Lab 9 Design of Beta-Multiplier Reference (BMR)- EE 420L

Authored by Jeremy Garrod

4/19/2017   

garrod@unlv.nevada.edu

Pre-Lab work

Pre-lab work


The adjusted transistor parameters from laboratory 8 will be used in the BMR design in this lab. Below is a summary of the transistor parameters and the hand calculations.





Below is the schematic of the BMR as well as the simulation results showing the current through both sides, vbiasp and vbiasn while VDD varies from 0-10V in 1V steps. The simulations match closely to my hand calculations. Vbiasn stays constant around 700mV and Vbiasp is always about 2.5V below VDD. The current through both sides is almost identical as well.







Lab Work


Design

Sadly, the wrong transistor array was used in laboratory 8 as well as the prelab. In order to do this lab, the correct transistor array needed to be characterized. The SPICE parameters can be seen below, I have skipped writing all of the calculations, as that has already been done that in the previous lab.


Another issue that was encountered was when the vbiasn was being measured, it would raised linearly until about 4.5V and then it jump and plateau. The vbiasn would then stay at this voltage until VDD was lowered to around 2V, which is what was expected. The only remedy that seemed to work was lowering the resistor value of the startup circuit. This in turn added more current to the bottom NMOS devices than I would have liked. This effect can be seen and will be further discussed below.







When the VDD is at 10volts, it adds approximately 1uA of current into the circuit, which causes the NMOS to have nearly double the current as the PMOS, which can be seen in the simlation results above. Also, the current in the right side of the BMR seems to be about 1/3 the current in the left side. I had initially assumed that this was due to the extra current from the start up circuit. With a 300MEG resistor, the gap is not as large, but there is still a sizeable gap there.  I am not sure what could cause this, my only guess is that the SPICE parameters are incorrect.

Experimental Results

In the table below, the current through M2 is more precise than the current through M1. This is due to the fact that M1 was measured directly with an ammeter that couldn't measure lower than a tenth of a microamp. The current through m2 was calculated by taking the voltage drop across the resistor and dividing it by the resistance.

VDDVbiasnVbiaspID M1 (uA)ID M2 (uA)
10.4930.017600
21.015.5360.91.22
31.1511.5451.41.36
41.5442.5691.71.44
51.5563.4631.91.54
61.5614.4372.11.60
71.5695.3982.31.68
81.5746.4212.91.75
91.5977.5333.21.83
101.6098.4293.31.89



It can be seen that the right side of the BMR has the exact behavior that you would expect to see. However, the left side is very distorted, which is more than likely due to the extra current from the start up circuit. Neither of the measured current are really close to the simulated values. The left side is roughly 1.5x the simulated current and the right side is about 2.5x the simulated current.


Design



 
The current through the NMOS current mirror is the same as the current in M1, which is expected.

The current in M5 is close to the current that flows in M4. It is close enough to where I'd say that the current mirror is working as intended. 


Experimental Results

VDDID NMOS Mirror (uA)ID PMOS Mirror (uA)
100
20.92.5
31.13
41.53.6
51.84.1
624.2
72.34.9
82.75.3
92.95.9
103.46.5

The currents that were measured in the lab were pretty far off from the simulations. The NMOS current mirror has roughly double the current while the PMOS current mirror has over 5x the simulated current. even though the values were off, the shapes of the graphs are very similar to those from the simulations. 

Design

The NMOS cascode is working as intended in the simulations, it is acting like a constant current source with very little variation. 

It looks like the PMOS cascode is working, but I am not too sure.

Experiment

I was not able to get my PMOS cascode to work in the lab. I do not know if there was an issue with the actual circuit or my testing methodology.

VDDNMOS Cascode (uA)
01
16
26
36
46.1
56.2
66.3
76.3
86.3
96.4
106.5



This experiment ended up being similar to the otherexperiments, the shape is very close to what was simulated, but the actual values are pretty far off.

While the shapes of my graphs seemed to follow those from the simulations, I could not get my measured values to match my simulations. I am assuming that my KP values that I am using are off from the real values. I am also putting a relatively large amount of current into the current mirror from the start up circuit, when it should be a negligible amount. I am positive that this has an adverse affect on the circuit, however, the smaller resistor was necessary in order to get the correct behavior.


Return to EE 420 Labs