• State TI Thinks Resolved
  • Locked Locked
  • Replies 4 replies
  • Answers 2 answers
  • Subscribers 34 subscribers
  • Views 387 views
  • Users 0 members are here
Support feedback
Options
Options
Related

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ONET1191P: Output showing a 6Ghz oscillation on custom PCB

AnBer
Mastermind 30110 points
Part Number: ONET1191P
Other Parts Discussed in Thread: ONET8501P, ONET1151P, ONET1151L

Team,

Could you please help with the below question:

For this project we want to amplify the signal delivered from an external fast photodiode amplifier of a third-party company via coaxial cable (bandwidth up to 10 GHz) so that it can be processed by a fast frequency divider .
We came across TI's "Optical networking ICs", for example the ONET1191P and ONET8501P. but have no experience with such devices.
We saw there are no EVM boards for these components so we designed our own PCB for evaluation (see pics attached) but we have issues the board.

The two boards with the ONET1191P show the following behavior:
"When the input is not terminated, the amplifier becomes an oscillator, and this is consistent with both PCBs.
This oscillating output frequency is typically around 6GHz and can be considered (at least in the two board we designed) as an "independent LO-level" in normal operation. This means that an input signals with lower levels (<-20dBm) and higher frequencies (>1GHz) (can) generate new output frequencies that are at n*F(e) +/- m*F(LO), thus appearing as a kind of "multimirror".
If signals above approx. 6GHz are applied to the input, no signal at all appears at the output that can be directly derived from the input, but only mixed signals from the upper sideband, and even the input level plays only an insignificant role.
This oscillating behavior can be influenced (if necessary) by putting a shield over the component - the undesired circuit behaviour may be reduced, or perhaps just shifted in the frequency products, or even underlined."
So the circuit does not seem to behave as we think it should.
I will provide picture of the boards offline as well as a block diagram of what we want to achieve in this project.

-Does this behavior sound familiar?
-Are there some ready to use EVM for ONET1191P and ONET8501P that can be used for evaluation?
-Could you provide feedback about the board/PCB we designed?
-Do we have some PCB guidelines, reference designs and even tips for the board design (like placement and type of connectors,..etc)

Thanks in advance,

A.

  • 0
    TI__Guru* 93670 points

    Hi Anthony,

    You had already sent email on this issue as well along with customer schematic. This is very helpful.

    Given the customer schematic comments, i will respond to this question shortly through email you had sent.

    Regards,,Nasser

  • 0 in reply to Nasser Mohammadi
    TI__Mastermind 30110 points

    Hi Nasser,

    I am copying your email answer below so that customer can answer directly:
    "I can see customer is using ONET1191P in single ended mode and it is AC coupled. Given the input is not terminated and we are AC coupling, it is possible to get oscillation since we are AC coupled. To prevent this oscillation, it is best to DC couple both positive and negative inputs.
    Has the customer tested the part with termination and active signals?  Do they observe any issue under this condition?"

    Also I have seen the below ref design for on TI.COM where Schematics/PCBs are available:
         https://www.ti.com/tool/TIDA-00088 (using ONET1151P / ONET1151L)
    Do we have further design documents (like schematics/PCB guidelines)?

    Thanks in advance,

    A.

  • 0 in reply to AnBer
    Prodigy 120 points

    Hi, I am not sure what you mean by DC-coupling. Can you please provide a snippet of a schematic how to do that properly?

    Do we need a termination resistor from each input to GND? Or should we just replace the input series capacitors with a 0-Ohm-resistor? But even then the input signal stays AC-coupled, because the output of the external photodiode amplifier where the input signal comes from is AC-coupled and we cannot change this.

    We connect the input pins of the ONET1191P in almost the same way as it is shown for the input pins of the ONET1151P in the reference design https://www.ti.com/tool/TIDA-00088 , except that we ground one of the inputs via a capacitor.

    Please advice in detail what we should try, thanks!

  • 0 in reply to Christian Rausch66
    TI__Guru* 93670 points

    Hi Christian,

    I need to correct my earlier statement. ONET1191P is designed to be AC coupled - meaning there must be Capacitors on DIN+, DIN-, DOUT+, and DOUT-.

    Also, typically we typically use ONET1191P in differential mode versus single ended. This is because device would see twice the amplitude of the signal.

    Looking at your schematic, the un-used input should be terminated with 50-ohm to GND. In your schematic i see AC coupling Cap directly to GND. We should  go to AC coupling Cap as you have in your schematic and then a 49.9-ohm Resistor to GND.

    Also, single ended input used should have active signal. This is because device has a low frequency cutoff(~30KHz). Otherwise we can observe oscillations on the output.

    Regards,Nasser