Variable DC Power Supply 0-20V +/-0.25V Accuracy

[bentx0rx0]

I need to provide between 0 and 20V, with an accuracy of +/- 0.25V, controlled by a microcontroller at will. I can only think of two solutions, an adjustable regulator with a variable resistor on the adjust pin, or a variable resistor on a 20V DC line. Both these options seem pretty ridiculous, I've never seen the first one done, and the second one wastes a TON of power. Is there a circuit or component that I don't know about that will be both simple and efficient? Thank you

[qno]

sounds like a 7 bit DA converter with an amplifier.

[tszaboo]

You make a -2V and +22V power supply with something or a 2V virtual ground. You put a http://www.ti.com/lit/ds/symlink/opa547.pdf between the two supplies. You set the gain 4. You connect a 0-5V DAC to the input. And put a big heatsink on the opa.

[Bored@Work]

Actually, this has been discussed here in probably 10th of threads, maybe even already 100 threads.

For adjustable three-terminal ("high side") regulators like LM317 or LM350 raise the voltage on the adjust pin to Vout - Vref. Typically requires a current-sinking buffer (opamp) following a DAC, or switched resistors (a DIY DAC).

For four-terminal ("low side") regulators inject an additional current into the node where the feedback pin is connected. Determining the right current range, possibly generated from a voltage DAC with an additional resistor or directly from a current DAC, requires a bit of calculation involving Kirchhoff's laws and Ohm's law.

If you want to waste money, and want to look stupid use an expensive digital pot to adjust a resistor in the feedback loop of either type of regulators.

For a discrete build regulator or opamp with power stage either replace the voltage reference with some DAC output, manipulate the feedback loop as described above, or wrap a (current) DAC around the error amplifier (a variant to manipulate the feedback loop).

[bentx0rx0]

The Opamp method looks good in theory. I'll have to provide upwards of 3.5A to the circuit, which I'm pretty sure kills all but the most expensive high current opamps

Bored@Work's idea seems promising, but I admit that I don't really understand exactly  what you mean. On the high side regulators if you raise the voltage to Vout -Vref, you'll get a desired output voltage. By inserting an opamp buffer, you can adjust the voltage to the adjust pin, giving the desired output. I'm assuming it's not a good idea to just connect a DAC input to the adjust pin, as that would harm it's ability to self regulate without an external feedback loop. Is this correct?

I can't find anything describing the difference between high side and low side regulators, so I don't really understand the second part about those kinds of regulators. I had no intention of using a digital potentiometer, but I knew those existed. Can anyone explain the hate for digital potentiometers. I've read that they're bad for audio, but is there any other reason? They're not really expensive(60 cents MCP4017), there's a MAXIM app note on the subject: http://www.maximintegrated.com/app-notes/index.mvp/id/225
Is there something I'm missing?

[tszaboo]

3.5A <- you should have mentioned this. This is not a simple project anymore. First you need to decide, if you want it to be switching, or linear. If linear, than look for the internals of some agilent E3xxx power supplies, that is great startpoint. Some have SCH posted on their website.
If switching, and you dont have already some idea how to do this, than look for pre-made modules, there are quite a few of these.

[bentx0rx0]

I should have, but I did want to see what you guys would recommend for a variety of amperages. I've made this circuit, which uses a TPS5450 5A regulator and an opamp on the adjust pin to set the output voltage. With a gain of 6, I should be able to set an output from 1.221 to the max of the TPS5450 with a DAC chip. What do you guys think, will this work?


I got the capacitor, inductor, and diodes from TI's Simple Switcher program. Simulations seem to indicate I'll get the desired voltage on the adjust pin, but my experience with Opamps is limited, I'm still studying them in uni

[ve7xen]

Your opamp is going to drive its output to DAC voltage * 6, as I think you expect. However you've got this directly conected to the Vsense pin, which is the voltage sense for the regulator's feedback loop, which has an internal setpoint of 1.221V here. It's going to turn the output completely off if your set voltage is > 1.221V, or completely on if it's < 1.221V, since the regulator no longer has a closed feedback loop.

One way to do it would be to put the opamp at the bottom of a voltage divider from the output, with Vsense tappinig off the centre of the divider. You can then create a virtual ground for the divider which will let you control the feedback loop by setting the virtual ground to the voltage that will cause the output to be what you want. In other words, you set the output of the opamp to be the voltage that causes 1.221V to appear on Vsense when the output is your desired voltage.

Of course you need to do a little bit of math to determine what voltage to set the opamp to. You also won't get below 1.221V without a negative supply rail.

[macboy]

The Opamp method looks good in theory. I'll have to provide upwards of 3.5A to the circuit, which I'm pretty sure kills all but the most expensive high current opamps

LM3886 (marketed as a audio amp, but it is of course just a power op-amp) can do around 10 A.
OPA549 is spec'd for 9 A continuous. It has a simple to use (but not super precise) current limit set via a resistor.

Of course you still need to take note of the power dissipation. So if you want high current at low voltage output, the regulating device will need to dissipate a lot of power. You may need to derate the current at lower voltages, and/or use a variable power supply for the op-amp (such as switching taps on your power transformer).

If this is a one-off project, then slightly more expensive parts make sense vs. spending time designing a "less expensive" alternative. Time costs money too. Consider a DAC with internal reference, microprocessor interface, and voltage output, like the TI DAC7741 (or obsolete DAC707/708/709). These provide very simple to use 0 to +10V output. Amplify that with your power op-amp, and you are nearly there. It will run you at least $10 (as will the op-amp), but it is a simple solution. Keep in mind that this will be a full source/sink supply, capable of sourcing or sinking current to keep the output at the set point.

You had specified zero to +20 V, then you suggested a circuit that could only go down to about 1.2 volts. So which is it? Do you need an output that goes to zero or not? This seems trivial, but it is a significant consideration in designing a regulated supply, since the requirement to go down to zero will eliminate many potential solutions.

[Harvs]

I need to provide between 0 and 20V, with an accuracy of +/- 0.25V, controlled by a microcontroller at will. I can only think of two solutions, an adjustable regulator with a variable resistor on the adjust pin, or a variable resistor on a 20V DC line.

What's it for? How many do you need to build?  Because the reality is you can get a serial controlled linear lab supply dirt cheap that'll do that spec.  If it's just to power some one off test jig or the like you're going to better off just buying a supply ready to go.

[bentx0rx0]

Dirt Cheap huh. Care to point one out? It might be useful for testing the prototypes.

I'm building a series of induction chargers based on the Qi standard as a project for university. I've got two designs one with a single coil and one with 4 arrays of 3 coils. One array of 3 coils draws 3.2A and needs to supply AC through a half bridge inverter, which was the topic of another thread. The spec says I should be able to do +/-0.25V, but seems to indicate that it works mostly at 12 or 20V. I don't think it's a requirement to get very close to 0V.

The design in the above post looks like it could work well. The parts are specced from 1.22V to 25V, and simulations suggest I can get good accuracy with the opamp controlling the DC power supply.

On the quad coil array device , I was going to use 4 of the above design with a quad opamp(LM324), 4 DACs and a microcontroller. This design setup seems a bit obtuse, not to mention somewhat expensive. I haven't figured out a nice way to get the AC wall power to the device itself. I was thinking about using a transformer/rectifier setup, which would make the device bulky but be relatively simple for me to implement. Most of the power supplies(for laptops) I've found can't really provide the power I seek(20V*13A = 240W).  could probably compromise here, as I've found several laptop chargers that can supply 180W.  More realistically the device will probably draw closer to 30W-75W, as a couple tablets and a phone or a lithium ion charger operate in this range.

The single coil device is quite a bit simpler. One power supply, one DAC, one coil. Likely to draw <3.2A, I decided to just reuse the above supply to make my life easier and give me some headroom in the device itself. I was thinking about adding a 5V->30V boost converter, or replacing the design in the thread, so that this one has the potential to be powered over a USB connection and by common phone/tablet wall warts with a microUSB connection. This one could have uses in industrial design and hobbyist projects, allowing hobbyist integration of inductive charging into robots, wearables, and things more easily. The major downside to this one is that the coil's have to be aligned in with magnets, which can be a pain. It'll be a lot smaller and sleeker though.

[JoeO]

Dirt cheap depends on what country you are in.  Set your country.

[tszaboo]

75W??? What are you charging, a water kettle? Most phones can hardly be charged by more than 5W, and with the awesome 30% of efficency of the wireless transfer you hardly need more than 50W for the whole system. Read this:
http://www.ti.com/lit/an/slua635a/slua635a.pdf
If you have a multiple coil charger only one coil is active at a time. It is the one, which is closest to the phone's coil, so you don't need to position the phone accurately. You also dont need a DC power supply, you need to drive the coil with a PWM, 20V or 12V and the duty cycle of the PWM signal will make a 0.25V accurate voltage inside the phone.
First, understand how the system works, and review project requirements. Buy a demoboard, they are cheap, compared to failures.