LMX2582: LMX2582 18nH design & MUXout issues

Phil Lin2

Part Number: LMX2582
Other Parts Discussed in Thread: LMK1C1102

Tool/software:

Dear Sir/Madam,

We are designing 18nH to get more higher gain. If we want to use RFoutAP to get single-end out, may I know how to solve RFoutAM, which is not used.

RE Synthesizer_TI_LMX2582-20250401-1.msg

In this attached e-mail, the RED mark there, the channel is not used, so does it

Pull 49.9ohm to GND?
Or other?

Please suggest. One more, may I know how to link MUXout in schematic?

Thank you.

5 months ago

0 Phil Lin2 5 months ago

Prodigy 20 points

One more, may I know whether one TXCO can support two LMX2582 such as this photo? Thank you.

0 Noel Fung 5 months ago in reply to Phil Lin2

TI__Guru** 108270 points

Hi Phil,

Your schematic on RFoutA is correct. The unused RFoutAM pin should be pulled-up with the inductor and then 50Ω AC shunt to ground.

Looks you are not using RFoutB, you can leave the pins floating.

if you want to read back the register, you should connect MUXout to a processor. If you want to have know the lock status, connect MUXout to a LED and/or connect MUXout to a processor.

if the TCXO output is clipped sine wave, it cannot drive 50Ω load. Remove R133 and change R135 to 0Ω. You can share CMOS TCXO output to two synthesizers, if TCXO output is clipped sine wave, suggest use a buffer to make two copies.

0 Phil Lin2 5 months ago in reply to Noel Fung

Prodigy 20 points

Dear Noel,
Thnak you for your help. About "use a buffer to make two copies," please suggest us TI buffer part. Thank you.

0 Noel Fung 5 months ago in reply to Phil Lin2

TI__Guru** 108270 points

Hi Phil,

We suggest LMK1C110x, they are very low additive jitter buffers.

0 Phil Lin2 5 months ago in reply to Noel Fung

Prodigy 20 points

Dear Noel,
Thank you for suggestion. There are 1:6 and 1:8 LVCMOS clock buffers. May I know whether you have 1:2 LVCMOS clock buffers? Thank you.

0 Phil Lin2 5 months ago in reply to Phil Lin2

Prodigy 20 points

Dear Noel,
I saw LMK1C1102 from TI web, which is 1:2 LVCMOS clock buffers. Thank you.

0 Phil Lin2 5 months ago in reply to Phil Lin2

Prodigy 20 points

Dear Noel,
Attached "2582 Schematic-20250407-1" with LMK1C1102. Please help to review it. Thank you.

0 vicente flores prado 5 months ago in reply to Phil Lin2

TI__Mastermind 29282 points

Hello Phil - in regards to the LMK1C1102 if the input is a Sine Wave - please bias CLKin input to VCC/2.

You can use EVMs as reference.

Schematic looks good though

Best regards,

Vicente

0 Phil Lin2 5 months ago in reply to vicente flores prado

Prodigy 20 points

Dear Vincente,
Thank you for your review and suggestions. I attach updated schematic below.

However, there are 2 questions, which are marked PINK color at above schematic, still need your help.

Question 1:
May I know whether Y0 & Y1 need to follow EVM schematic such as below RED rectangles? One more, may I know the trace impedance is 50 ohm or not?

Question 2:
From page 8 at LMX2582 datasheet, it mentions below. May I know whether "not using RFoutB" needs to add 50 ohm and then link to GND each? Or, just the same as Noel mentioned "leave the pins floating"? Please suggest.

Thank you.

0 Michael Srinivasan 5 months ago in reply to Phil Lin2

TI__Genius 9110 points

Hi Phil,

Phil Lin2 said:
May I know whether Y0 & Y1 need to follow EVM schematic such as below RED rectangles? One more, may I know the trace impedance is 50 ohm or not?

I would recommend biasing the output and adding a decoupling capacitor as the EVM schematic does. And yes, you can assume the trace impedance is 50 ohms.

Phil Lin2 said:
From page 8 at LMX2582 datasheet, it mentions below. May I know whether "not using RFoutB" needs to add 50 ohm and then link to GND each? Or, just the same as Noel mentioned "leave the pins floating"? Please suggest.

The section in the datasheet you are referring to indicates what to do when you are driving a differential output and only want to use the P or the N side. If you are not using RFoutB, then you should leave the pins floating to reduce the power consumption.

Thanks,

Michael

0 Phil Lin2 5 months ago in reply to Michael Srinivasan

Prodigy 20 points

Dear Michael,
Thank you for feedbacks & suggestions. Below is the updated schematic

May I know
1. these" biasing the output and adding a decoupling capacitor as the EVM schematic" close LMK1C1102 or LMX2582?
2. decoupling can be changed from 5pF to 5.1pF because lab cannot find 5pF?

Thank you.

0 Michael Srinivasan 5 months ago in reply to Phil Lin2

TI__Genius 9110 points

Hi Phil,

Phil Lin2 said:
1. these" biasing the output and adding a decoupling capacitor as the EVM schematic" close LMK1C1102 or LMX2582?

This would be best placed close to the input to the LMX2582.

Phil Lin2 said:
2. decoupling can be changed from 5pF to 5.1pF because lab cannot find 5pF?

This is fine.

Thanks,

Michael

0 Phil Lin2 5 months ago in reply to Michael Srinivasan

Prodigy 20 points

Dear Michael,
Thank you for your help with confirmations. Thank you.

0 Phil Lin2 5 months ago in reply to Phil Lin2

Prodigy 20 points

Dear Michael,
Sorry to bother you.
For signal "VIH - High Level Input Voltage" min value, datasheet shows at least 1.4V. From our current schematic, which follows EVB, is 3.3V * (12K/(12K+12K)=1.65V, it is very close 1.4V whether it ensures the desired ON/OFF function. Please suggest. Thank you.

0 Phil Lin2 5 months ago in reply to Phil Lin2

Prodigy 20 points

Add discussing area graph.

0 Noel Fung 5 months ago in reply to Phil Lin2

TI__Guru** 108270 points

HI Phil,

Min. VIH is 1.4V, so a 1.65V input will be consider a logic HIGH.

Indeed, you don't need to copy EVM schematic, the 12kΩ resistors were added to make the digital control pins compatible with 5V logic.

If your MCU is 3.3V logic, simply series a 33Ω resistor is good enough.

0 Phil Lin2 5 months ago in reply to Noel Fung

Prodigy 20 points

Dear Noel,
Thank you for your feedback. About "simply series a 33Ω resistor is good enough," I have a little confuse, so I update below schematic. Please help to check whether ORANGE color area is correctable or not. Thank you.

0 Noel Fung 5 months ago in reply to Phil Lin2

TI__Guru** 108270 points

Hi Phil,

The 12kΩ shunt resistors are not needed.

For the CE pin, the pull-up resistor should be in kΩ, for example, 10kΩ.

0 Phil Lin2 5 months ago in reply to Noel Fung

Prodigy 20 points

Dear Noel,
Thank you for your suggestions and review.
The whole 12Kohms are "(Do not SMT Component)", so these will be not soldered.
About the CE pull-up resister is changed to 10Kohm.
Below is the updated schematic. The whole discussed parts are at ORANGE area. Please help to check whether there is any mistake.

Thank you.