What advice would you give?

[Indico]

Hi all. What would have been the best advice you wished you were given when starting analogue IC design? Also what advice would you give to someone that is starting in this field and/or wanting to get into this field?

Thanks in advance.

 

[vivekroy]

My advice:

• Perseverance. Do not expect that you will be given a ton of independence on day 0. People need to trust you before they will give you responsibility. One mistake on your part can cost a company millions.
• Read, read and read. There is no substitute for reading papers. Recreate the results in the papers if you want to get your hands dirty.
• Answer questions on Edaboard. Will keep your basics sharp. Don't hesitate to call out someone who may have an experience level of 100 if you think they are mistaken. Of course, be polite. I have learned a lot, and also I have learned that even the best may not know a thing or two.

Something that I wish I was told:

• Health. Don't ignore it. Tapeouts will come and go, there will always be more things to do, more verification to be done, more papers to be read. But that is not an excuse for missing out on an hour of exercise every day. IC design is a long-haul career. You want to be healthy.

 

[timof]

What's your situation, your starting point?
E.g., do you have a BS or MS or PhD degree in EE or in IC design, or any other degree, are you already employed, etc.?

 

[deep_sea]

I will combine all as advices because it does not matter if I personally have done them or not

• Writing notes is a must. You can revisit the topic in half an hour instead of wasting one working day to study it again. Of course notes with citing references. You might find your notes are not enough and want to read the source again.
• Deep knowledge is essential. If you understand something 60%, it does not help.
• Get your hands dirty like already told by vivekroy.
• Trade the line very carefully between specialisation and general knowledge. It is practically impossible to know ADC/DAC, power management IC, RFIC, special purpose circuits, filters and IO. You have to focus but also know the fundamentals of most (all) of them.
• Ask colleagues and experts how things are really done. There is a "hidden" part in analogue IC that is not written in books. In particular biasing, PVT and ESD.
• Look at the designs of experienced colleagues in details.
  

 

[Indico]

My advice:

• Perseverance. Do not expect that you will be given a ton of independence on day 0. People need to trust you before they will give you responsibility. One mistake on your part can cost a company millions.
• Read, read and read. There is no substitute for reading papers. Recreate the results in the papers if you want to get your hands dirty.
• Answer questions on Edaboard. Will keep your basics sharp. Don't hesitate to call out someone who may have an experience level of 100 if you think they are mistaken. Of course, be polite. I have learned a lot, and also I have learned that even the best may not know a thing or two.

Something that I wish I was told:

• Health. Don't ignore it. Tapeouts will come and go, there will always be more things to do, more verification to be done, more papers to be read. But that is not an excuse for missing out on an hour of exercise every day. IC design is a long-haul career. You want to be healthy.

Thanks for your message. I understand that and actually it is something that I think it is valid in regards to the responsibility.

In regards to reading it is something that I need to get again into the habit. Used to read a lot but not much these days due to personal life.

I will try to participate in the forum and learn from everyone.

Yes, health it is important. I wonder how many hours people work in this field in general? I have heard some people saying that you can work very long hours, all the time. Some others only during peak close to tape-outs. Another ones saying that it is like they are always during peak times. That's is very unsettling. I guess it depends which market you work for but how do people manage being always in the peak?

 

[Indico]

What's your situation, your starting point?
E.g., do you have a BS or MS or PhD degree in EE or in IC design, or any other degree, are you already employed, etc.?

Hi timof,

I have a MSc in IC design, but have been working in a parallel field but now going to start in a company as IC design engineer. I have been learning from more experienced over the last couple of months but it is not been easy since people are not open and don't have patience.

 

[Indico]

I will combine all as advices because it does not matter if I personally have done them or not

• Writing notes is a must. You can revisit the topic in half an hour instead of wasting one working day to study it again. Of course notes with citing references. You might find your notes are not enough and want to read the source again.

Makes sense and it is was I always try to do although it is not easy since we are on the go. But need to improve on that one.

• Deep knowledge is essential. If you understand something 60%, it does not help.

How do you gain this deep knowledge?

• Get your hands dirty like already told by vivekroy.

Yes, definitely.

• Trade the line very carefully between specialisation and general knowledge. It is practically impossible to know ADC/DAC, power management IC, RFIC, special purpose circuits, filters and IO. You have to focus but also know the fundamentals of most (all) of them.

Can you elaborate a bit more on this one? How do you do this?

• Ask colleagues and experts how things are really done. There is a "hidden" part in analogue IC that is not written in books. In particular biasing, PVT and ESD.

Yes, definitely. I find ESD to be something that I don't have much knowledge but there isn't much information available. What you mean by biasing? Like current generators, bandgap generators?

• Look at the designs of experienced colleagues in details.

Yes, I try to do that but for example sometimes I ask this: I don't understand why this is done like this. I try to understand but then it comes down to system level requirement. So it can be specific or done in a way to overcome some issue. But it is always good to understand in the context to expand your knowledge.
Another thing I ask my self is, how the colleague did this? Why this way? What is the starting point? Could've done in another way? What makes you go to one or other option? Would I be able to do this? But I don't have much knowledge so maybe only experienced people, but then how will I become experienced? Till reaching a point to for example design a complete bandgap generator

Thanks for you inputs. I have commented and asked another question in your quote.

 

[timof]

Hi timof,

I have a MSc in IC design, but have been working in a parallel field but now going to start in a company as IC design engineer. I have been learning from more experienced over the last couple of months but it is not been easy since people are not open and don't have patience.

Indeed, that's a big problem.

You can learn the basics form the books, or from university courses and labs, but the real knowledge and intuition comes from the real experience in the industry.
And such things are not explained and described in the books.

Finding a good company, or group, or mentor, who would be open and patient, is not easy.

A few examples, of what happens in real design experience, and where the time is spent on, are explained in this blog:

https://streetsmartanalog.blogspot.com/

 

[deep_sea]

Makes sense and it is was I always try to do although it is not easy since we are on the go. But need to improve on that one.

Taking notes means also when you study from a book or a lecture. It is compulsory not optional.

How do you gain this deep knowledge?

You have to know why things are done. You have to understand specs and small signal models. The reasons behind choosing size and selecting topologies. Why a circuit is made this way rather than a simpler way.
Never plug circuits in simulator and sweep variables to get results. You will never be a good designer that way. Even if you get results by chance, you have to analyse and understand them.

Yes, definitely.

Can you elaborate a bit more on this one? How do you do this?

Set realistic goals. Do not say I am going to design 16b ADC while you are in early career. Try to master first the fundamental circuits. Current mirrors and opamps are essential and a good point to start with. Take your time and do not jump to other topics before finishing.
In bandgap, LDO, ADCs current mirrors and opamps are used. Do not try to master everything together because you will end up with nothing.
on the other hand, try to know a little bit about all topic but not deeply until you start working with them (either for personal development or for professional work).

Yes, definitely. I find ESD to be something that I don't have much knowledge but there isn't much information available. What you mean by biasing? Like current generators, bandgap generators?

Circuits are biased by current mirrors. In textbooks you see a small signal coming to an input. This input has a DC level which plays a role in common mode requirements. The current coming from current mirror decides your gm depending on the size. It does not work like (VGS-Vth) increase or decrease rather the gate voltage is decided by the size of device not the opposite.
Startup circuits are overlooked.
The square law is not accurate even to start with before simulation. There gm/ID (V*) methodology as an alternative.

Yes, I try to do that but for example sometimes I ask this: I don't understand why this is done like this. I try to understand but then it comes down to system level requirement. So it can be specific or done in a way to overcome some issue. But it is always good to understand in the context to expand your knowledge.
Another thing I ask my self is, how the colleague did this? Why this way? What is the starting point? Could've done in another way? What makes you go to one or other option? Would I be able to do this? But I don't have much knowledge so maybe only experienced people, but then how will I become experienced? Till reaching a point to for example design a complete bandgap generator

Ask questions. Nobody is perfect. "Classical" IC designers usually know the conservative way of doing things. They are more likely doing it the right way. This does not mean that it is the only or best way but your first guess should be that experienced designers are doing something for a reason and try to understand or ask for this reason.

Thanks for you inputs. I have commented and asked another question in your quote.

 

[Justinli]

I think maybe it's persistence and hard work.
Insisting on studying every day in case you forget what you studied before.
The hard work is to do more, I think make a project by yourself can remember more than write it down.