DS90CR287 failures

Hi Tom,

The DS90CR287 and DS90CR288A are designed to be used in an LVDS envrionment, and therefore are designed to comply or exceed TIA/EIA-644-A LVDS specifications. Therefore, I searched in both the datasheet and LVDS specification to see if I could shed some light on the questions you asked.

1. The Common Potential Difference (Vcpd) parameter is also described as "Common-Mode Noise Margin." LVDS supports an input voltage that is measured from GND to +2.4 V. Given that the transmitter transmits with an offset voltage of 1.2 V, this means you have at most a +/-1V shift in the common mode voltage between transmitter and receiver to account for potential ground differences and common-mode noise. The maximum voltage applied on the terminals at the receiver should not be greater than 2.4 V or less than 0 V with respect to the receiver GND.

2. The LVDS output from the DS90CR287 is limited by the transmitter IC, and therefore it will not exceed a VOD differential output (TXOUT+ minus TXOUT-) of 450 mV (or 900 mVpp). However, according to the LVDS spec, the receiver can technically tolerate an input differential voltage that exceeds 450 mVpp differential. The maximum allowable input differential voltage that the receiver can handle is 600 mV (or 1200 mVpp). There is no direct consequence listed for exceeding this limit, but it is possible that the receiver pins can be damaged by overstressed voltage.

3. The datasheet ensures that the maximum Vos delta is 35 mV from the transmitter. In the LVDS spec, the allowable offset voltage range is actually less than or equal to 50 mV. Exceeding this Vos delta difference can lead to the introduction of common-mode noise into the differential signaling.

4. The latch-up tolerance abs max is not a standard item that normally goes into the datasheet, and I was not able to find specific information corresponding to the history of how we took the data. However, I believe this spec is the amount of supply current that the transmitter or receiver can draw before it goes into a latch-up state and is damaged beyond practical use. I am not sure whether you will find any incentive by attempting to drive over 300 mA into the Tx or Rx device.

Please let us know if we can help with the debugging. Do you have any more details about the failure? For example, what is the operating frequency, and how much loss or cable skew are you expecting due to the cable? Any scope shots perhaps?

Thanks,

Michael

Sorry I don't have many more details about the failures at this point. These transmitters are failing in the field. Our customer produces the equipment on the transmit side. Their customer who is remote, is the receive side and is somewhat a black box.

All transmit units are tested before being sent into the field (to their customer). ~30% of the units are failing with the transmitter damaged although there is no physical indication of the damage. One the transmitter is damaged, it never works again until the DS90CR287 is replaced and then the units works reliably again.

The cable length is in the 30m range I believe and the transmit and receive units power and ground are independent of one another. I believe the shield of the cable is connected to the digital grounds on each side. Our customer has concerns over the ground potential difference and is trying to figure out the following regarding your device.

What level of ground potential difference can result in data integrity issue?
What level of ground potential difference can result in a device damaged?

Their hops is that they can define parameters for their customer to adhere to.

Hi Tom,

I think to ensure proper data integrity and to avoid device damage, I recommend that the maximum amount of ground potential difference corresponds to the ground potential difference that would cause either of the LVDS signal polarities to reach the 0 V minimum or 2.4 V maximum at the receiver end. The specification defines that the incoming signal on the receiver side must be within the range of 0 V to 2.4 V, but the LVDS signal amplitude may arrange from 100-600 mV. Therefore, for example, if the LVDS signal is +/-100 mV with common-mode voltage at 1.2 V, then the ground potential of the receiver compared to the ground potential of the transmitter can be a difference of +/-1.1V so that the maximum LVDS differential voltage does not exceed 2.4 V if ground potential becomes 2.3V and the minimum LVDS differential voltage does not fall below 0 V if ground potential becomes 0.1V.

I do not have a clear picture of what amount of ground potential difference will cause permanent damage to the device.

Thanks,

Michael

Other LVDS standards like RS422 define an allowable ground potential difference of +/-7V.

See the bottom of page of this TI document.
www.ti.com/.../snla049b.pdf

Hi Tom,

To clarify, only TIA/EIA-644 is considered LVDS (which is also slightly different from standards defining M-LVDS and B-LVDS). RS-422 and RS-485 in themselves are other interface standards for high-speed signaling rates. You are correct that RS422 has a larger allowable ground potential difference, but it is also not equipped to handle the data rates that the LVDS interface is capable of.

I found a document comparing RS-422/RS-485/LVDS in an application note from Fairchild that is pretty insightful regarding differences in signaling amplitude and common mode voltage range:

Thanks,

Michael