hypothesis is that the AM26C31QD output amplitude drops at cold temp however AM26C32 input is not as sensitive as that of AM26LS32. There are several experiments you can try. 1) shortening the distance between TX and RX; 2) removing the termination at the input of RX side (AM26C32). 3) increasing the supply voltage of TX side from 5V to 5.5V. Please let me know how the results look like.

Regards,

Hao 

Thank you for your reply. Kindly find my answers for your queries,

What's the distance from TX to RX? 

JS: 4700 Mils (~12 cm)

What is the media, like cable? 

JS: PCB (with controlled impedance)

What's the data rate?

JS: 1MBPS (it is non-standard baud rate)

One hypothesis is that the AM26C31QD output amplitude drops at cold temp however AM26C32 input is not as sensitive as that of AM26LS32.

JS: Can you please let me know the rationale behind this hypothesis (Is it through previous experience?). The datasheet states the operating condition is -55 Deg C.

There are several experiments you can try.

JS: As it is the PCB and fixed product, these experiments may not help me.

1) shortening the distance between TX and RX;

JS: It is not possible to change the PCB traces

2) removing the termination at the input of RX side (AM26C32).

JS: The response is same.

3) increasing the supply voltage of TX side from 5V to 5.5V.

JS: It is not possible to increase the supply voltage (as it will affect all other ICs in the board)

The product is proven with ‘AM26LS32AMDREP’ and it is in production stage. As part#: AM26LS32AMDREP is obsolete without any replacement part (Ref: Product Withdrawal/Discontinuance Notification Details TI, PCN Number: 20150406000, Dated: 04/08/2015). As I mentioned in my first post, I have found part#: “AM26C32MDREP” as replacement part with almost same electrical characteristics and better SWITCHING characteristics.

Further experiment details:

To check the waveform characteristics at low temperature (From IC inside the environmental chamber to DSO outside the chamber), the have tapped the following pins using 26AWG-Multi strand wire

1. Input pins:
   1. Pin#6 (RS-422_D+)
   2. Pin#7 (RS-422_D-)
   3. Output pins:
      1. Pin#5 (TTL Out)

With this setup, the test is passed even at -40 Deg C also. To isolate further, I have removed the tapping wires from input pins #6 & #7 (As I suspected the issue is at RS-422 side) and tested at -40 Deg C. The test is even passed now.

When I removed the tapping wire from Pin#5 and tested at -40 Deg C. The test is failed now.

I suspect that there is load capacitance mismatch at the TTL output side for IC part# ‘AM26C32MDREP’ compared to Old IC part# ‘AM26LS32AMDREP’. The load capacitance details are not available in the both the datasheet. Is my understanding on the issue correct?

Thanks for your detailed information. It looks the marginal place is actually the receiver output (pin 5). If you compare the related spec Iol in the recommended operation conditions (6.3), AM26C32 does have smaller driving capability (6mA) than AM26LS32 (8mA). You can check the com port to see if you could somehow lessen the load. Another option would be to replace by SN65LBC173A.

Regards,

Hao

1)      The output of “AM26C32MDREP” is connected to “SNJ54LVTH162245WD”, whose input current is 20uA max and no other load id connected. Hence, 6mA is sufficient. Ten, loading issue should not be there.

2)      It is noted that if I’m connection tapping wire at TTL output Pin# 5 of “AM26C32MDREP”, the test is passed. It means that if we are increasing capacitive load (wire capacitance) to the output pins, we are getting correct data.

Kindly clarify the above two points.

Also, I will check the technical specification of suggested IC “SN65LBC173A-EP”.

Thanks for clarifying these. Another thought is that the issue was due to the fact that AM26C32 was too fast. Unfortunately there is no rise/fall time spec in AM26LS32 datasheet. Sometime the single-ended unterminated TTL signal might get some reflection and have some ringing if the transition edges are sharp. It would be helpful to have the waveform to debug this kind of signal integrity issue. In this sense, SN65LBC173A may not help since it has faster rise/fall time than AM26C32. But at least you can tell from the result that would better or worse.

Regards,

Hao