Some folks special love digital circuits, some analog. Many problems have good solutions in both domains. Some are best solved in one or the other, some are best solved with a hybrid solution.

How do you choose which direction to go? How do you decide to take a hybrid approach?

Do you have some heuristics for selecting the best route?

 

Been using analog a lot longer (started in the military) so I will pick analog well past when it makes sense to use digital.

I guess it would be called the stubborn approach.

Most times I find out that digital was the better route when some youngster makes a statement like "Dude, that's an interesting solution" and then proceeds to digitalify it. Then the light goes on and I hate digital (and kids on my lawn) even more!

 

Go digital when you can... with digital it [usually!] either works or it doesn't, with analogue it can work, but with every shade from working well to only just + noise.
It also depends a bit on how many components it takes to do the job.

 

Each has its natural domain that depends on the technology you have access to. The mathematical foundations of the solutions (digital implementations of analog electrical functions) are the same but when you need a fundamental function like very short time measurements, using analog for the base electrical element of change makes more sense than a pure digital implementation if your digital technology is limited by the short time periods measured.


A good example is the CTMU unit in a controller.

 

I tend to go digital where possible, most Micro's have the analogue input option, but no analogue out module.
Max.

 

Some folks special love digital circuits, some analog. Many problems have good solutions in both domains. Some are best solved in one or the other, some are best solved with a hybrid solution.

How do you choose which direction to go? How do you decide to take a hybrid approach?

Do you have some heuristics for selecting the best route?

I have been around much longer than I expected. I started learning analogue electronics in the early 50s. I started learning digital in the 60s. I started learning how to program microprocessors in the early 80s. They each became just another resource to me. I am still learning. I use whatever makes sense and what I have on hand at the time. I do have a very good stock of components and an endless junk box. Quite often my creations contain analogue, digital and microprocessor circuits all working together. most problems have more than one solution. I usually use the simplest way to achieve the results I want.

 

How do you choose which direction to go?

My background is biochemistry and chemical engineering and not electronics. Most of the inputs and outputs I might be interested in are analog and slow or DC, and so I tend to think analog. But I've also learned that just about everything in between an input and an output is easier to accomplish in the digital world. Once your data exists in digital form, the manipulations you can accomplish are limited only by your imagination.

So there's a big toolbox with a whole bunch of tools I don't know how to use and just a very few that I do.

 

I have been around much longer than I expected. I started learning analogue electronics in the early 50s. I started learning digital in the 60s. I started learning how to program microprocessors in the early 80s. They each became just another resource to me. I am still learning. I use whatever makes sense and what I have on hand at the time. I do have a very good stock of components and an endless junk box. Quite often my creations contain analogue, digital and microprocessor circuits all working together. most problems have more than one solution. I usually use the simplest way to achieve the results I want.

I could have written that verbatim.!
Max.

 

On reviewing a few projects, seems to divide into 1/3 mechanical, 1/3 digital & 1/3 analogue. I generally prefer digital if it's held to 3 or 4 ICs. Some day I'll take time to consider micros but at 92 don't think that will happen.

 

I could have written that verbatim.!
Max.

Same here.

 

On the same note; I occasionally still do a bit of consulting. I recently helped the developer of a different kind of turbine to make cavitation measurements. It took several thousand dollars worth of off-the-shelf equipment a working with a computer to identify the problem. He then needed to monitor several installations so I designed and built an analogue circuit that would make the same measurements for a tenth of the cost of the off the shelf solution. It all depends on what you need to do, and what resources are available.

 

So for me digital seems to always require less components but adds compile and programming time for the IC sometimes more work with JTAG or something else...
With analog,what you build is what you get so no programming needed. But component count is usually higher.

I like the weigh the two and use what makes more sense. Less components usually mean smaller size...

 

As a rule of thumb: Use analog only where absolutely necessary (e.g., interface with the physical world), and digitize early in the signal path. With very few exceptions, digital is far superior to analog in terms of noise-immunity, flexibility, and maintainability (firmware/software update vs PCB redesign!). And, if you have an MCU or FPGA, digital has unmatched processing capabilities.

 

As a software engineer and systems architect for 50 years, guess which I prefer?

I see problems solved in code in my sleep. Simple solutions take me ten minutes to wire up, code and test.

Not saying digital is better. Just that it’s easy peasy for me...

 

Since the real world is totally analog, the choice of circuits must depend on what the results need to be and the complexity to get from input to output. The virtue of analog circuits is that they are simple to understand and monitor, and the advantage of digital approaches is that discrete digital circuits are also easy to work with. So discrete digital logic can do a whole lot very reliably and with more immunity to electrical noise. BUT beware, because soon as programmable functioning becomes involved suddenly another skill set is required, and usually there is also expensive hardware required to convert that thought train to functional code, and load it into a system. And then there is that whole arduino world where they use different words for just about everything, so that communication with others is less convenient.

So the bottom line on which to use is that it depends mostly on what needs to be achieved.

 

Once the product ships, software changes ( updates & patches ) usually cost a lot less than hardware changes

 

Once the product ships, software changes ( updates & patches ) usually cost a lot less than hardware changes

Certainly correct. THEREFORE it is important to get the design right the first time. Most companies don't embrace the microsoft business plan of "ship junk, and then send out fixes as required." We always made sure a job was right before it ever shipped.
Besides that, the TS was asking about personal projects, at least that is what I think that I read. AND it is simpler to get hardware correct the first time.

 

I tend to go digital where possible, most Micro's have the analogue input option, but no analogue out module.
Max.

PSOC, 5LP family, has 4 DACs, I or V or any combo of I & V.

Regards, Dana.

 

As a rule of thumb: Use analog only where absolutely necessary (e.g., interface with the physical world), and digitize early in the signal path. With very few exceptions, digital is far superior to analog in terms of noise-immunity, flexibility, and maintainability (firmware/software update vs PCB redesign!). And, if you have an MCU or FPGA, digital has unmatched processing capabilities.

All true but unless designed carefully for signal fault identification, digital control systems can have wicked software corner-case/edge-case failure modes that can transform real-world analog inputs into sequences of analog output commands almost impossible with pure analog control loops. I've worked with some vendors for months to de-glitch modern digital updates to old analog control racks. These systems worked perfectly on a test-bench with all specification meet but failed in operational use because they were actually too fast when reacting to noisy signal changes the analog system integrated to nothing. My solution was to old-school analog pre-filter the main culprit signals before and post-filter after the digital stage, their 'solution' was to just idle (maintain current settings) digital control during the upsets.

 

I too have been around much longer than I expected! I always wanted to learn computer circuits, but just never had or took the time. I always wanted to design a analog computer for my car. I have it all down on paper, but I know how that goes. I was just working on the hardware for the capacitor discharge ignition, which was the hardest for me to design. It will have variable advance and a variable RPM limit. Something to do on days that are too cold to work outside!