How can electrical engineering be self-taught?

[russ_watters]

So essentially, the P.ENG. is useless to Electrical Engineers or more specifically, Computer Engineers.

A PE is basically useless for anyone not getting government or other permits for their work, yes. It is completely useless for a chip designer in Silicon valley.

Or from the other direction: you cannot become competent at making permit documents unless you work for someone who makes permit documents. A chip designer job won't prepare you for that so it is not a flaw in the PE requirements.

 

[analogdesign]

Yeh, those days, IC was relative simple, the excitements were in discrete designs. Now everything is connecting from one big IC to another big IC. Designs are transmission lines impedance matching of the traces from IC to IC or to external world. Everything exciting is inside the IC.

I worked in SONET related system, all we saw was the transceiver ICs with differential outputs. We just design the differential traces to interface the circuit to another IC! Yes, at 10GHz, pcb material makes a difference, every little kink, every little bent has to be considered. Pads has to be shaved off here and there etc. But it is just a glorified pcb layout! You have all the fun! I don't even think you can do a lot of 10+ GHz circuit design in discrete form as the $\lambda$ is just too short and loss is too high to run traces. It is a lot more suitable to be inside the IC where dimension of devices are much smaller.

IT's too bad now the degree is so much more important. As I said before, education does not make a person more creative or more intelligent. I just seen more morons with a degree...and they are still morons! I hired a student that is in the process of getting a MSEE in San Jose State...Let's just say...he got fired after a month.

Well, my days have passed, I had my fun in my days. It's been 8 years since I retired. It's a different world. When I was working in LeCroy in 82, we did a 200MHz 8bits ADC front end using subrange configuration. It was a big big deal. You are doing multi GHz ADC! We had hybrids that we put surface mount transistors inside the hybrids!

How interesting! I'm with on you the moron angle... in my experience education and capability are somewhat correlated, but they don't have a one-to-one correspondence at all!

PCB design at 10 GHz or even 40 GHz (!) is possible but very challenging and specialized! You're right that it is primarily connecting highly integrated ICs but getting multi-gig signals across a backplane is tough! There are some good new materials (like the various Rogers mixes) that really extend the bandwidth... you can't just use FR-4 these days! You have to keep the signals balanced and minimize changing layers whenever possible... also you typically have to back drill vias to get rid of the stubs. Very expensive! The SERDES I'm working on right now has an equalizer in it that can compensate somewhat for the board parasitics, but you're right, it's much easier in the IC (and even then not easy!)

The scope front-end designers are always on the cutting edge! That must have been exciting at LeCroy. I have a buddy who does ICs for scopes and they typically use a InP or GaAs S/H amplifier followed by a massively parallel CMOS ADC. Going parallel is the real trick. The 10 GHz ADC I did was eight 1.3 GHz ADCs in parallel. You get all kinds of issues due to timing, gain, and offset mismatches but you can fix those in DSP. Fun stuff!

 

[analogdesign]

A PE is basically useless for anyone not getting government or other permits for their work, yes. It is completely useless for a chip designer in Silicon valley.

Or from the other direction: you cannot become competent at making permit documents unless you work for someone who makes permit documents. A chip designer job won't prepare you for that so it is not a flaw in the PE requirements.

One other thing to keep in mind is that government regulations typically don't favor EEs. In California at least, a Civil PE or Mechanical PE can stamp electrical drawings as well as civil and mechanical drawings, but an EE PE can't sign off on civil or mechanical drawings!

My advice is if you want to work on civil-engineering type projects, be a Civil Engineer!

 

[yungman]

How interesting! I'm with on you the moron angle... in my experience education and capability are somewhat correlated, but they don't have a one-to-one correspondence at all!

PCB design at 10 GHz or even 40 GHz (!) is possible but very challenging and specialized! You're right that it is primarily connecting highly integrated ICs but getting multi-gig signals across a backplane is tough! There are some good new materials (like the various Rogers mixes) that really extend the bandwidth... you can't just use FR-4 these days! You have to keep the signals balanced and minimize changing layers whenever possible... also you typically have to back drill vias to get rid of the stubs. Very expensive! The SERDES I'm working on right now has an equalizer in it that can compensate somewhat for the board parasitics, but you're right, it's much easier in the IC (and even then not easy!)

The scope front-end designers are always on the cutting edge! That must have been exciting at LeCroy. I have a buddy who does ICs for scopes and they typically use a InP or GaAs S/H amplifier followed by a massively parallel CMOS ADC. Going parallel is the real trick. The 10 GHz ADC I did was eight 1.3 GHz ADCs in parallel. You get all kinds of issues due to timing, gain, and offset mismatches but you can fix those in DSP. Fun stuff!

Oh yeh, FR-4 is out of the question. I have been using Roger materials in quite a few occasions already. Changing layers on the trace is the tricky part. That's where shaving pad and all comes in.

At the time in LeCroy, we used the sub-range configuration. We first use a 4 bits ADC that had 8 bits accuracy to digitize the signal, then use a DAC to transform back to an analog signal with only 16 levels ( 4 bits) and sum together with a delay version of the original analog signal. This create the difference signal. Then we digitize the difference signal with the second 4 bit ADC. The first ADC create the M.S 4bits, the second ADC creates the L.S. 4 bits. We had to use RG58 coax lines to delay the analog signal for summing. Matching the amplitude and phase of the delayed signal is very tricky or else the two signal do not sum correctly and you can see the wave form breaking up. Everything was about timing, amplitude and phase compensation. Even big RG58 needed a lot of compensation to match the signal. We did not have any DSP or anything for correction everything was done in analog.

We were in the days even before Comlinear amplifier. We had to design our in house hybrid opamp and sample & hold to get the BW of 200MHz. Those were the fun days that I put in 18 hours a day for a period of time.

The time I left LeCroy, they were talking about "bucket brigade" stuff, which was the idea CCD that digitize the signal fast and spit it out slowly for a slower ADC to digitize the signal. I guess that did not happen. At the time, we even talked a little about driving the analog signal onto a transmission line, then digitize the signal at different taps with a number of ADC, so every time you digitize, you get a few data points closely spaced to increase the throughput.

Writing to memory was not easy either. There were no 100 MHz static RAM. All the gang switching and all.

It was hard to leave LeCroy, but I got the opportunity in Exar, so I had no choice.

 

[Dembadon]

A PE is basically useless for anyone not getting government or other permits for their work, yes. It is completely useless for a chip designer in Silicon valley.

Or from the other direction: you cannot become competent at making permit documents unless you work for someone who makes permit documents. A chip designer job won't prepare you for that so it is not a flaw in the PE requirements.

I know this is probably a rare case, but wouldn't there be a need for a PE if the chip were, say, to be part of a control system that runs an autopilot system or something similar? Or are these type of chips not developed by electronics companies in Silicon Valley?

 

[yungman]

I worked on military contracts in a company called Pacific Consultants that got bought out by Penstar. We were actually working on the Land Warrior system that connects the solders in the battle field together. The stuffs like you see in movies etc. They never ask anything about whether I had a PE, just citizenship.

This is the first time I ever heard about PE. Is this something for power transmission stuff?

 

[AlephZero]

I know this is probably a rare case, but wouldn't there be a need for a PE if the chip were, say, to be part of a control system that runs an autopilot system or something similar? Or are these type of chips not developed by electronics companies in Silicon Valley?

Probably not, unless the chip was designed and manufactured for that single application, and that's not very likely.

Certainly in the case of aircraft engine electronic control systems, the design responsibility and certification of the complete system is the responsibility of the engine manufacturer. That would certainly include a requirement to have a traceable source of parts with well-defined specifications, but most if not all those parts are not specific to an ECS, which conceptually is not much different from any other embedded computer system, and built from the same types of parts.

Maybe a better version of the question would be something like "what about the design and manufacture MIL-SPEC parts in general" - and I don't know the answer to that.

 

[analogdesign]

Maybe a better version of the question would be something like "what about the design and manufacture MIL-SPEC parts in general" - and I don't know the answer to that.

This is from the California Bureau of Consumer Affairs, Board for Professional Engineers, Land Surveyors, and Geologists website:

In California, electrical and mechanical engineering work performed by
employees of the Federal Government, and employees of manufacturing,
mining, public utility, research and development, or other industrial
corporations is exempt, and thus, such employees may serve as
references whether or not they are licensed as long as they are qualified
to appraise the technical competency of the applicant.
Typical examples of exempt employers would be aerospace companies
such as Boeing, or McDonnell Douglas; public utilities such as San Diego
Gas & Electric or P.G.&E.; manufacturers of electrical or mechanical
equipment such as Allen Bradley, Square D, Trane or Carrier, etc.
Examples of non-exempt employers are consulting electrical or
mechanical engineering firms and state and local governments.

So, I would say you do not need a PE license to develop MILSPEC components, or do virtually any type of defense work for that matter. A PE for Electrical Engineers is primarily relevant for EEs who do support work for Civil and Structural Engineers. For example, the engineers who develop control systems and electrical systems for buildings would need a PE to sign off on the design. These engineers typically work for consulting firms.

 

[Dembadon]

Probably not, unless the chip was designed and manufactured for that single application, and that's not very likely.

Certainly in the case of aircraft engine electronic control systems, the design responsibility and certification of the complete system is the responsibility of the engine manufacturer. That would certainly include a requirement to have a traceable source of parts with well-defined specifications, but most if not all those parts are not specific to an ECS, which conceptually is not much different from any other embedded computer system, and built from the same types of parts.

Maybe a better version of the question would be something like "what about the design and manufacture MIL-SPEC parts in general" - and I don't know the answer to that.

Makes sense; such systems are not purely electrical/electronic. Components of it might be, but then sole responsibility does not fall on a single person.

 

[MinnesotaState]

Self teaching? Circuit analysis. Absolutely worthless!

Ended up flipping back the pages to study the history. Why? Because learning what's already refined didn't do a dang thing for me. Studying the history satisfied those "how's" & "why's" I was full of.

My knowledge of EE is about as low as it gets, but hey at least I know what I'm working with now.

"The Battery" by Henry Schlesinger & BBC's "The Story Of Electricity"

^Just thought I'd mention those because it's where I started!

 

[dan12]

If you are not going to school and you want to get into EE, get a low level job. I don't know what's your knowledge level, try get a technician or even an assembler position and study at the same time.

I never have a degree in EE, I had an AA from Heald that I never even advertized as it is not very useful nor would I recommend. I studied everything on my own. I started as a field service technician for Norelco office dictator equipments. Job was just cleaning recorder heads, belts, motors etc. Then I studied the schematics, I studied textbooks at the same time. I even got fired because I rush through the job so I can stay in the car to study! Then I landed onto a production tech in a satellite transmitter company. Then slowly move into engineering tech. Then got promoted to be a designer engineer. I had been an engineer and manager of EE for 30 years since. Other than I stayed home and studied 18 hours a day for 3 months after I got fired from the first job, I never stop working. I gained broad knowledge by keep switching companies of completely different technologies after I mastered one technology. I went from designing data acquisition( oscilloscope type), to linear integrated circuits, to ultra sound imaging medical equipments, mass spectrometers, RF military equipments. All just to gain knowledge. Use the job, maximize return of your time!

Point is try to use the job as an experimental ground to learn. Once you get your foot in, you are surrounded by people in this field and you learn from them. Just the fact you are in the environment, it rub off on you. You'll be surprised how much designs you can do with so little knowledge to start with. I did studied along the way, but nothing to get a degree. After I retired, I spent 8 years studying everything required for BS and way beyond now. But that's after I retired. I don't know the job situation now a days, it worked for me. You have to be hungry, you have to work hard, put in lots of hours, be aggressive to get result.

I did do some experiment at home, you can buy breadboard like on ebay:http://www.ebay.com/itm/70pcs-Breadboard-Jumper-Cable-Wire-PCB-Protoboard-Test-Circuit-Board-400Points-/360667847168?pt=LH_DefaultDomain_0&hash=item53f97a9a00 to build circuits. Get some experimental books and build the circuits. I went to a electronics surplus store and bought a used oscilloscope and function generator for cheap and use it at home.

If you have more specific question, post back.

Hi Yungman,

First off, hats off to you for being so dedicated to self-educating yourself all of these years. I am currently a low level RF lab technician at a small company here in the Silicon Valley. I started out as an assembler and have done QA, and over the years done enough job shadowing and gained enough practical experience to be called a "low level tech". By "low level tech", I mean I know what to do practically through experience, but I don't know the theory behind it. I never had formal training in electronics and know very basic AC/DC theory from trying to teach myself the theory off and on over the years. I must admit, my dedication to learning electronics and passion for it is not at your level and electronics is very math heavy (something I was never good at and had very interest in while in school).

I was thinking of attending Heald College in Milpitas to gained the knowledge on theory, but tuition is crazy high and my company will not reimburse for the tuition cost. The field I'm in is very specialized, and a good RF/Microwave tech is very hard to come by from what I hear. I wouldn't call it my ideal career choice, but at my age (30 years old), and since I already have years of tuning/testing/troubleshooting experience under my belt, I want to dedicate my time from now on to self-teaching myself electronics. I have researched what books are out there. For AC/DC (I'm using Grob's Basic Electronics), Semiconductors (I am using Electronic Principles by Malvino), analog (using Tom Wheelers book on Electronic Communications for Technicians), digital (using Digital Fundamentals by Thomas Floyd). All these books on written on a technician level which I am comfortable with due to my weak math and science background. I will specialize in RF/Microwave and I see Microwave Engineering by Pozar is popular book, but the math in that book from taking a quick glance is way over my head.

I definitely need to improve my math and science skills, and working in this industry, I can see that it doesn't hurt to have skills in mechanical drafting/design either. I plan to take courses at a local community college next semester and work my way up to engineering level math and physics, but all of the electronics theory and lab experiments will all be self-taught through books at home.

My question to you are as follows:

1) From your experience through the years, would you hire someone who has no knowledge on theory, but can do the job even if its on a low level? I guess I would call myself a level 2 out of 5 to describe where I'm at. I primarily work on low noise and medium power amps tuning, testing, aligning, troubleshooting down to component level. I know calibration procedures for the rf equipments and also know some testing methods for checking harmonics, IP3, oscillation. I can take test data, know what to look for when looking at datasheet of a FET or MMIC. However, if you were to ask me any technical questions, I wouldn't have a clue.

2)What books do you recommend to learn RF/Microwave theory in depth since this will be my area of specialization.

3)Just wondering if attending a trade school better prepared you for the entry level jobs you had?

Thanks for your time,
Dan