Storing a ramp signal voltage at a random time

[mordak]

Hi,

I appreciate if someone can help me with my problem: There is a ramp signal and a switch, that whenever the switch is closed the ramp signal should stop ramping up and the voltage should stay constant. For example, we have a ramp with slop of one and normally goes from 0 to 1V, and the switch randomly would be closed, say at 0.4 V. I need the ramp signal to stay at 0.4 V and does not increase anymore. I wonder if there is any simple circuit (no sample and hold ....) that can do that (circuit, not discrete components) and save the voltage for days.

Thanks

 

[rahdirs]

what is the problem with a sample & hold circuit ?

 

[schmitt trigger]

Hi,

I appreciate if someone can help me with my problem: There is a ramp signal and a switch, that whenever the switch is closed the ramp signal should stop ramping up and the voltage should stay constant. For example, we have a ramp with slop of one and normally goes from 0 to 1V, and the switch randomly would be closed, say at 0.4 V. I need the ramp signal to stay at 0.4 V and does not increase anymore. I wonder if there is any simple circuit (no sample and hold ....) that can do that (circuit, not discrete components) and save the voltage for days.

Thanks

No sample and hold?? How are you going to avoid the voltage from dropping, once that you've stopped the ramp?

I consider myself an analog guy, but this is one of those instances where it is far simpler and more accurate, and probably even cheaper, to do it with a single 8-pin microcontroller.

 

[mordak]

what is the problem with a sample & hold circuit ?

No sample and hold?? How are you going to avoid the voltage from dropping, once that you've stopped the ramp?

I consider myself an analog guy, but this is one of those instances where it is far simpler and more accurate, and probably even cheaper, to do it with a single 8-pin microcontroller.

Thanks guys for your comments. I try to avoid sample and hold, not only because it increases complexity, it also can't save the voltage for a long time since parasitic will discharge the sampling cap. Unless there is any dynamic analog voltage refreshing approach (don't know if that exists) it is not possible to keep the voltage for days.

Besides, I won't use any off-the-shelf component, I want to design at the transistor level, so I have more freedom.

 

[crutschow]

Thanks guys for your comments. I try to avoid sample and hold, not only because it increases complexity, it also can't save the voltage for a long time since parasitic will discharge the sampling cap. Unless there is any dynamic analog voltage refreshing approach (don't know if that exists) it is not possible to keep the voltage for days.

Besides, I won't use any off-the-shelf component, I want to design at the transistor level, so I have more freedom.

The only practical way to store such a voltage for days is to analog-to-digital convert the voltage and store the digital answer in memory.

Alternately you can generate the ramp with a sequence of digital words into a D/A converter and just stop the digital word at whatever voltage you want.

If you want to built those circuits with discrete parts, have fun. Shouldn't take more than a month or two. ;-)
What "freedom" are you talking about?

 

[Warpspeed]

You could use a tracking analog to digital converter, then just hold the count at that point "forever".

The basic idea is that you have an up/down counter that drives a digital to analog converter, which could be just a bunch of resistors.
(R - 2R network)

You compare analog output voltage of the a/d resistor network to your incoming ramp. If its higher than the ramp you count down to meet it. If its lower you count up to meet the ramp.
The under/over difference being detected by a voltage comparator.

It will rock back and forth by one count, but it will follow the ramp upwards all the way within one count accuracy. When you get to the "hold" point, you just stop clocking the up down counter, and it will lock both the analog output voltage and the digital count value at that value forever.

It can be done either in hardware, or in software, whichever you feel most comfortable with.

 

[schmitt trigger]

If you want to built those circuits with discrete parts, have fun. Shouldn't take more than a month or two. ;-)
What "freedom" are you talking about?

My same thoughts.

My first project I built 45 years ago used vacuum tubes.
I went thru germanium, then silicon transistors.

The project that the OP is proposing, which would be able to store an analog value for days..... I don't really think it can be done with discrete semiconductors only. It would have to be some sort of electromechanical contraption, like a motor-driven potentiometer.

Back in the 1970 I designed a light chaser, like the ones that they would use on theater marquees.
After struggling mightly with a discrete ring oscillator (the temperature effects would cause to change its speed wildly or even to stop oscillating), I settled for a cam-driven array of limit switches.
The contraption worked until the theater was torn down.

 

[FvM]

In the analog signal processing age, ramp-and-hold circuits (integrators with large, high quality capacitors) with days storage time have been designed. But obviously, DAC based mixed signal circuits can easily outcompete it these days.

 

[schmitt trigger]

Yes FvM, that is correct, but special layout techniques are still required to minimize board current leakage.

I remember one such integrator, used for some sort of military project. The PW board area where the sample and hold circuit was located had been cut out, and an insert made of teflon was installed in its place.

And if I remember properly, everything was in an oven to prevent moisture condensation.

 

[FvM]

I remember one such integrator, used for some sort of military project. The PW board area where the sample and hold circuit was located had been cut out, and an insert made of teflon was installed in its place.

And if I remember properly, everything was in an oven to prevent moisture condensation.

I also remember a 70er years color TV with analog frequency storage. The circuit was molded in high epoxy.

There are in fact many reasons why you won't use these techniques today, unless demonstrating ancient technology.

 

[Warpspeed]

From my point of view it depends on the particular situation.

If you are planning to go into production and build thousands, then a single chip low end microcontroller is the obvious solution.

If its a hobby project, or even a professional design assignment for a unique one off, it depends on where your skills lie.

As an analog designer I could get a breadboard prototype of something like this working and tested in less than half an hour.
There is no way I personally could write and fully test and debug code as well as assemble the microcontroller hardware in that time, but no doubt there a great many people here that probably could.

Its not a complex problem to solve with simple hardware.

But even a microcontroller will still probably need level shifting or other interfacing support hardware to make it usable.
So in the end, it may use almost as many parts and cost about the same.