Utilizing both TTL and RS-232 level signals from transceiver

Question

In the past I have used Maxim's MAX213CAI part as my RS-232 transceiver with FTDI VCOM chips. In our application we utilize both, the TTL level and the RS-232 level signals, depending on which device we attach to our board. This means I have both signals routed to the connector:

Where TXD1/RXD1 are the TTL level, and TX1RS232/RX1RS232 are the RS-232 level signal, for the same COM port. This works just fine with Maxim's MAX213CAI.

However, due to the high price of the MAX213CAI and the very attractive price for EXAR's SP213EHCA I decided to switch. I got new boards for testing some other changes and included this new Exar part, and lo and behold, it doesn't work.

I tried with a lookback and it just doesn't work. The EXAR part pulls the line high and the FTDI part (in the loopback), or the device attached to the connector, can't pull it low enough to receive data.

I took this screenshot from my scope:

As you can see, the RXD1 line doesn't drop bellow 3.5V, not enough to detect a "LOW". This is not the behaviour of the MAX213CAI part.

The 2 parts are 100% pin compatible, and this is how I have them hooked up:

PIN 24 is the ENABLE pin, if set to HIGH, the outputs are "Enabled", and if set to "LOW" the outputs are set to "Tri-State". The problem I described above is when the EN pin is set to HIGH. If I set the pin to "LOW", then the TTL level signal works just fine, but the RS-232 signal stops working.

Any ideas why the Maxim part has no issues with this, but the Exar part does?

EDIT:

These are the traces for the charge pumps:

PIN 15 (C2+):

PIN 16 (C2-):

PIN 144 (C1-):

PIN 12 (C1+):

Answer 1

That's what happens when you rely on undocumented behavior.
Unless I'm misinterpreting your question, you seem to expect to be able to override the xxx213 logic outputs while they are enabled ...
Why do you expect to be able to do this?
The behavior you're seeing indicates that the SP213's logic outputs are better able to source current to drive a high level output than the MAX213's are.
The datasheets for these parts don't seem to give a spec for how much current they can source on these pins so your 'off-label' usage is difficult to predict - but really the results you're seeing are unsurprising.
You've just been lucky with the MAX parts up to now, but if Maxim made an internal change to the IC for whatever reason you could end up stuck with the same problem even with these parts.