MAX3232: Hot swap connection

Hi Laurent,

So the TTL pins of the device are mostly hot-swappable but there are cases where this isn't true. 

The DIN pins are not reliant on VCC to be present - abs max is -0.3V to 6V regardless of VCC - this implies that there is an ESD diode to GND but not one direct to VCC. 

The Rout pin however is refenced directly to VCC:

So if the ROUT pin is pulled up by something else when the power is off that could cause issues. 

If you have a schematic of just the RS-232 transceiver that would be helpful to double check (I understand if you can't share on the public forum - but if you can it would be helpful) - but the only pins that should be exposed to hot swap realistically are the DIN/RIN/DOUT/ROUT pins which are all okay regardless of VCC except ROUT. 

There could also be damage if for whatever reason the DOUT pin sees the 24V on the RS-232 bus side. 

Also what kind of damage is being seen and on what pins are affected?

Please let me know!

Best,

Parker Dodson

Hi Parker,

First, Thank you for your reply! I did not expect an answer that quick. :-)

Sorry for the delay...

I am not sure how to insert a picture on here. But I can try to describe the connections. The only pins being hot swapped are DOUT1 and RIN1. The TTL side pins are not disconnected. But the Power supply is. The power supply consist of a AP63357DV-7 powered from 24V set to output 12V followed by a AZ1117 to create the 3.3V supply used to power the MAX3232. The 24V side gets hot swapped.

I am not sure how to check which gate gets damaged on the MAX3232... How would you check this?

So, to recap: DOUT1, RIN1, GND, and 24V gets hot swapped the rest stays put.

Thank you,

Laurent

To insert an image, you must use the "Insert" / "Image" menu entry below the edit box.

As far as I can see, the DIN pins are overvoltage tolerant (even when powered off), but the ROUT pins are not. In any case, disconnecting GND does not sound like a good idea, because then any voltage can be lower than the floating GND, and that is forbidden by the absolute maximum ratings.

Hi Clemens,

When the GND is disconnected, so is everything else... This board is a module that is only connected to the rest of the equipment via +24V, GND, TX and RX on the RS232 side. These 4x signals are on the same 4 way connector. It could be that the GND is disconnected first by a fraction of seconds depending on how the user pull on the connector while disconnecting...

Hi Laurent,

Thank you for the clarification and you don't need to apologize for the delay. You also generally can expect we will get back to you within 1 business day - we aren't perfect at that but that is our general goal. 

So RIN and DOUT should be resistant to the hot swap in that case unless there is a transient on the DOUT pin that exceeds the +/-13.2V - it is very common to add a protection diode here - even in non-hot swap cases this is a common addition. 

As far as damage goes - just a basic continuity test - you should see ~5k to GND on the RIN pins (could swing between 3k and 7k - but typical is 5k) and it should be high impedance w.r.t. to VCC. DOUT pins should be high impedance when system is off. Essentially most common form of damage is for there to be a short to V+ or V- if damage occurs on DOUT pin (which I would imagine if there is damage it is on the DOUT pin - the RIN pins are generally more robust and I don't think I have seen a case of RIN damage - but I have definitely seen DOUT damage due to electrical transients that may/may not be induced by a hot swap application. ) 

Do you have protection diodes on the DOUT and RIN pins? 

Best,

Parker Dodson

Hi Clemens,

"The GND contact should be longer to prevent this."  That is actually a very good idea!  Thank you.

Hi Parker,
I have no protection diodes on TX or RX... Do you mean schottky diodes to GND ?
But if GND is disconnected, would they still help?

Thank you,
Laurent

Thank you Clemens for the clarification. ;-)

Hi Laurent,

Where you able to do a continuity test on the DOUT/RIN pins? 

While I still think the diodes should offer an extra layer of protection that could help mitigate this issue in the future; I still want to confirm that is where the damage is. My main concern here is that the output pins are fine but something could be damaged on the charge pump - it is rare for that to be the issue - it usually is some type of damage on the RS-232 bus pins, but it has happened before and I just want to confirm that isn't the case here - based on application where failure is most likely occurring it probably isn't the charge pump (at least only the charge pump - depending on transient a shock on the DOUT pins could potentially damage charge pump - but there have been cases where only the charge pump was damaged and that is what I want to confirm.)

Best,

Parker Dodson

Hi Parker,

I've checked those values on multiple boards with damaged chips, and they seem right: I get around 5k between RIN and GND, and high impedance with VCC; And DOUT to VCC and GND is high impedance too.

However, we've noticed that on damaged chips, the negative charge pump only gets to around -3V instead of -5V on V- (compared against a known good chip).

Laurent

Hi Laurent,

Thanks for the confirmation. 

On the damaged device is it possible to measure the VOH/VOL for a logic high input/logic low input? I'd image that you would be getting a -3V (around there) when you send a logic high; this would confirm that the V- charge pump is dysfunctional. If you could capture a scope image of the DOUT pin during the high/low output as well as a scope shot of the V- pin that would be helpful. There have been instances before where V- may be damaged when DOUT doesn't appear to have direct damage - usually the fix is still the diode - but seeing the scope shots would help rule out something completely unexpected essentially. 

Also if possible - measure the DOUT pin during a hot insertion event to measure the voltage w.r.t. device GND would be a good way to see if there is a potential large transient as well. 

Best,

Parker Dodson

DOUT doesn't show any kind of activity when the logic input is either high or low.

Thank you Parker!  This is a great help. I will implement all of these changes and comeback to you with some result ASAP.

Laurent

Absolutely no problem Laurent!

Please let me know if you need any more support or help and I will see what I can do!

Best,

Parker Dodson