Hi Chen, For VM TX we usually play with the impedance to modulate the swing when supply is fixed. Could you elaborate, how the supply is varied for swing adjustment? As calibrated impedance would change when supply is varied on the fly? Please let me know if I am missing something?
Hi vars008rcb, nice to meet you, and thank you for the good question. The supply can be adjusted by the regulator. But, the impedance calibration would not be applied on-the-fly; otherwise, the BER or SI would be degraded a lot.
@@circuitimage Hi Chen, Maybe I was not clear. I meant to say we calibrate the impedance for a given supply and process corner. Now if the supply is adjusted after impedance calibration, aren't we disturbing the calibrated 50 ohms?
@@vars008rcb Hi vars008rcb, thank you for the good clarification. The impedance calibration can be done at power-up, which is an offline calibration. In that case, you are correct. The impedance would be less matching. On the other hand, the impedance calibration can be done with a replica tracking loop, which is a background calibration running continuously. If the loop bandwidth was designed well, the signal integrity could be still maintained well. Thank you so much. 🙂
@@vars008rcb Hi vars008rcb, thanks for the clarification. You are right, if the supply is adjusted after impedance calibration is done, the new impedance won't match well and that does disturb the 50ohms; therefore, people may have other options of doing the replica calibration loop background to calibrate the impedance continuously, which would be adjusted when the supply changed. Thank you so much. 😄
Hi 正宇, 感謝您的反饋和好問題。你是對的。 1. TX FFE 或 RX CTLE 都是信號緩衝器。 2. TX FFE 將驅動 100ohm（差分），這是低阻抗和高功耗。 3. 正如你提到的，TX 驅動程序將在通道之前包括 FFE 和 TX 終端；因此，我們也需要在通道之前進行阻抗匹配。 Hi, 正宇 Thanks for the feedback and good question. You are correct. 1. Either the TX FFE or The RX CTLE are signal buffers. 2. The TX FFE would drive the 100ohm (in differential), which is a low impedance and high power consumption. 3. As you mentioned, the TX driver will include the FFE and the TX termination before the channel; therefore, we need the impedance matching before the channel as well.
The fast SR would contain more frequency content. The reflection comes from a wide band of frequency matching. Usually, you may not have a good matching at the high frequency because of the big ESD requirements. For example, if your Z matching only matches at 1GHz, but you're sending 10GHz content from the fast SR in the TX driver output, you must receive terrible reflection because of a poor matching at 10GHz (very poor return loss)
Thanks for your share, the videos are all valuable for me.😄
Hi Zm Lee, nice to meet you,😀 and thank you for all your feedback. I'm glad that helps.
Great video, as always!
Glad you enjoyed it!
Hi Chen, For VM TX we usually play with the impedance to modulate the swing when supply is fixed. Could you elaborate, how the supply is varied for swing adjustment? As calibrated impedance would change when supply is varied on the fly? Please let me know if I am missing something?
Hi vars008rcb, nice to meet you, and thank you for the good question. The supply can be adjusted by the regulator. But, the impedance calibration would not be applied on-the-fly; otherwise, the BER or SI would be degraded a lot.
@@circuitimage Hi Chen, Maybe I was not clear. I meant to say we calibrate the impedance for a given supply and process corner. Now if the supply is adjusted after impedance calibration, aren't we disturbing the calibrated 50 ohms?
@@vars008rcb Hi vars008rcb, thank you for the good clarification. The impedance calibration can be done at power-up, which is an offline calibration. In that case, you are correct. The impedance would be less matching. On the other hand, the impedance calibration can be done with a replica tracking loop, which is a background calibration running continuously. If the loop bandwidth was designed well, the signal integrity could be still maintained well. Thank you so much.
🙂
@@vars008rcb Hi vars008rcb, thanks for the clarification. You are right, if the supply is adjusted after impedance calibration is done, the new impedance won't match well and that does disturb the 50ohms; therefore, people may have other options of doing the replica calibration loop background to calibrate the impedance continuously, which would be adjusted when the supply changed. Thank you so much. 😄
老師您好，想請問您
driver 的明確定義，有包含哪些部分。
以FFE來說，是接在Channel之前。 即0:39處，
那我可以說ffe equalizer都算是一種driver嗎
還是討論tx driver時，是包含ffe 以及阻抗的termination。
即訊號先經過FFE 再經過 阻抗匹配才會送進Channel。
Hi 正宇,
感謝您的反饋和好問題。你是對的。
1. TX FFE 或 RX CTLE 都是信號緩衝器。
2. TX FFE 將驅動 100ohm（差分），這是低阻抗和高功耗。
3. 正如你提到的，TX 驅動程序將在通道之前包括 FFE 和 TX 終端；因此，我們也需要在通道之前進行阻抗匹配。
Hi, 正宇
Thanks for the feedback and good question. You are correct.
1. Either the TX FFE or The RX CTLE are signal buffers.
2. The TX FFE would drive the 100ohm (in differential), which is a low impedance and high power consumption.
3. As you mentioned, the TX driver will include the FFE and the TX termination before the channel; therefore, we need the impedance matching before the channel as well.
@@circuitimage Thanks for your reply!
it’s helpful
@@王正宇-i1l You're very welcome and I'm glad that was helpful :)
could you explain how fast SR will impacts us in terms of frequency domain .I didn't understand about reflection .
The fast SR would contain more frequency content. The reflection comes from a wide band of frequency matching. Usually, you may not have a good matching at the high frequency because of the big ESD requirements.
For example, if your Z matching only matches at 1GHz, but you're sending 10GHz content from the fast SR in the TX driver output, you must receive terrible reflection because of a poor matching at 10GHz (very poor return loss)
@@circuitimage got it .I understand it now.
Thank you for your reply.
@@rangababuganta7725 You're welcome :)