Author Topic: gilbert cell mixer implementation  (Read 2899 times)

0 Members and 1 Guest are viewing this topic.

Offline promachTopic starter

  • Frequent Contributor
  • **
  • Posts: 878
  • Country: us
gilbert cell mixer implementation
« on: March 26, 2019, 09:32:47 am »
In this paper 15.1 mW 60 GHz up-conversion mixer with 4.5 dB gain and 57.5 dB LO-RF isolation , I have three questions :

1) Have anyone designed a balun in spice software ? Is it doable ?

2) How to implement TL1 and TL2 (inductor implemented using transmission line) in spice software ?

3) How to derive the conversion gain expressions ((1) and (2)) for the up-conversion mixer circuit with and without negative resistance compensation in the paper ?

 

Offline Paul Rose

  • Regular Contributor
  • *
  • Posts: 139
  • Country: us
Re: gilbert cell mixer implementation
« Reply #1 on: March 27, 2019, 12:58:21 am »
Question 1) only:

I've never done this for 60 GHz, but you can make a balun in SPICE.

For a 3 port balun, use three inductors.

Set coupling between inductors with K statement:

  K1 L1 L2 L3 1

The 1 is the coupling.  1 is perfect, .9 is 90%, etc.

This one K statement should be the same as the the following three statements ( in case your software only allows setting coupling between two inductors at a time ).
  K1 L1 L2 1
  K2 L2 L3 1
  K3 L1 L3 1


Use L1 as your unbalanced input ( ground one leg ).

Observe phasing of L2 and L3.  Connect "dotted" end of L2 to "plain" end of L3, and connect this common point to ground.  Your balanced output  is the "plain" end of L2 and the "dotted" end of L3.

In LTSpice, for example, your inductor won't have a visible phasing dot until you add a K statement.


Set the inductance to match the characteristics of the balun you are simulating. 
 

Offline promachTopic starter

  • Frequent Contributor
  • **
  • Posts: 878
  • Country: us
Re: gilbert cell mixer implementation
« Reply #2 on: March 27, 2019, 03:57:50 am »
Quote
For a 3 port balun, use three inductors.

Set coupling between inductors with K statement:

  K1 L1 L2 L3 1

The 1 is the coupling.  1 is perfect, .9 is 90%, etc.

Could you use a picture to explain your balun schematics structure ?


By the way , see figure 3 of A 60‐GHz double‐balanced mixer with negative resistance compensation for direct up‐conversion using 90‐nm CMOS technology

 

Offline Paul Rose

  • Regular Contributor
  • *
  • Posts: 139
  • Country: us
Re: gilbert cell mixer implementation
« Reply #3 on: March 30, 2019, 10:29:09 pm »
Pic attached.

Did this in a paint app, based on google pics of LTSpice.  Don't have schematic editor on this linux laptop.

Inserting the K statement makes the phasing dots show up ( at least in LTSpice ).

Connect three inductors as shown.  Pick ratio of inductance to model turns ratio of your balun ( impedance proportional to turns squared).   Also pick inductance to set reactance according to your frequency.  You want your inductive reactance to be a lot higher than the load impedance ( like 4 times or more ). 

I've never worked with anything higher than 500 MHz, but it does show the concept of a balun in spice.

You don't have to use ground as the reference point.  Whatever you connect to the center tap, the signal will be balanced with respect to that point.
« Last Edit: March 30, 2019, 10:31:43 pm by Paul Rose »
 

Offline promachTopic starter

  • Frequent Contributor
  • **
  • Posts: 878
  • Country: us
Re: gilbert cell mixer implementation
« Reply #4 on: April 01, 2019, 07:58:27 am »
Thanks for balun picture schematics

But how do I interface your balun with mixer using differential current RF+ and RF- input signals ?



By the way, someone else told me to use the following :

but the load impedance for LO+ and LO- is almost infinite because it is mosfet gate ?





« Last Edit: April 02, 2019, 02:59:56 am by promach »
 

Offline Paul Rose

  • Regular Contributor
  • *
  • Posts: 139
  • Country: us
Re: gilbert cell mixer implementation
« Reply #5 on: April 02, 2019, 11:59:46 pm »
OK.

I haven't read the paper you linked, so I can't claim that this will be useful to you.

But simulating a current balun instead of a voltage balun is possible too.

This time I used QUCS, since I have it installed on my linux box.

I used their built in transformer component and turned it "sideways".

I fed in unbalanced voltage and loaded the outputs with two different sized resistors to ground.  I inserted a current probe in line with each resistor.

I read equal currents through each resistor ( and thus different voltages ).

Instead of using a transformer component, you could build the same thing with two coupled inductors.


 

Offline Paul Rose

  • Regular Contributor
  • *
  • Posts: 139
  • Country: us
Re: gilbert cell mixer implementation
« Reply #6 on: April 03, 2019, 12:21:34 am »
Of course that one provides no DC isolation.

But that is easy to fix.

« Last Edit: April 03, 2019, 12:23:30 am by Paul Rose »
 

Offline promachTopic starter

  • Frequent Contributor
  • **
  • Posts: 878
  • Country: us
Re: gilbert cell mixer implementation
« Reply #7 on: April 03, 2019, 02:15:50 am »
@Paul

I have no electrical knowledge on transformer and how it achieves the purpose of single-ended to differential conversion mathematically.

And why are you not using LC balun which I had highlighted in my previous post. Are there limitations for LC balun ?
 

Offline BrianHG

  • Super Contributor
  • ***
  • Posts: 8275
  • Country: ca
    • LinkedIn
Re: gilbert cell mixer implementation
« Reply #8 on: April 03, 2019, 03:07:06 am »
@promach, I know you are dealing with an on die 60GHz circuit, but, perhaps, you would get a few more responses if you moved your topic the the 'RF, Microwave, Ham Radio' section on this forum as RF mixing transformers is kind of their thing...
 
The following users thanked this post: TheUnnamedNewbie


Share me

Digg  Facebook  SlashDot  Delicious  Technorati  Twitter  Google  Yahoo
Smf