TCA9803: 400kHz fast mode rise/fall time check

Fred Cheng

Expert 5830 points

Part Number: TCA9803

Hi team,

I'm trying to convince our customer to use our buffer on their design, they sent the current waveform now.

does this rise/fall time seems ok to you?

1. rise time : I recall the rise time must below 300ns in 400KHz, the rise time below is over 300ns..

2. fall time: ??

over 2 years ago

0 Clemens Ladisch over 2 years ago

Guru 317480 points

The I²C specification requires an upper limit of 300 ns to ensure interoperability. The TCA980x works with any rise time.

The TCA980x's current source on the B side will speed up rising edges.

0 Fred Cheng over 2 years ago in reply to Clemens Ladisch

TI__Expert 5830 points

Hi Ladisch,

1. customer can speed up the the rising time by decreasing the pull-up resistor as well, is there breaking point for us?

2. so how about fall time?

since fall time bottom limit is 20x3.3/5.5=12ns, their fall time is 16ns now, which is close ,

will the fall time be even shorter if we add TCA9803?

0 Clemens Ladisch over 2 years ago in reply to Fred Cheng

Guru 317480 points

1. The I²C specification limits the current to 3 mA, or 6 mA for a high-capacitance bus with relaxed V_OL. The TCA9803 is rated for 6 mA. Please note that you must not use pull-up resistors on the B side of the TCA980x; the pull-up current is supplied by the TCA980x itself.

2. The TCA980x's outputs conform to the I²C specification, i.e., the fall time will be OK. (It does not change the fall time of other devices on the same bus.)

0 Fred Cheng over 2 years ago in reply to Clemens Ladisch

TI__Expert 5830 points

Hi Ladisch,

so you also agree the rise time of customer seems slow and needs buffer?

just want to confirm with you on this

0 Clemens Ladisch over 2 years ago in reply to Fred Cheng

Guru 317480 points

I do not know if the oscilloscope measures the rise time in the same way as the I²C specification (between 30 % and 70 % of V_DD).

The signal waveforms actually look OK.

0 Tyler Townsend over 2 years ago in reply to Fred Cheng

TI__Mastermind 49548 points

Hi Fred,

The max rise time is 300ns for i2c spec at 400kHz from 30% to 70% of VCC. Does the fall time measure from 30% to 70%, or from a different range? If the fall time measures from say 10% to 90%, then this waveform capture looks to be within spec.

If you want to make the rise time faster, increase strength of pull-up resistors or think about buffering solutions.

Regards,

Tyler

0 Fred Cheng over 2 years ago in reply to Tyler Townsend

TI__Expert 5830 points

Hi Tyler,

but the rise time as you can , already exceed 300ns spec,

it's 30%~70%, which means it's out of spec already?

0 Tyler Townsend over 2 years ago in reply to Fred Cheng

TI__Mastermind 49548 points

Hi Fred,

Just to clarify, you are saying that the oscilloscope reads rise time from 30% to 70% VDD? I know some other scopes that read rise time from 10% to 90%.

If your scope is reading 10% to 90%, and the rise time = ~300ns, then you would pass I2C spec since 30% to 70% is a smaller margin.

Regards,

Tyler

0 Fred Cheng over 2 years ago in reply to Tyler Townsend

TI__Expert 5830 points

Hi Tyler,

I got your mean,

but if this is 30% to 70% instead, that means they doesn't pass the spec ,right?

+1 Clemens Ladisch over 2 years ago in reply to Fred Cheng

Guru 317480 points

Yes; if that oscillope is using 30% to 70% (which I doubt), then the signal would be outside the spec.

Interface

Interface forum

TCA9803: 400kHz fast mode rise/fall time check