Learn More
Hi, I’m Michael
Engineering Student from the Bay Area
If you want to learn more about me, feel free to explore this page!
Full Adder Chip Fabrication
This is a 2-bit full adder which is a digital circuit that can add binary numbers. We worked in groups and used L-edit to create the mask shown in the image to the right with four layers (diffusion, vias, oxide, and metal). Then we ordered the masked and was able to successfully fabricate the circuit onto a silicon wafer shown in the dark square on the mask.
We tested the I-V characteristics of the circuit concluded that the transistors were working perfectly but the full adder was not working as we expected.
The general procedure for fabrication are as follows:
cleave wafers, clean wafers, apply photoresist, mask align, develop, repeat from step 2 for each mask then do metal evaporation after mask 4.
During this process, we work with many chemicals and acids such piranha, HF, HMDS, and toluene.
If you want to know more about what we did please contact me below.
MFT TAP Correction
This was research on probabilistic computing done with my former research group at OPUS Lab.
Mean field theory is a method to solve large stochastic models with less computations. Traditional MFT is done by assuming an average for each p-bit in a system and then using the weight (J) matrix to iterate the system until there is very little change and we arrive at the average for each p-bit. This result can be used to train machine learning models.
This method of finding averages is efficient but can be very inaccurate when you start working with larger systems. So, there is a second degree correction called the “TAP correction.” This method is slower but much more accurate. To verify this, you can see from the first image that the average error is less with TAP correction than with traditional.
Now that I’ve verified that TAP is better, I can now try this method on the MNIST set. After many hours of training I finally arrived at some results. You can see from the second image that the training results for the TAP and Traditional MFT are about the same. Thus, I concluded that there was no advantage in using the TAP correction for machine learning models.
If you want to learn more about this, contact me below and I’d be happy to go into more detail.
Electrocardiogram
This was a project chosen for a lab. I created an Electrocardiogram (ECG) with a programmable gain on a microcontroller which can also output the BPM of the heart.
This circuit is created with an instrumentation amplifier which reduces the common mode noise and amplifies the heart signal. This passes into a low pass filter with 250Hz cutoff to remove the high frequency noise. This passes a bandpass filter from 0.5Hz-150Hz which is the range of the heart signal. Finally, we pass this through a notch filter at 60Hz to remove powerline interference. For all the filters we use the Sallen-Key circuits. The DAC and high pass filter was integrated in the microcontroller. The first image shows how the circuit is laid out.
After everything was set up, I put some electrodes across by heart and grounded my body. The oscilloscope showed the signal in the second image which is my heartbeat. This signal can be adjusted via the microcontroller for more or less gain.
Experience and Projects
Hobbies
Working Out
Web Design
Investing
Gaming
Contact Me
I love connecting with new people! Feel free to contact me via email or Linkedin.
About Me
Hi, my name is Michael Chen. I am currently a UC Santa Barbara student pursuing a degree in electrical engineering. My interest is in semiconductors. I will be graduating in 2025 and I’m currently looking for a job for next summer.
I recently started doing web design which I am having a lot of fun with because I get to share my creative side with others and they are able to benefit from it as well.
I am currently interning in Saint Louis at Emerson Electric as an electrical engineer.
“Anyone can follow directions. It's more difficult when you have to create them”
-Michael C.
(C) Copyright Michael Chen 2024. All Rights Reserved
Privacy Policy
Terms and Conditions
michaelchen1142003@gmail.com
619-786-0244
Learn More
Hi, I’m Michael
Engineering Student from the Bay Area
If you want to learn more about me, feel free to explore this page!
Hobbies
Working Out
Web Design
Investing
Gaming
Contact Me
I love connecting with new people! Feel free to contact me via email or Linkedin.
Full Adder Chip Fabrication
This is a 2-bit full adder which is a digital circuit that can add binary numbers. We worked in groups and used L-edit to create the mask shown in the image to the right with four layers (diffusion, vias, oxide, and metal). Then we ordered the masked and was able to successfully fabricate the circuit onto a silicon wafer shown in the dark square on the mask.
We tested the I-V characteristics of the circuit concluded that the transistors were working perfectly but the full adder was not working as we expected.
The general procedure for fabrication are as follows:
cleave wafers, clean wafers, apply photoresist, mask align, develop, repeat from step 2 for each mask then do metal evaporation after mask 4.
During this process, we work with many chemicals and acids such piranha, HF, HMDS, and toluene.
If you want to know more about what we did please contact me below.
Experience and Projects
About Me
Hi, my name is Michael Chen. I am currently a UC Santa Barbara student pursuing a degree in electrical engineering. My interest is in semiconductors. I will be graduating in 2025 and I’m currently looking for a job for next summer.
I recently started doing web design which I am having a lot of fun with because I get to share my creative side with others and they are able to benefit from it as well.
I am currently interning in Saint Louis at Emerson Electric as an electrical engineer.
“Anyone can follow directions. It's more difficult when you have to create them”
-Michael C.
(C) Copyright Michael Chen 2024. All Rights Reserved
michaelchen1142003@gmail.com
619-786-0244
MFT TAP Correction
This was research on probabilistic computing done with my former research group at OPUS Lab.
Mean field theory is a method to solve large stochastic models with less computations. Traditional MFT is done by assuming an average for each p-bit in a system and then using the weight (J) matrix to iterate the system until there is very little change and we arrive at the average for each p-bit. This result can be used to train machine learning models.
This method of finding averages is efficient but can be very inaccurate when you start working with larger systems. So, there is a second degree correction called the “TAP correction.” This method is slower but much more accurate. To verify this, you can see from the first image that the average error is less with TAP correction than with traditional.
Now that I’ve verified that TAP is better, I can now try this method on the MNIST set. After many hours of training I finally arrived at some results. You can see from the second image that the training results for the TAP and Traditional MFT are about the same. Thus, I concluded that there was no advantage in using the TAP correction for machine learning models.
If you want to learn more about this, contact me below and I’d be happy to go into more detail.
Electrocardiogram
This was a project chosen for a lab. I created an Electrocardiogram (ECG) with a programmable gain on a microcontroller which can also output the BPM of the heart.
This circuit is created with an instrumentation amplifier which reduces the common mode noise and amplifies the heart signal. This passes into a low pass filter with 250Hz cutoff to remove the high frequency noise. This passes a bandpass filter from 0.5Hz-150Hz which is the range of the heart signal. Finally, we pass this through a notch filter at 60Hz to remove powerline interference. For all the filters we use the Sallen-Key circuits. The DAC and high pass filter was integrated in the microcontroller. The first image shows how the circuit is laid out.
After everything was set up, I put some electrodes across by heart and grounded my body. The oscilloscope showed the signal in the second image which is my heartbeat. This signal can be adjusted via the microcontroller for more or less gain.
Learn More
michaelchen1142003@gmail.com
619-786-0244
Hi, I’m Michael
Engineering Student from the Bay Area
If you want to learn more about me, feel free to explore this page!
Hobbies
Working Out
Web Design
Investing
Gaming
Contact Me
I love connecting with new people! Feel free to contact me via email or Linkedin.
Experience and Projects
“Anyone can follow directions. It's more difficult when you have to create them”
-Michael C.
MFT TAP Correction
This was research on probabilistic computing done with my former research group at OPUS Lab.
Mean field theory is a method to solve large stochastic models with less computations. Traditional MFT is done by assuming an average for each p-bit in a system and then using the weight (J) matrix to iterate the system until there is very little change and we arrive at the average for each p-bit. This result can be used to train machine learning models.
This method of finding averages is efficient but can be very inaccurate when you start working with larger systems. So, there is a second degree correction called the “TAP correction.” This method is slower but much more accurate. To verify this, you can see from the first image that the average error is less with TAP correction than with traditional.
Now that I’ve verified that TAP is better, I can now try this method on the MNIST set. After many hours of training I finally arrived at some results. You can see from the second image that the training results for the TAP and Traditional MFT are about the same. Thus, I concluded that there was no advantage in using the TAP correction for machine learning models.
If you want to learn more about this, contact me below and I’d be happy to go into more detail.
Electrocardiogram
This was a project chosen for a lab. I created an Electrocardiogram (ECG) with a programmable gain on a microcontroller which can also output the BPM of the heart.
This circuit is created with an instrumentation amplifier which reduces the common mode noise and amplifies the heart signal. This passes into a low pass filter with 250Hz cutoff to remove the high frequency noise. This passes a bandpass filter from 0.5Hz-150Hz which is the range of the heart signal. Finally, we pass this through a notch filter at 60Hz to remove powerline interference. For all the filters we use the Sallen-Key circuits. The DAC and high pass filter was integrated in the microcontroller. The first image shows how the circuit is laid out.
After everything was set up, I put some electrodes across by heart and grounded my body. The oscilloscope showed the signal in the second image which is my heartbeat. This signal can be adjusted via the microcontroller for more or less gain.
Full Adder Chip Fabrication
This is a 2-bit full adder which is a digital circuit that can add binary numbers. We worked in groups and used L-edit to create the mask shown in the image to the right with four layers (diffusion, vias, oxide, and metal). Then we ordered the masked and was able to successfully fabricate the circuit onto a silicon wafer shown in the dark square on the mask.
We tested the I-V characteristics of the circuit concluded that the transistors were working perfectly but the full adder was not working as we expected.
The general procedure for fabrication are as follows:
cleave wafers, clean wafers, apply photoresist, mask align, develop, repeat from step 2 for each mask then do metal evaporation after mask 4.
During this process, we work with many chemicals and acids such piranha, HF, HMDS, and toluene.
If you want to know more about what we did please contact me below.
About Me
Hi, my name is Michael Chen. I am currently a UC Santa Barbara student pursuing a degree in electrical engineering. My interest is in semiconductors. I will be graduating in 2025 and I’m currently looking for a job for next summer.
I recently started doing web design which I am having a lot of fun with because I get to share my creative side with others and they are able to benefit from it as well.
I am currently interning in Saint Louis at Emerson Electric as an electrical engineer.