TLK2711EVM-CVAL: About RX_CLK and PRBS_PASS

<Translation to English>

Hello.

TLK2711EVM is being used to perform loopback tests.

Enter 100MHz for GTX_CLK before connecting the cables between TX and RX.

however, the CLK waveform is output to RX_CLK. Is the output correct even if the cable is not connected between TX and RX?

The next question is PRBS_pass, with the cables connected between TX and RX,

However, if you disconnect from either TXP or TXN, you get PRBS_pass and also get RX_CLK.

Is this correct behavior?

The jumper selection for J7 is PRBSEN Hi LCKREFN Hi enable Hi.

Please answer.

Hi,

I'm assuming PRBSEN, LCKREFN, and ENABLE are high for both scenarios.

1. TXD0-15 inputs are sampled on the rising edge of GTX_CLK. Even though these inputs are left floating, my assumption is that the device is still sampling these inputs and looping them back to RXD0-15. This is likely why you see an output clock on RX_CLK. I recommend connecting input signals to TXD0-15 before connecting GTX_CLK.

2. What signal is being transmitted to TXD0-15 in this scenario?

Best,

Lucas

こんにちは、ルーカス。

この度は迅速な対応をいただき、誠にありがとうございました。

確認している回路は、取扱説明書の図2-2です。

TXとRXの間に用意したケーブルを接続して、問題がないか確認するのが目的です。

PRBSEN、LCKREFN、および ENABLE を Hi に設定し、疑似ランダム ビット パターンを送信し、PRBSPASS (RKLSB) ピンが Hi を示していれば問題ないと理解しています。

擬似ランダムビットパターンはデバイス内部に送信されるため、TXD0-15に信号は入力されないと理解しています。

私の理解に何か問題がある場合はお知らせください。

TXとRXの間のケーブルを外すとRX_CLKが出力されないのはわかりますが、それは間違っていますか?

また、ケーブルを抜いているのにPRBSPASS(RKLSB)ピンがHiになっている理由がわかりません。

説明してください。

添付の図は

(1)確認中のブロック図

(2)TX-RXをケーブルで接続して波形を確認

CH1:PRBS(Hi) CH2:RXD0 CH3:RX CLK(80MHz)

(3)TX-RX間の接続が切断されたときの波形

CH1:PRBS (Lo) CH2:RXD0 CH3:RX CLK(80MHz)

(4)TXPのみ切断時の波形

CH1:PRBS(Hi) CH2:RXD0 CH3:RX CLK(80MHz)

<Translation to English>

Hello, Lucas.

Thank you very much for your prompt response.

The circuit we are checking is Figure 2-2 in the operation manual.

The purpose is to connect the cables you have prepared between TX and RX and check for problems.

I understand that if I set PRBSEN, LCKREFN, and enable to Hi, send a pseudo-random bit pattern, and the PRBSPASS (RKLSB) pin shows Hi, I understand that this is OK.

Because the pseudo-random bit pattern is transmitted inside the device, we understand that there is no signal input to TXD0-15.

If you have any problems with my understanding, please let me know.

I know that when I disconnect the cable between TX and RX, RX_CLK is not output, is that wrong?

Also, I do not know why the PRBSPASS (RKLSB) pin is high when the cable is unplugged.

Please explain.

The attached figure is

(1) Block diagram under review

(2) Connect TX-RX with cable and check waveform

CH1:PRBS(Hi) CH2:RXD0 CH3:RX CLK(80MHz)

(3) Waveform when the connection between TX-RX is lost

CH1:PRBS (LO) CH2:RXD0 CH3:RX CLK(80MHz)

(4) Waveform when TXP is disconnected only

CH1:PRBS(Hi) CH2:RXD0 CH3:RX CLK(80MHz)

Hi,

I'm still looking into this and will get back to you with more feedback tomorrow.

Best,

Lucas

<Translation to English>

Hi Lucas, Thank you for your quick response. Followed power-on reset procedure. It was confirmed that the waveform after reset was at the level described in the data sheet. We then set PRBSEN to Hi without connecting a cable between TX and RX and verified that the PRBS Lo RX CLK is at a low level. I understand that there is no problem so far, but please let me know if you have any comments. So when I connected the cable, the RX CLK was output from the moment I touched the connector, and it was still output even after the cable was disconnected. There was phase variation compared to TX CLK. If you can work around this issue, let me know what to do. Or, if it is possible with this device, we will also understand it. Thank you very much.

About the attached image

(1) J7 jumper setting and waveform recording during power-on reset

(2) Wave recording after PRBSEN is set to Hi (cable between TX and RX is not connected)

(3) Waveform recording of RX CLK output from the moment the connector is touched 

Hi,

Thank you for your follow-up, I agree it is unexpected that RXCLK output continues after DOUTTX pins are disconnected from DINRX pins.

I'm wondering if it's possible that the LCKREFN pin is somehow getting pulled low. When LCKREFN is low, the receiver clock is frequency locked to TXCLK and the device is placed in a transmit-only mode. This would explain why RXCLK output continues when DINRX inputs are floating.

Can you perform the same experiment, but measure voltage on the LCKREFN pin at 3 different times? Can you share these measurements with me?

1. After pulling PRBSEN low. DOUTTX pins are still disconnected from DINRX pins. (I expect there is no RXCLK output at this time.)
2. After connecting DOUTTX pins to DINRX pins. (I expect RXCLK output begins at this time.)
3. After disconnecting DOUTTX pins from DINRX pins. (I expect RXCLK output continues at this time, even though this behavior is abnormal.)

Best,

Lucas

こんにちは、ルーカス。

迅速な指示をありがとう。

測定結果をお送りします。

LCKREFN は常に "Hi" でした。

その他、測定や確認が必要なことがあればお知らせください。

感謝。

<Translation to English>

Hi,

Thank you for taking these measurements. I see that the LCKREFN pin remains high through the duration of the test. I discussed this behavior with my colleagues and now have a better understanding of why RXCLK output may be continuing after DOUTTX pins are disconnected from DINRX pins.

The TXCLK input is used as a reference clock to source the PLL. The PLL locks to the incoming signal on the DINRX pins and recovers the bit clock which is output to RXCLK. We believe that the PLL may not be losing lock when the input signal is removed from the DINRX pins, therefore allowing the RXCLK output to continue.

Although this behavior is unexpected, it is unlikely to cause any issues in typical system implementations. Do you anticipate the DINRX input signal being intermittently turned on/turned off? Is there a particular reason you are investigating RXCLK behavior when DINRX input is removed?

Best,

Lucas

こんにちはルーカス

迅速な対応をいただきありがとうございます。

PLL ロックは失われておらず、RXCLK 出力は続行できると理解しています。

RXCLKについては、原因が特定されたと理解していますので、大丈夫です。

また、前回の回答では、ケーブルのTXP側を抜くとPRBSが「Hi」になり、RXP電圧からRXN電圧を差し引くことで入力データ値を正しく解釈できるとおっしゃっていました。

今回は接続ケーブルを短いものと長いものの両方で測定してみましたが、TXP側を抜くと、PRBSは常に短いケーブルで「Lo」になっていました。長いケーブルでは、PRBSは常に「Hi」でした。

どちらも可能であると理解して大丈夫ですか?

こちらで考えているのは、通信中にケーブルが断線しているかどうかを判断するためにPRBSのステータスを確認することですが、これは不可能であると理解して正しいですか?

よろしくお願いいたします。

添付画像について                                                                                                                                                                                                                           (1) 50cmケーブルによるPRBSステータス                                                                                                                                                                                     (2) 50cmケーブルによるPRBSステータス(TXP断線)                                                                                                                                                                     (3) 5mケーブルによるPRBSステータス                                                                                                                                                                                         (4) 5mケーブルによるPRBSステータス(TXP断線)                                                                                                                                                                       (5) 5mケーブルによるPRBSステータス(TXP断線)(100ns/DIV)

<Translation to English>

Hello, Lucas

Thank you for your prompt response.

I understand that the PLL lock is not lost and the RXCLK output can continue.

As for RXCLK, I understand that the cause has been identified, so it's okay.

The previous answer also stated that when the TXP end of the cable is unplugged, the PRBS becomes "Hi" and that the RXN voltage is subtracted from the RXP voltage to correctly interpret the input data value.

This time, I tried to measure both short and long connection cables, but when I pulled out the TXP side, the PRBS was always "Lo" with a short cable. On long cables, the PRBS was always "Hi".

Is it okay to understand that both are possible?

What we are thinking is to check the status of the PRBS to determine if the cable is broken during communication, is this correct with the understanding that this is not possible?

Thank you very much.

About the attached image

(1) PRBS status over 50cm cable

(2) PRBS status via 50cm cable (TXP disconnected)

(3) PRBS status via 5m cable

(4) PRBS status via 5m cable (TXP disconnected)

(5) PRBS status via 5m cable (TXP disconnected) (100ns/DIV)