Category: Electrical Engineering

• Life update

  Oh boi, it’s been a year since I updated this site.

  There are lot of things I need to share:

  1. I just graduated from UW with electrical engineering degree. My concentration was in Analog Circuit Design.
  2. I built so many cool things over the year, it helped me to think like a real engineer (thanks to Professor Darling and Professor Rudell)
  3. got a job working on 5G stuff

  Anyways, 5G eh? Long story short, my capstone project that I did remotely due to Covid19, gave me a little insights about 5G. Our team built a homodyne (direct conversion) receiver for 2.5GHz 5G application. And 2.5GHz is actually band41 that T-Mobile got from their recent merger with Sprint (guess who’s working on it now ;))

  High level explanation, the receiver itself consist of 3 main blocks:

  1. Low noise amplifier (LNA)
  2. Voltage-controlled oscillator (VCO) utilized MOS varactors
  3. I-Q mixers

  I personally designed the VCO block. I used an LC tank to set the frequency, some buffer, and frequency divider using D-Flip Flops to produce a 90 degree phase shift that are needed by the I-Q mixers.

  

  

  

  I built these on Cadence Virtuoso Circuit Design Environment, employed with a 65nm CMOS process. However, due to time constraints, we did not build the whole layout for this receiver, someday I will come back to this and finish it.

  

  Complete Receiver Front-End Schematic

  For the future post(s), I will share the detail on how to build this (the math behind it etc), instrumental amplifier (IA) applications, and I will talk about 5G.

  Can’t wait to blog more.

  Cheers,

  Olce

  September 8, 2020

  ---

• Eye Diagram Using LFSR

  Heya,

  Unfortunately I am back to school again, but I only have one hard class (designing op amp), and 2 garbage classes, so I probably will continue working on this.

  Last time, I managed to transmit signals over 1000 foot UTP cable. However, to check whether it was garbage signals or not we need to generate eye-diagram to analyze our signals. I had no luck creating appropriate eye-diagram due to the code I uploaded to the Arduino. I needed something that have separate clock and something random.

  Talking about random, I thought about linear–feedback shift register (LFSR). LFSR is a shift register and with a feedback of two or more “taps”. Based on the circuit schematic shown in Picture 1, I tapped bit 0 and 2. You can of course tap more than two, however, there are certain of pattern you need to follow (refer to Picture 2) that can be referred to the length of the loop.

  

  

  Moreover, LFSR would spit out pseudo-random numbers that I need for my eye-diagram testing. The previous set up with Arduino was bad because it spit out constant pattern that of course you can’t analyze whether your signal is bad or not.

  So, I built LFSR with d-flip flop (using TI SN74LS273N), and XOR gate ( HD74LS86 )

  

  Using the circuit set up from Picture 1 (a), I use function generator as a clock for the D-flip flop, and output it from pin #6 of the d-flip flop (since I’m only using 2 taps).

  anddddd there she is, the not so pretty eye-diagram (lol). But it’s a progress!

  

  

  

  October 6, 2019

  Circuit Analysis, Electrical Engineering, LFSR, RS485, Signal Integrity, XOR Gate

  ---

• Fixed schematic and Circuit Board

  

  

  The difference between the new one and the previous schematic would be I added 4 120 ohm resistors (R1, R2, R3, R4) from the outgoing side (PIN#9 and #10) of transceiver and incoming side (PIN#12 and #11) of the transceiver.

  So far it looked like I managed to transmit 1Mbps over 1000 foot UTP cables, however at this point I can’t confirm whether the signal I transmitted was smooth or mostly jitters. I still need to figure out how to set up the eye-diagram with external trigger. Will keep you posted!.

  September 14, 2019

  Electrical Engineering, RS485, Signal Integrity, Transceiver

  ---

• Why RS-485?

  Hello people!

  It’s been a while since my last post. I actually did a little summer activities (hiking, gardening, rearranging pantry, making soap etc). BUT I also managed to design the circuit schematic for my serial communication project.

  The schematic as shown in Figure 1 is very drafty. I was using Altium Designer for it, but since I am not making PCB at this moment, I thought this should be doable as my breadboarding reference. Again, this schematic is awful because you can’t really see the flow from the original to the isolation section. I will fix it later!

  

  Breadboarding part I: Look at those loops!

  

  Then I realized that those loops could introduce unwanted inductance, so I revised.

  

  As you can see, they look so much better! FLAT. You’d want them flat like that. It’s a tedious work, but I shall get used to it.

  For multigiga serial links (with increasing bitrates), it would suffer from signal distortions. While SerDes receiver designed to compensate those distortions, by creating a bigger eye opening (see: eye diagram) for reliable sampling, the bit rates also rising faster than Rx, PCB, and package HDI. This results in decreasing reachable total channel. RS-485 is a type of repeater that could restore the signals, hence RS-485 extending the channel reach.

  

  RS-485 use differential signaling over twisted pair, which is beneficial. We know from circuit theory, that differential signaling technique reduces electromagnetic interference (EMI) that could cause cross-talk. In addition, RS-485 also use 3-state logic (1, 0 and high Z) that allows individual transmitter to be deactivated.

  The schematic and circuit board above (see Figures 1,2,3), uses RS-485 as a repeater and it connected it to digital isolator to ensure the power-supply voltage to the MCU is compatible and DC-DC converter to rectify the whole process back to the DC.

  If you notice, the circuit diagram for DC-DC converter contains Inductor(L) and Capacitor (C). The LC circuit is a simple low pass filter to filter out unwanted signals that may present above the wanted pass band.

  So in theory, once we connect all of these out to the UTP cable, 50m cable should not signal faster than 2Mbit/s. We’ll see!

  Will keep you updated.

  Source: https://www.intel.com/content/www/us/en/io/serial-bus-white-paper.html

  August 25, 2019

  Electrical Engineering, repeater, RS485, Signal Integrity

  ---

• Eye Diagram on Oscilloscope

  Heya people!

  It’s been a quiet week, but I managed to review and ordered few things that I needed to get for my on going project. By next week I should be getting my DC-DC converter, Digital Isolator, RS-485 transceiver, and some passive components. I might post the detail on why or how to pick these components to make it compatible to your microcontroller (MCU).

  Further, why did I mention eye diagram on the topic? What is eye diagram?

  Since Signal Integrity is the main topic of my project, an eye-diagram, which is an oscilloscope display (literally it look like an eye), could tell us the quality of serial bit stream and it can be used to debug signal integrity issues. According to Sciencedirect, amplitude noise and timing jitter could impact our system’s performance. However, using simple Bit Error Rate (BER) measurement alone can’t tell the difference between amplitude noise and timing jitter. Eye diagram, on the other hand, provides more information regarding both issues.

  

  We can take high-bandwidth instantaneous samples of high speed digital signal and the sum of the samples (superimposing ‘0’ and ‘1’) are the eye diagram! To make it more clear, just remember those timing diagram you did in your System Verilog class. The eye diagram would look EXACTLY like one of those state diagrams!

  I was playing with my oscilloscope earlier to display the eye diagram ( I own Tektronix TBS1102B), and unlike the advanced version, I had to manually set up the closest eye diagram I can get from it (see short video below) and I was using high baud rate (2Mbps) to demonstrate the best display of eye diagram.

  void setup() {
    Serial.begin(2000000);
    pinMode(4, INPUT);
    digitalWrite(4, LOW);
   }
  
  void loop() {
      Serial.write(0xAA);
  }

  From this short clip, we can see clearly the transition state from High to Low and so on.

  If you are using the same type of Oscilloscope, this can be accomplished by running code above (connect it with your Arduino, of course), Autoset it, then pick sample (by adjusting the horizontal knob) to the non-static rising edge, then select utility > display > enable persist (to as long as you want).

  ps: I found this artwork (or maybe enlarged screenshot) on Instagram and I want one for my office! lol

  https://www.instagram.com/p/BSE7K2eg8Yu/?utm_source=ig_web_button_share_sheet

  August 16, 2019

  Electrical Engineering, eye diagram, jitter, noise, Signal Integrity, timing jitter

  ---

• Test of Arduino’s USART

  Heya folks,

  Happy weekend to all of you! We had heavy rain on Saturday morning, and it was nice and good for my plants.

  We did a simple linear test with Arduino by using the Tx pin (USART) and to test the highest baud rate it can handle before it introduce some weird stuff on oscilloscope.

  //Again, pin#4 being LOW means for HIGH Z
  //the baud rate in this code is the highest rate that we tested and it had some nasty wave on oscilloscope
  void setup() {
    Serial.begin(2500000);
    pinMode(4, INPUT);
    digitalWrite(4, LOW); 
  }
  void loop() {
     Serial.write(0xAA);
  }

  Baud Rate(Bps)	Period ( μs)	Frequency(Hz)
  9600	~104.2	9596
  19200	~52.07	19204
  38400	~26.04	38402
  76800	~13.02	76804
  153600	~6.512	153562
  230400	~4.509	221778
  460800	~2.004	499001
  921600	~1.002	998003
  1843200	~0.5012	1995000
  2000000	~0.5011	1995609
  2500000	~0.5014	1995609

  Based on data above, we noticed that although it maintain the linear behavior, the higher the baud rate (the corresponds frequency value should be very close to the baud rate), the higher the deviation of the frequency rate.

  Check out the screenshots from two different baud rate:

  

  

  Just by eyeballing those two screenshots, we know that at period 250nanoseconds, the square wave behave abnormally (it has some negative peak, which we don’t want).

  We could say that it is feasible to transmit 1Mbs, BUT definitely less than 2.5Mbs!

  Tomorrow we will use two Arduino to test both Tx and Rx modules. That’s all for now, enjoy the rest of your weekend.

  Also, check out Thom Yorke’s new album:

  Cheers,

  August 11, 2019

  Arduino, Digital, Electrical Engineering, Oscilloscope, Signals, Testing, Transmitting

  ---

• Clock Slips

  Heya folks,

  Few days ago I received my some of the tools for my electronic projects (such as: oscilloscope, Arduino and Multimeter). We (William and I), configured an Arduino to test the oscilloscope by creating a simple square waves. Based on today’s experiment, we accomplished/discovered two interesting things:

  1. Near perfect square wave with no phase shift (117kHz frequency).

  We tested to see the frequency for 50% (no scale) by using analogWrite() function (reference).

  For this configuration, we noticed no phase shift on the analog side of the equation. However, after we tested with digitalWrite() (see reference), we noticed 1/4 of phase shift on our square wave (will update with image later).

  To solve this clock slips/phase shift problem, we did a little research and concluded that one of the reasons of the shift problem is due to interruption on the main (Arduino) code. Hence, we modified the code by adding noInterrupts() (see: reference).

  //The commented code were used to test elimination of phase 
  // shift before using the noInterrupts() function
  //unsigned int usdelay =1; 
  void setup() {
    noInterrupts();
    pinMode(3,OUTPUT);
    pinMode(4,OUTPUT);
    digitalWrite(3, LOW);
    digitalWrite(4, LOW);
  }
  
  void loop() {
    // put your main code here, to run repeatedly:
    digitalWrite(3, HIGH);
    digitalWrite(4, HIGH);
    //delayMicroseconds(usdelay);  
    digitalWrite(3, LOW);
    digitalWrite(4, LOW);
    //delayMicroseconds(usdelay);
  }
  

  

  For amplitude 5V, we managed to produce desired frequency: 117kHz (1 pin as INPUT) or 60.9kHz (for two pins as INPUT) without phase shift with the noInterrupts() solution.

  2. Cross-talk problem using serial connection

  We tested the serial connection with 9600bps and using 0xAA (Binary: 1010 1010 <– you can also imagine this as LOW, HIGH, LOW, HIGH,…, HIGH) pattern, and somehow our configuration introduced a cross-talk problem from the pin#4 we set-up as input (!!!).

  void setup() {
    Serial.begin(9600);
    pinMode(4, INPUT);
    digitalWrite(4, LOW);// pin#4 LOW means High Z
    }
  
  void loop() {
    Serial.write(0xAA);
  }

  

  As we can see from Figure 2, channel 2 (blue) is the square wave from pin#4 and channel 1 (yellow) is the square wave from the serial connection (tx) from the Arduino. We noted that the amplitude of the pin#4 is very low (148mV) compared to the amplitude of channel 1 (5.12V).

  That’s all for today. Will test 2 Arduinos later.

  Peace, Love, Mango

  August 7, 2019

  Arduino, electrical engineer, Oscilloscope, Signal Integrity, square wave

  ---