Do I need an MPU or MCU?

Other Hardware 3rd Party Boards
1

I need 10 software controlled voltage regulators with digital feedback.

This is what I have now

PWMrefsup.jpg
PWMrefsup.jpg2250×1281 231 KB

The microcontroller will fight to maintain the intended voltage so if it is higher or lower then the voltage I set then it will self-correct.

I need to ditch the RC filter to rid myself of the time constant and go with a full blown dac solution.
I also possibly need more ADC/PWM resolution? I'm not sure if the step resolution on 8bit is too small for stable and accurate operation.

I need to control 10 different regulators and measure various currents so I need probably 24 ADC and 10 DAC ports.

I'll also end up needing a display that will show a bunch of different data and allow visual adjustment of operating points.

Do I need to use an MPU solution with dedicated ADC/DACs and whatnot or is there some other solution?

2

What you absolutely do not want to do, is to insert a microcontroller in the feedback loop of a voltage regulator. Can you guess why?

You need DAC arrays feeding proper regulator elements. If you want feedback, you monitor the control signals of the regulators - they will tell you whether the voltage is correct or not.

3

I don't follow, mind explaining that to me in laymans terms? What's wrong with using an ADC to measure the output voltage and use that to tell whether the micro needs to increase or decrease the output?

4

Just have a think about what computers often do that you do not want them to.

5

I have no idea, my computers do everything I want them to. How bout just telling me?

6

I expect you'd not want a component that could potentially go MIA (restart, dead time while starting up, hang etc) in this feedback loop leaving your system in an uncertain state.

Geoff

7

Why would a simple ADC feedback loop restart or hang on me? (by the way you guys must be using old computers if they still hang on you :P)

8

I was just speculating on what Paul was asking. There's the possibility the uC won't be running all the time the feedback loop needs it during startup.

9

Can you tell us more about your application? Specifically, "What is the load on each of your regulated voltage outputs?", "Is the load stable or does it fluctuate?", "What is the range of output voltages you need?, and "How fast do you expect the output voltage values to change?"

As drawn, your circuit has high impedance (~1k) for a voltage source which normally should be much lower than your load impedance or else it will be very sensitive to load changes. Your feedback loop is relatively slow (A/D conversion time divided between 10 controlled voltages) so load variation of even a few 10s of hertz is going to reek havoc on the voltage stability. If, on the other hand, you're setting a nominally fixed voltage across a nominally fixed load, then PWM with heavy filtering might be ok.

If you really need 10 DACs, I doubt you're going to find them on a single microcontroller.

10

In industry this kind of regulation circuits is often used to compensate slow load changes. Using this kind of circuit or a pure analogue one is a choice to be taken based on the boundary conditions you need. What is the regulation speed you need? How quick can fluctuate your load and what steps does it do?
In any case you may consider using a PID type of regulation and you must be sure about your regulation parameters otherwise it will start oscillating.
What is the application for your project?

MrMark:
If you really need 10 DACs, I doubt you're going to find them on a single microcontroller.

The PWM output, R9 and C1 are already behaving like a DAC here. So he only needs 10 independent PWM outputs. That is possible to do with a Mega or Due. Otherwise for more accuracy he could use MCP4922 external SPI driven DACs.

11

MrMark:
"What is the load on each of your regulated voltage outputs?"

I've changed it from a common emitter regulator to a common emitter regulator that feeds a source follower, this seems to be the best way to fake a resistor without having unwanted effects.
So the load of the regulator should be zero.

MrMark:
"Is the load stable or does it fluctuate?"

It depends really, I have multiple circuits I intend on using this in. If you mean load resistance then I expect it to be mostly stable.

MrMark:
"What is the range of output voltages you need?

0-600v

MrMark:
How fast do you expect the output voltage values to change?

I'm not sure, I only want to stabilize the DC voltage though.

MrMark:
As drawn, your circuit has high impedance (~1k) for a voltage source which normally should be much lower than your load impedance or else it will be very sensitive to load changes. Your feedback loop is relatively slow (A/D conversion time divided between 10 controlled voltages) so load variation of even a few 10s of hertz is going to reek havoc on the voltage stability. If, on the other hand, you're setting a nominally fixed voltage across a nominally fixed load, then PWM with heavy filtering might be ok.

I intend to ditch the filter and replace it with a dac because I want to rid myself of these issues.

flxmx:
What is the regulation speed you need?

Only DC level.

flxmx:
How quick can fluctuate your load and what steps does it do?

Not sure what you mean

flxmx:
What is the application for your project?

The essence is I have a bunch of different things that have operating points based on resistor values, I want to have a digitally controlled variable resistor so I can manually control these devices.

flxmx:
The PWM output, R9 and C1 are already behaving like a DAC here. So he only needs 10 independent PWM outputs. That is possible to do with a Mega or Due. Otherwise for more accuracy he could use MCP4922 external SPI driven DACs.

I would need to decrease the time constant and increase the filtering for that setup, I'm thinking of just replacing them with pure dacs so I can program in the time constant and not have to worry about filtering.
Also, while my current set-up does work, the accuracy is off, for example if the output voltage is set to 5v, the output will stabilize at like 4.9v or something. I'm not sure if that is due to the dac or the adc.

Also I want to measure current and voltage at quite a few points in my circuits so the amount of DACs/ADCs I need may be a lot more than I originally thought. I'm wondering if it would be easier to just buy a dedicated multichannel DAC/ADC and then control it with an MPU.
Although from what I read MPUs require dedicated power management ICs and memory ICs and a whole bunch of other stuff. I'd be clueless on how to properly connect all of that. I've yet to be able to decipher the pins of a uC let alone properly connect them.

Although, in the end I intend to have a high res display that will show various amounts of data and allow real-time adjustments and it will also handle various operational tasks such as varying the operating points based on certain parameters and providing limits to certain operations for safety.

I wonder if all of this will need the power of an MPU anyway?

Maybe I need to make some sort of sooped up MPU based arduino with loads of ports?
I certainly would want to keep the arduino based programming, it would suck if I had to convert to some more difficult method of coding.
I've read that MPUs have some latency since they don't have on board memory and such, I wonder if the latency will be an issue here? Perhaps what I need is just a bunch of MCUs strung together?
Maybe I can multiplex my way to a bunch of high bit dacs/adcs on some unos? Is there some sort of bandwidth limit on that?