TCA9800: 1.8V-3.3V level shift, 1:3 MUX, bi-directional application

Jerry Chen

Part Number: TCA9800
Other Parts Discussed in Thread: TCA9617A, PCA9546A, TCA9546A, TCA9509

Hi team,

My customer is looking for a 1.8V-3.3V level shift, 1:3 MUX, bi-directional I2C solution. They chose the wrong solution exactly the same with the below post, which TCA9617 B side is interfacing with the B side of another 9617.

I would like to confirm. Could they P2P replace TCA9617 with TCA9800 to solve this issue? They can either just replace the right one, or replace both TCA9617.

Or, do we have better solution for this application?

Thanks.

TCA9517: B side connect to B side through PCA9546 MUX - Interface forum - Interface - TI E2E support...

e2e.ti.com

Part Number: TCA9517 Hi Sir, As my understand, 2 pcs TCA9517 can't connect both B-side together. And I want to double confirm if we add a NXP PCA9546 I2C MUX

over 4 years ago

0 Danny Bacic over 4 years ago

TI__Mastermind 26776 points

Jerry,

Would you be able to verify the part number for me? Did you mean that they're using the TCA9517? Or the TCA9617A or TCA9617B?

The TCA9800 features current driving, so it would not necessarily be P2P. It might just come down to a matter of DNP-ing a resistor. Are you able to share any more details about this part of the design?

Best,

Danny

0 Jerry Chen over 4 years ago in reply to Danny Bacic

TI__Expert 5110 points

Hi Danny,

I attach the schematics. TCA9617A B side is interfacing with another TCA9617A B side. As VOL(0.53V) is higher than VIL(0.4V), B side can't drive B side, so I'm thinking changing TCA9617A to TCA9800, whose VOL is 0.22V to drive its B side VIL 0.3*VCCB.

Thanks.

0 Danny Bacic over 4 years ago in reply to Jerry Chen

TI__Mastermind 26776 points

Jerry,

A few comments.

First, I highly recommend switching the PCA9546A to the TCA9546A if possible. This is a drop-in replacement and is the newer technology.

Next, you're correct that the TCA9800 could be used. For this device, you have to ensure that the B-sides do not face each other just like the TCA9617A. Additionally, the B-sides of the TCA9800 should not have pull-up resistors. See the image below for an example:

This would mean connecting the B-side of the device to the controller, which would require a board spin, since it would not be a drop-in fix for the current schematic if a board design has already been made.

Another potential solution is to use the TCA9509, which could serve as a drop-in replacement for the current routings of the TCA9617A devices above. This device allows the B-sides to face each other:

Note that the A-sides of the TCA9509 require no pull-up resistors, so you'd have to DNP any of these pull-up resistors in design (shown as "Bus A" and "Bus C" in Figure 6 above).

Please let me know if I can help with this any further.

Best,

Danny

0 Danny Bacic over 4 years ago in reply to Danny Bacic

TI__Mastermind 26776 points

Jerry,

I figured I'd add a little more explanation to my response. The TCA9800 B-sides can't face each other because the B-side uses a current source for this device. There cannot be more than one current source on the same bus, as this would cause the device to fail. This is a different reason than why the TCA9617A device B-sides can't face each other.

The TCA9509 is very similar, but it has the current sources on the A-side. This device doesn't allow the A-sides to face each other, which is not a problem in the current design.

I hope this explanation helps!

Best,

Danny

0 Jerry Chen over 4 years ago in reply to Danny Bacic

TI__Expert 5110 points

Thanks, Danny.

I don't understand why more than one current source on the same bus would cause the device to fail. Is it because the current is too big?

0 Clemens Ladisch over 4 years ago in reply to Jerry Chen

Guru 299980 points

The TCA980x measures the current to determine whether to transmit from A to B or from B to A. If another device (or pull-up resistor) on the bus segment sources current, this measurement fails.

0 Danny Bacic over 4 years ago in reply to Jerry Chen

TI__Mastermind 26776 points

Jerry,

Clemens is correct! You already correctly mentioned that the reason B-sides of the TCA9617A can't face each other is due to voltage offset. For clarification, this offset exists so that the device doesn't incorrectly redrive the A side by seeing its own low signal on the B-side, causing a stuck bus. But, if a B-side device pulls down, it will pull lower, so the input to the B-side will only recognize that it needs to drive the A-side when the voltage drops even lower, which the B-side of another TCA9617A can't do.

Jerry Chen82343 said:

I don't understand why more than one current source on the same bus would cause the device to fail. Is it because the current is too big?

Now, to clarify with the TCA9800, it doesn't rely on the voltage level to determine whether a B-side device is pulling down. Instead, it relies on current. This is what allows the V_OL of the TCA9800 to be so low compared to the TCA9617A. As Clemens said, the presence of another current source on the B-side of the TCA9800 would disrupt the current measurement of the TCA9800, which is what it relies on to know whether to drive the A-side low.

For the TCA9509, the current driver is on the A-side, so you also wouldn't be able to connect the A-side of the TCA9509 to the B-side of the TCA9800, for example.

I hope this helps clarify! Please let me know if I can help any further!

Best,

Danny