AM5728: ESD device of AM572x EVM

g.f.

Part Number: AM5728
Other Parts Discussed in Thread: TPD12S016, TPD1E05U06

Hi,

My customer have a question about AM572x IDK and GP EVM.

On AM572x IDK EVM, TPD12S016 is used for ESD between AM572x HDMI I/F and HDMI connector.
But on AM572x GP EVM, TPD1E05U06 is used.

Are there any reason why different ESD device are used on this two EVMs?

best regards,
g.f.

over 5 years ago

0 Bill Taboada over 5 years ago

TI__Mastermind 42545 points

Hi g.f.
Both devices are rated for ESD protection for the HDMI interface but the TPD12S016 is a single device for the entire interface making it more difficult to route. The GP EVM used the single line TPD1E05U06 which can be more easily placed near each of the interface lines. The TPD12S016 was still used for the I2C, hot plug detect and the CEC lines. This device is good for 3.3V signalling but does have some issues when used as a 3.3V to 1.8V level converter. This has been discussed in E2E posts such as the following.
e2e.ti.com/.../2771535
Regards, Bill

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

Thank you for the reply.
I understood the reason why TPD12S016 and TPD1E05U06 are used.

best regards,
g.f.

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

I'm sorry but I have a related question about TPD12S016.

I'm confusing about the following part of above post:
/////////////////////////////////////////////////////////////////
"The device is good for 3.3V signalling but does have some issues when used as a 3.3V to 1.8V level converter."
/////////////////////////////////////////////////////////////////

I read the E2E post which you attached and also the linked post inside , but I'm not clear yet.
e2e.ti.com/.../2771535
e2e.ti.com/.../649475

I want to know the reason why there are issue when using 1.8V signaling for I2C
and the reason why TPD12S016 can be used when I2C signal is 3.3V.
I also read the datasheet of AM572x and TPD12S016 but could not find the problem.
Could you please tell me the details?

best regards,
g.f.

0 Bill Taboada over 5 years ago in reply to g.f.

TI__Mastermind 42545 points

Hi g.f,

As discussed in some of the posts, the AM572x does not provide a signal level that is low enough to meet the Vil threshold for the TPD12S016 for the load provided by the device. It does provide the proper voltage levels for 3.3V signalling.

Regards, Bill

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

Thank you for the reply,
But I'm sorry, I'm not clear yet.

AM572x I2C VOL3 is 0.2V * VDDS(MAX) when I2C Fast/Standard Mode - 1.8V
so that VOL3 = 0.36[V](MAX) at 3[mA] sink current.
And when I2C Fast/Standard Mode - 3.3V, it's VOL3 is 0.4[V] at 3mA sink current.

TPD12S016 I2C VIL is 0.082*VDD(MAX) so it will be as follow:
When 1.8V, VIL=0.1476V
When 3.3V, VIL=0.2706V

From above, it seem AM572x doesn't provide a signal level which meet TPD12S016 VIL threshold
in either 1.8V or 3.3V case.

But as Jim commented in following E2E, VIN of low level will be
0.021[V]@1.8V and 0.043[V]@3.3V .
So, it seems signal level from AM572x will be low enough against TPD12S016 threshold and
it seems there are no ploblem. Is my understanding wrong?
e2e.ti.com/.../649475

best regards,
g.f.

0 Bill Taboada over 5 years ago in reply to g.f.

TI__Mastermind 42545 points

Hi g.f,

The analysis provided by Jim includes a number of assumptions and we can't guarantee it is accurate for all parts. We do have information from the TPD12s016 team that the input stage uses a circuit that requires a Vin low reach a lower voltage level than is published in the UM10204 Rev 6 I2C specification. It will operate correctly if you meet the Vin low specified in the TPD12s016 data sheet. Although the AM572x drives the output to a low enough voltage to meet the I2C specification, we can't guarantee that it will meet the limit published in the TPD12s016 data manual.

Regards, Bill

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

A Happy new year!
Thank you for the reply.

I'm very sorry but may I ask following question?
I understood you can't guarantee that AM572x will meet the VIL limit of TPD12S016.
But then what is the reason for "AM572x does provide the proper voltage levels for 3.3V signalling"
which you told me in previous post?

From datasheet of AM572x and TPD12S016, we can't decide that TPD12S016 can be used with AM572x HDMI I2C.
So,does user actually need to check the output level of the AM572x HDMI I2C by mounting on the custom board
that whether TPD12S016 can be used with AM572x HDMI?

best regards,
g.f.

0 Bill Taboada over 5 years ago in reply to g.f.

TI__Mastermind 42545 points

Hi g.f,
It is true, based on the Vol specification for the AM572x. that the AM572x would not provide a signal level low enough to work with the TPD12S016. I had to reassemble some discussion from a while back on this subject. The I2C interface specification for Vol is based on a full current load of 3mA. That is much higher than the expected load from the TPD12S016 which should be closer to 100uA. The assumptions made in the other posts were based on the similarity between the dual voltage I2C IO buffer and the dual voltage buffer used for GPIOs. When we looked at the specs for that buffer at the more reasonable 100uA, the voltage level for 3.3V I2C operation for Vol was at 0.2V which is within the spec for the TPD12S016. This was not true for the interface using 1.8V signaling. It did not drop below the 0.1476V requirement for the TPD12S016.
Based on this we told customers that the AM572x would operate with the TPD12S016 using a 3.3V I2C interface. We can't change the data manual since I2C must be able to operate with the higher loads but a single connection to the TPD12S016 should work fine if you are using 3.3V.
Regards, Bill

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

Thank you for the reply.

>That is much higher than the expected load from the TPD12S016 which should be closer to 100uA.
How was this "100uA" calculated?
According to the TPD12S016 datasheet, this device have 10kohm(typ) pullup resistor internally at I2C line.
Since I2C is open drain, if VOL(max) is 0.2[V] as you said
the current load would be (3.3V - 0.2V)/10kohm = 310 uA, isn't it?

> When we looked at the specs for that buffer at the more reasonable 100uA,
> the voltage level for 3.3V I2C operation for Vol was at 0.2V

Since the specifications of the I2C buffer and the LVCMOS buffer used for GPIO are similar,
are you assuming that the VOL of I2C is also 0.2V(Max)@100uA?

best regards,
g.f.

0 Bill Taboada over 5 years ago in reply to g.f.

TI__Mastermind 42545 points

Hi g.f,
The 100uA wasn't calculated, it was a number based on observations for similar devices. We don't have characterization information on the I2C buffers for either the TPD12S016 or the AM572x beyond what was provided in the data manual. Based on these, we shouldn't be using the TPD12S016 for the I2C of the HDMI interface circuit but it was included on the EVM before this was understood. Based on internal discussions, we determined that the risk for 3.3V I2C was low but that the risk for the 1.8V interface was higher. We have not observed any issues with 3.3V but the EVM is used in a lab environment and we have not tested across temperature or for parts at different process levels. Based on this we recommend the use of a separate level translator to avoid this issue.
Regards, Bill

0 g.f. over 5 years ago in reply to Bill Taboada

Guru 15510 points

Hi Bill,

Thank you for supporting me so long time.
I understood and I will introduce other separate level translator to my customer.

best regards,
g.f.

Processors

Processors forum

AM5728: ESD device of AM572x EVM