0
\$\begingroup\$

What part of a level shifter datasheet indicates drive strength?

I have an MCU and sensor connected over UART (115200 baud). The MCU is 3.3V and the sensor is 1.8 V, so I'm using a level shifter in between. There is a 3 m cable between the two.

It appears the drive strength of my level shifter is insufficient, causing an SI problem. The signal from the sensor to the shifter looks nice, but the output from the level shifter suffers from low voltage and very rounded highs. This was also suggested by the sensor manufacturer.

Digital interpretation above, actual analog signal below.

I am currently using the RS0102 shifter. I also looked at other datasheets now that I am swapping the component, and they confuse me a bit... Listing I_OH=-20 uA seems common. I interpret this as <20 uA flows into the pin when it is in its high state. I'm thinking current should flow OUT of the pin when it's high, at least on one side of the shifter. 10 kΩ internal pull-up also seems common. But at ~3 V with 10 kΩ I'd expect ~300 uA, so what does the -20 uA from the datasheet mean? Speed does not seem to be the issue either, as 115k baud is quite slow compared to the capabilities listed in the datasheets. It feels like I am missing something. Perhaps this is the wrong value to look at altogether? What are the important parameters to find a suitable shifter? What is likely my issue?

My current shifter has a note in the datasheet re. that round trip should be <30 ns. That means ~<4.5 m at 115200 baud, right? Perhaps the cable is still the issue. What should I look for in a level shifter in order to drive the signal over 3 m cable?

\$\endgroup\$
3
  • 2
    \$\begingroup\$ Maybe the problem isn't the level shifter, but the circuit it is in. Please add schematics of everything between the MCU, level shifter, and sensor. Which sensor it is? How is the MCU pin configured? How much wiring or cable there is between the devices? Was this particular level shifter suggested by sensor manufacturer? Where can we find these manufacturer recommendations for level shifters? \$\endgroup\$
    – Justme
    Commented Mar 13 at 9:15
  • \$\begingroup\$ Perhaps you are right. The circuit is minimal, ESD and internal pull ups, but there is also 3m cable which might be the issue... I updated the question wrt the cable, what chip do I need for that use case? \$\endgroup\$
    – Alexander Ohm
    Commented Mar 13 at 10:35
  • \$\begingroup\$ And no, the manufacturer was not consulted before the shifter was designed. Nor do they provide recommendations. \$\endgroup\$
    – Alexander Ohm
    Commented Mar 13 at 10:50

2 Answers 2

Reset to default
3
\$\begingroup\$

Sometimes, the recommended operating conditions table show values for IOH and IOL. But in general, these values are specified as test conditions for VOH and VOL in the electrical characteristics table:

RS0102 VOH/VOL

The guaranteed drive strength for high-level signals indeed is 20 µA. (A negative value means that the current flows out of the pin.)

This device is a passive switch, but has edge-reate accelerators that increase the drive strength for a short time during rising edges. This works only if the capacitive load is not too high; see the section "Output Load Consideration" in the datasheet. I suspect that your cables are too long.

For unidirectional signals, a unidirectional, buffered translator like the TXU0202 or (SN)74AVC2T245 would be a better option.

\$\endgroup\$
9
  • \$\begingroup\$ Thank you. I have a 3m cable so indeed that could be my issue... Thanks for pointing it out! The note mentions that round trip should be <30ns. That means ~<4.5 m at 115200 baud, right? I suspect it could still be the issue. Will add this to the question. This is bidirectional, so your suggestions won't work unfortunately. But "buffered" is the keyword to look for? \$\endgroup\$
    – Alexander Ohm
    Commented Mar 13 at 10:31
  • \$\begingroup\$ UART signals are not bidirectional; you have two signals, each of which is unidirectional. \$\endgroup\$
    – CL.
    Commented Mar 13 at 12:30
  • \$\begingroup\$ Yes, that is true. I think you understand how I read it differently. With your suggestion I would need one TXU0202 for the RX line and one for the TX line, despite the chip having two bits. Yes? @cl \$\endgroup\$
    – Alexander Ohm
    Commented Mar 13 at 12:33
  • \$\begingroup\$ No; one chip can handle both signals. \$\endgroup\$
    – CL.
    Commented Mar 13 at 12:38
  • \$\begingroup\$ Well.. Only if I actively provide direction control logic? I cannot change either interface in my case. I only have supply, gnd, rx & tx. Right? \$\endgroup\$
    – Alexander Ohm
    Commented Mar 13 at 12:40
2
\$\begingroup\$

You can't expect a logic level communication to work over 3m, with that level shifter, and at such high speed.

You should switch to something that is intended for 3m communication between devices, such as RS-232 transceivers. On the 1v8 side, there would be a level shifter with few cm on each side.

\$\endgroup\$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.