# Gilbert Cell

Discussion in 'General Electronics Discussion' started by GilbertCell, Jul 23, 2013.

1. ### GilbertCell

6
0
Jul 23, 2013
Hi I am trying to simulate the Gilbert Cell circuit shown but unsure how. what connection is f1 and what is f2?

File size:
43.2 KB
Views:
258
2. ### john monks

693
2
Mar 9, 2012
What is f1 and f2?

3. ### GilbertCell

6
0
Jul 23, 2013
The Gilbert cell is a frequency mixer so it should take two inputs the frequency (f1 and f2) to get the new frequency. So do I set it up so I have two frequency's going into RF and this will give me the sum and difference at LO and IF? I am trying to simulate to see the waveforms

4. ### duke37

5,364
772
Jan 9, 2011
The diagram shows which are the inputs and output.
LO is local oscillator.
IF output will contain LO-RF (or RF-LO) and LO+RF.

If the cell is well balanced, the output will not contain LO or RF.

5. ### john monks

693
2
Mar 9, 2012
Then it seems to me that f1 would go into RF Input and f2 would go into LO Input. Your new frequency would be your IF Output.

This corresponds to a radio converter where the RF is the received signal, the local oscillator or LO corresponds to your LO input and the IF corresponds to the intermediate frequency that is past by the IF strip.

Am I missing something?

6. ### GilbertCell

6
0
Jul 23, 2013
Ok so I am building this with NI, do I need additional components or can I just attach a function generator to the 2 inputs and an oscilloscope to the output? Also what kind of values do I use for the frequency or does it not matter?

Also is the current source just going to ground?

Last edited: Jul 23, 2013
7. ### duke37

5,364
772
Jan 9, 2011
Who is NI? Can he help?

The current source goes to ground.

The RF input will need a positive bias (2V?) and a balanced signal.
The LO will need to be 2V more positive and balanced.
The output needs to be at higher voltage still and will have a balanced output.

The frequencies do not matter if they are within the capability of the transistors. If you are running a simulation then low frequencies would be best.

8. ### GilbertCell

6
0
Jul 23, 2013
NI meaning the software national instruments. Do you know if I connect RF to a voltage source with positive and negative going to the two separate ends or both ends to positive and the negative to ground? Yea I am lost, do you know a working Gilbert cell schematic as I can't find any with values on them.

9. ### duke37

5,364
772
Jan 9, 2011
The only component that you want for a simulation is the current source. You can use a resistor for this giving about 1mA.

The LO signal should be about 100mV +/- from the mean in order to turn the transistors fully on and off. This is often done with a centre tapped transformer with the centre connected to the positive bias.

The RF signal must be much smaller if linearity is required and is connected similarly to the LO.

The output is also taken using a centre tapped transformer winding.

The cell can be used without proper balancing but there is not much point in degrading the performance.

10. ### GilbertCell

6
0
Jul 23, 2013
Hi have I done something wrong as the output waveform does not looks right.

File size:
153.3 KB
Views:
240
11. ### duke37

5,364
772
Jan 9, 2011
You have no bias on the inputs.
You have the inputs shorted so there is no signal in.
You have no power supply for the output.
You have shorted the output signal.

I am surprised that you got any output signal.

I attach a sketch on how the cell should be fed and how the signal should be extracted.

File size:
4.3 KB
Views:
239
12. ### GilbertCell

6
0
Jul 23, 2013
Hi does this look right? How do I get it so that the waveform to change to something like that of the attached image? Also what is a perfect transistor and how do I simulate a perfect transistor

#### Attached Files:

File size:
130.5 KB
Views:
275
• ###### waveform.jpg
File size:
20.4 KB
Views:
176
Last edited: Jul 24, 2013
13. ### duke37

5,364
772
Jan 9, 2011
The bias to Q1 to Q4 is suspect. The base voltage is not well defined.

You have used unbalanced inputs and the output looks like a simple switching mixer with no attenuation of the input signals. I think one of the inputs should be much more than 10mV.
I have not looked at the waveform of a Gilbert cell.

The waveform on the right is just an addition of the two input signals. No mixing has been done. You could generate this with two resistors !