TRS3386EIPW does not work in lower temperature

Hello,

The TRS3386 is only rated for a minimum of 2.25 for VL, so you're correct. To what specifications do you need them to be similar? Package, voltage levels, temperature?

Regards,

Robert Regojo.

Hello Jerry,

As you have noted, our device is not compatible with 1.8V VL at this time.
I suggest the Maxim MAX3386E as a replacement.
http://www.maxim-ic.com/datasheet/index.mvp/id/2059/t/al

Regards,
Ron Michallick

I have run into a similar problem. I still have a copy of the old April 2009 revision of the TRS3386E

which clearly says:

• VL Pin for Compatibility With Mixed-Voltage

Systems Down to 1.8 V on Logic Side

How can it happen that such a fundamental design spec is changed without

assigning a new part number? We use it for an OMAP3 design and that one

simply has 1.8V logic levels. Spending a level shifter for a RS232 converter

with split supply looks a little strange. 

 The device is an EXACT EQUIVALENT in functionality and parametrics to the compared device. 

Hello Nikolaus,

The transistor sizing in the voltage translation circuit of the TRS3386E causes a propagation delay that becomes quite long when VL is 1.8V and low temperature is applied.
This is true for all date codes of TRS3386E and TI MAX3386E. TI may or may not decide to fix the transistor sizing in the future and reinstate the 1.8V specification again.

When VL is lower than 2.5V, I suggest the Maxim MAX3386E as a direct (drop in) replacement device for both TI MAX3386E and TI TRS3386E.

Regards,
Ron Michallick

Hello Ron,

ok, I understand. So the problem was already existing and not introduced by a chip redisign or similar.

Does it mean that if we have not seen any noticeable problems with 1.8V at typical temperatures (10-35 degrees), it is safe to still use the TRS3386E?

And, we use the 0-70 deg variant TRS3386ECPWR (not the TRS3386EIPW as mentioned in the title). So we don't expect it to work below 0 degrees.

Thank you very much,

Nikolaus Schaller

Nikolau,

Here is a chart I made during the evaluation with customer returned units (CRU) and a sample unit (CON).

Data was recorded at 25C and -40C, 0C can be interpolated. The 1.8V VL is on a steep prop delay slope, so the application's minimum VL must be considered.
The most important considerations are bit rate (actually bit time) and minimum VL. The driver signal will become unreliable just before prop delay equals bit time.

Let me know if you have questions about applying this chart to your application.

Regards,
Ron

Ron Michallick said:

Nikolau,

Here is a chart I made during the evaluation with customer returned units (CRU) and a sample unit (CON).

Data was recorded at 25C and -40C, 0C can be interpolated. The 1.8V VL is on a steep prop delay slope, so the application's minimum VL must be considered.
The most important considerations are bit rate (actually bit time) and minimum VL. The driver signal will become unreliable just before prop delay equals bit time.

Let me know if you have questions about applying this chart to your application.

Regards,
Ron

Ron,

thank you very much for this analysis.

So it looks as if we are at the corner of these curves.

We run the 3386E on the UART3 of the OMAP3530 which by default has 115200 bit/s. And, we usually operate this interface at 20-24 C. But the OMAP system can heat up the whole board. Finally, the TPS65950 provides the 1.8V VIO specified with +/-4%. This means 1.73 - 1.87 V.

So we will have up to 20uS @ 1.74 V / 0C. This would be beyond 1 bit time.

But the typical operation point would have 5 uS @ 1.8V / 25C which is half of a bit.

Another question: your curve is specified for VCC=5V. Does this have an influence? What about VCC between 3.2 and 4.2 V (directly driven from a LiIon battery)?

And one more: will there be an upgraded chip? The TI chip is a little cheaper than the proposed second source :)

Thank you very much,

Nikolaus

Nikoluas,

The voltages that matter are VL and V+, The boost supply is regulated, so VCC has little effect, but lower VCC generally reduces V+ a little (better prop delay). Heavier RS232 load generally increases V+.

I have the raw data pasted below. I also have an logiritmic interpolated 0C prediction. VL is in Volts and other other numbers are prop delay in uS.

I consider it risky to use the device at 1.8V espically at 115,200bps.

Regards,
Ron.