Help Needed: Suggestions for Final Project - Solid State Electronics Class

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
It's time to plan another Final Project. This most of the semester has been about opamps. Once we are done with opamps we will do some work on power regulation linear and switching, briefly cover AM and FM then ADC/DAC and various measurements.

We've done some great labs with oscillators, filters, integrator/ differentiator and log/ antilog conversions.

I am at a loss for ideas.. perhaps you guys have something interesting.

Few ideas that I had in the past were analog synthesizer, theramin, differential probe... possibly an AM radio transmit and receiver.
 

jpanhalt

Joined Jan 18, 2008
9,125
This "idea" may be completely worthless, but I will toss it out anyway.

Doing math operations with microcontrollers can require tables with extrapolation or use of CORDIC-like approximations. You could play with this approach:

digital value --> DAC --> analog value --> op-amp for calculation --> ADC and continue

I would compare the op-amp approach with existing approaches for speed, accuracy, lines of code, and memory utilization. Here are a couple of application notes to get you started:

https://www.analog.com/media/en/technical-documentation/data-sheets/AD538.pdf
http://www.ti.com/lit/an/snoa139/snoa139.pdf
 

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
The AD an interesting chip. I’ll have to dig into it a bit. I can see 3 log conversions going on and unfortunately cant see into the TI chip.

i would guess that analog computing is immediate with only slew rate and propagation delay. Digital ADC is very fast now a days and so darned useful but there’s always some delay and buffering depending on the source and use it may be relevant.

I would also guess that if you need immediate reaction analog would be the way to go.

I’m sure the professor would love a project that proves analog is still better in this regard...
 

jpanhalt

Joined Jan 18, 2008
9,125
In the old days, the mass spectrometer crowd looked at GLC (chromatography) as an inlet system. The gas chromatography crowd looked at mass spectrometers as detectors. Of course, they augment each other. That was basically my suggestion to have analog do what it does best to augment further digital processing.

I am glad you didn't think the idea was foolish. Please keep us updated.

John
 

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
Not foolish at all... but very complex question with complex answers. There are limitations to analog as there are with digital. There are benefits to analog as there are for digital. I'm not sure if it's really possible to do a simple comparison... for instance:

For summing, products, exponential - analog is quick... of course there will be challenges to accuracy which will demand even more work and circuity such as addressing thermal issues.

For integration and differentiation, digital might win since it won't be as hampered by capacitance.

More I look into this, it would be complex and perhaps too much for this final project. It is an interesting conundrum and one we are always faced with.

The US Navy had a hard time getting rid of Analog computing (mechanical) because it worked so well. There were huge power/weight and maintenance requirements. Electronics won in the end for various reasons.

I did a lot of audio work and I've got to say that digital isn't any worse and it's definitely better for S/N requirements. Look at the tiny Class D amplifiers... they're incredible and so efficient. I remember when 100 Watts was something that meant a lot of weight and power (Heat).

I've been out of microwave for a while but I'm sure there's a lot happening in that world that's digital now.

Again very complex answers and testing would get quickly slanted one or other without care...
 

JoeJester

Joined Apr 26, 2005
4,259
The US Navy had a hard time getting rid of Analog computing (mechanical) because it worked so well. There were huge power/weight and maintenance requirements. Electronics won in the end for various reasons.

You mean analog ISN'T electronics?

Digital won out for the obvious reasons.

Divide by 10 counters using
has fallen by the wayside to digital.

Even Admiral Hooper's computer bug

messed up the operations.

I have seen gecko's knock us off air, snakes causing havoc, and one unit had a mountian lion take refuge in their foyer to escape the cold winter night in montana.

So, we must adapt and overcome, then and now.
 
Last edited:

JoeJester

Joined Apr 26, 2005
4,259
Perhaps I wasn't clear: Mechancial Analog Computers:
I agree the electronics have replaced the targeting equipment. GPS increased the delivery accuracy, touting it to be within 50 meters CEP (circle error probable).

I doubt the Zumwalt has people moving that turret or loading the gun at that rate. There are electro-mechanical portions still in operation which if there is a failure, all the electronics in the world won't help.

I have no doubt the old electro-mechanical computers were a bear to maintain. The gearing on a purely mechanical computer would be worse.
 
Last edited:

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
I cant imagine the work that went into profiling a new cam for different loads... it is incredible watching that video. I believe the M9 gun director work by Bell Labs was the turning point for electronics. The amount of power requirements and sheer weight of the mechanical computers must have been incredible.

I was doing some research into analog PID vs digital. Analog is faster but due to time it takes to control things like valves it didn’t matter.
 

atferrari

Joined Jan 6, 2004
3,930
I cant imagine the work that went into profiling a new cam for different loads... it is incredible watching that video. I believe the M9 gun director work by Bell Labs was the turning point for electronics. The amount of power requirements and sheer weight of the mechanical computers must have been incredible.

I was doing some research into analog PID vs digital. Analog is faster but due to time it takes to control things like valves it didn’t matter.
Had the chance to watch one "table" solving distances (¿?) Impressive.
 

Hymie

Joined Mar 30, 2018
802
To impress your lecturer, design & build something not currently on the market.
Often when an RCCB trips, if it is an intermittent fault it can be very difficult to isolate the cause (faulty piece of equipment connected or the RCCB itself).

To assist with the above, what is needed is a system that is logging the earth leakage current (capable of measuring down to a mA using a current clamp and up to say 200mA). The AC signal from the current clamp could be amplified and rectified to give a logable voltage signal. Given the mains frequency and the speed of operation of the trip, readings would need to be made at least 100/second. The logging system could be some form that is commercially available – otherwise the project is likely to be too complex for a final project.

Ideally the system should indicate the leakage current in real time. It might be possible to limit the required data storage by overwriting older data (where no trip occurred), and stop logging once a trip event occurred.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
I've submitted my project idea, to demonstrate an analog opamp based PID controller. It will control a motorized potentiometer or some sort of motorized device which will feedback a voltage.

It will utilize LM741's which I have in abundance here and at school with following cascaded: variable gain, integral and derivative funtions. It will sum and feedback on itself. It's simple but very useful and not covered this semester. I will add the controls for each stage and we should get immediate feedback from the motor.

This devices should oscillate, jitter and overshoot without PID closed loop.

I'm interested in this as it might be a good device to take to my First Robotics team where I mentor so I can demonstrate the PID feedback and why it's important and how to use it to control the robots.

I have a friend who is doing active controls for amateur rocketry and this circuit might help him get more stable flights... I can see oscillations in his trajectrory... again... very important stuff... of course he should just use code but analog might be quicker.
 
Last edited:

atferrari

Joined Jan 6, 2004
3,930
I implemented an analog PID controller IIRC using all TL072. Started from scratch by reading all I could.

My stumbling block was how to avoid windup. Solved using a comparator with a floating span center voltage.

Interesting project.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
I have approval, it should be interesting. Initially I wanted to use a motorized potentiometer but I plan on using an old Maxon DC motor I have lying around. Plan to mount it and couple it with a potentiometer using a belt or similar.

the motor driver will have MJE3055 and complementary MJE2955 in push pull... nothing really challenging in the circuit but it will be a fun one and will be useful to demonstrate what it does and why it’s important for my first robotics group.

Due to Covid 19, we were told to stop at school to pick up scopes, function gens, power supplies and components.

I‘m using LM358s so I shouldn’t have latch up issues.
 
Last edited:

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
Just ran a sim... its been simplified a little to run on LTSpice. Looking forward to the build.



1584583854228.png

1584583901887.png
 
Last edited:

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
In the actual circuit it will be a potentiometer for position control of a motor at the other end. The feedback provided by the motor's potentiometer.
 

Thread Starter

Wolframore

Joined Jan 21, 2019
1,743
not sure what part of it you dont understand... here’s an interesting demonstration although this one uses a motor speed control and the algorithms are done in a microcontroller... the concept is the same.

https://hackaday.com/2016/11/08/helicopter-pendulum-is-pid-licious/

Here is another I believe this one is also done In programming but it shows a similar position control to one I will be demonstrating and perhaps it explains why it’s important.

https://hackaday.io/project/12171-robotresto-restoring-a-4-axis-arm/log/41880-moving-the-arm-with-some-pid-dc-motor-control

My project is entirely analog since the assignment requires use of linear semiconductors. Maybe next semester when I’m taking microcontrollers I can work on a vision based design.
 
Last edited:
Top