MAX3232: Problem with version G4

Juan,

The -G4 suffix was introduced in order to denote "green" (Pb-free, etc.) package options; the same chip design is used for both variants.  So, I wouldn't expect a performance issue to affect only the -G4 version and not a non-G4 version.  Can you provide any more detailed info about the problems you are seeing so that we can help debug further?

Regards,
Max

Hi Max, my problem is this:

With MAX3232I G4 I got this signal:

The signal below is the signal that I transmit through a PIC32MX795F512L and the signal above is what I receive from a machine. 0xFF, 0xFB and 0xFF are replicas when receiving data from the machine, they are sent by mistake and are not sent by the PIC32.

With MAX3232I G3 I got this signal:

there are no replicas when receiving data from the machine. I think it's the MAX3232I because it's the only thing different in both designs, but I can be wrong.

Thanks for your time

Hi Juan,

Thanks for the extra info.  Just as a point of clarification, when you talk about the "G3" version you just mean a version that is not denoted as "G4," correct?  (I don't know of a version from TI with "G3" in the part number.)  If you can give the full orderable part number listed on the packaging it may be useful to us.

Where are you measuring each of these signals?  Are they the "DIN" lines at the respective transmit sides or the "ROUT" lines at the receive sides?  (Or, is this measured on the RS-232 interface itself?)  Would it be possible to look at the analog waveforms on an oscilloscope rather than a logic analyzer?  I'm wondering if something like crosstalk between the TX and RX channels could be causing the detection of erroneous "start" bits (which then result in reception of garbage bytes of mostly "ones").

Regards,
Max

Hi Max, 

chips with G3 that I have:

37P2MXAG3
MAX3232I

31P2L3AG3
MAX3232I

chips with G4 (and much more):

14VMAY3
MAX3232ID G4

18BA3F3M
MAX3232I G4

I'm measuring on the cable between machine and my board, then RS-232 interface. I don't have an oscilloscope available at this time. The data that I receive on the board and those that the machine receives are fine, the problem itself is that when the board receives the data from the machine it is replicating those 0xFF and 0xFB and it is presented when the chips are marked with G4

Thanks for providing the markings.  I see now that some of these devices would have the "G3" letters printed on the IC itself even if it isn't part of the orderable part number.  These devices also all have the TI logo printed on top as well, correct?  I just want to make sure since the MAX3232 part number is available from multiple vendors.

I want to make sure I understand your set-up.  Is the following description correct?

• Transmit path: PIC32 output --> MAX3232 DIN, MAX3232 DOUT --> Machine RS-232 input
• Receive path: Machine RS-232 output --> MAX3232 RIN, MAX3232 ROUT --> PIC32 input.

If so, are the signals probed on the logic-level interfaces between MAX3232 and PIC32?  Is it possible to check the signals at different points along the signal chain in order to find out where the extra characters are being introduced?

Regards,
Max

The pictures are (sorry for quality):

The description is correct.

I will try to try other points, but it is difficult because the board is made in 4 layers.


Thanks for attention.

Juan,

Thanks for the photos.  The markings indicate that these were most likely assembled in 2013 (top photo) and 2011 (bottom photo) at different sites, but both would use the same silicon from the same fab.

Unless there is something in your PIC software that would trigger sending those characters for some reason, my best guess is that there is some kind of noise coupling from the ROUT to the DIN lines of MAX3232.  If that is the case, you might try some things like adding some filtering capacitance or a pull-up/pull-down resistor on DIN to help keep its voltage stable.  Of course, if you can get access to an oscilloscope it would make it much easier to debug.

By the way, about how many units (out of how many tested) seem to show this behavior?

Regards,
Max

Hi Max,

Can you show me some filters? with pull-up and pull-down resistors I got the same behavior.

I have a lot of G4 chips on boards. These boards communicate with machines through a serial protocol, I'm using the same boards with a new protocol, the ones that work well are older boards (made 6 or 7 years ago) that use the G3 chip, the new ones use G4 and there are hundreds, I have not tried all but you see a pattern in them.

Juan,

I was thinking something along the lines of 1 nF to ground on the DIN and ROUT ports as a starting point.  That should be small enough to not degrade the signaling too much but large enough to effectively attenuate any higher-frequency coupling.

If the failing behavior is consistent and repeatable for a given system, would it be possible to swap a G4 unit from a newer failing board with the G3 unit from an older working board to see if the issue follows the chip or the PCB?

Regards,
Max

Hi Max,

With filters I have the same problem. Then, I measured the signal before MAX3232 and it seems good (chip G4 is still the suspect). I will change chip G3 for G4 in some boards, but I have to send them to a specialized center.

Hi Juan,

I wanted to check back in - have you been able to perform any follow-up testing on the re-worked PCBs?

Regards,
Max

Hi Max, 

I could not make the changes (exchange G4 by G3 on the boards that work correctly), I hope to do it as soon as possible

Hi Juan,

Do you have any updates on the debug efforts?  Or, have you been able to resolve the issue?  Please let us know if there is something we can do to help.

Regards,
Max

Hi Max,

I got no updates, sorry