• TI Thinks Resolved

TXB0108: Significant noise observed on channels

Prodigy 110 points

Replies: 8

Views: 110

Part Number: TXB0108

Hi TI community,

We're working with a ADC chip test setup that uses the TXDB0108 to step up 1.8V data lines to 2.5V that are connected to Kintex7 FPGA.

The pin-out is as follows:

 VCCA is 1.8V, while VCCB is on a 2.5V rail.There are series resistor (value 10 ohm) on each of the data lines on both sides of the level translator.

The sample rate of the ADC chip under test is 2.5MSPS. OE is pulled up to VCCA through a 10 K Ohm resistor. We have 0.1uF bypass capacitors on both supplies. The issue that we are currently facing is that whenever ADC starts conversion and the data is available on the lines we observe high frequency noise that gets coupled to the supply. When there is no data on the channels of the level translator we do not see this noise. To confirm if this noise is coming from the ADC or from the level translator we disconnected the chip board of ADC and supplied a square wave to one of the channels. The frequency of the signal was 2.5MHz and we observed the same noise pattern on the data lines and the supply.

In this image channel 1 is the 2.5MHz signal at the B side of the channel of the level translator and channel 2 is VCCA of the level translator which is 1.8V. We are observing oscillations in the signal. 

 The image below has channel 1 showing the A side of the channel of level translator which is the supplied 2.5MHz signal from a source and channel 2 showing the VCCB of the level translator which is 2.5V. The noise appears on both supplies.

As you can see there is significant noise that appears on the signal which is getting coupled to the DC supply. 

Now we had previously installed 0.1uF bypass caps close to the level translator for proper operation but after observing this noise we soldered 10nF in addition to the 0.1uF caps. To our disappointment there was no improvement seen. 

We need your help to remove this noise as this is reducing the SNR of the ADC chip. Please let us know if you need any more information. Urgent help would be appreciated.

Maven M.

  • The TXB uses edge accelerators. Apparently, an ege on the signal line causes ringing that is large enough to be detected as the beginning of an edge in the opposite direction, which triggers the TXB to drive in the opposite direction.

    Are there long traces, connectors, cables, or anything else that could cause a capacitive load on the signal lines?

  • In reply to Clemens Ladisch:

    Hi Justin,

    Can we get a schematic of your circuit? 

    Check out our new AXC family of Voltage Translators! 

    Watch the Introduction to AXC Family to find out more about the family.

  • In reply to Clemens Ladisch:

    We have pin headers as connectors on evaluation board of ADC chip and a ribbon cable to connect it to the FMC breakout board XM105 of KC705. The ribbon cable is 5 inches long. 

  • In reply to Justin Kiln:

    Section 8.3.3 of the datasheet says:

    TI recommends careful PCB layout practices with short PCB trace lengths to avoid excessive capacitive loading and to ensure that proper O.S. triggering takes place.

    Apparently, the cable has increased the total capacitive load to more than 70 pF.

  • In reply to Clemens Ladisch:

    We captured these photos after disconnecting the cable. How short do the PCB traces need to be for the chip to function without this happening? 

  • In reply to Dylan Hubbard:

    Let me know if you need anything else?

  • In reply to Justin Kiln:

    The datasheet says:

    PCB signal trace-lengths should be kept short enough such that the round-trip delay of any reflection is less than the one-shot duration.

    An open connector without termination can also lead to reflections.

    Is the situation better with the actual circuit?

  • In reply to Clemens Ladisch:

    Hi Justin,

    I'm not seeing anything wrong with the schematic. These problems typically result from heavy loading on the device I/O as Clemens has pointed out. Can I get two more scope shots showing both A port and B port I/O, one with the ribbon cable connected and the without.

    Check out our new AXC family of Voltage Translators! 

    Watch the Introduction to AXC Family to find out more about the family.