[AVR]MCU controlled power supply project, thoughts of structure and communication.

[BrunoARG]

I don't understand. Why do you want a reference generator for an ADC and a DAC? Reference generators are used mostly for accurate (and not variable) supplying (for most PC CPU there's a programmable "reference generator" which generates a range of voltages to supply the proccesor), another application is for analog signal proccesing with op amps and other stuff.

If the voltage you need is the same, you MUST use the same reference generator, everytime make sure that it can supply enough current (if not going to high impedances) or use a buffer if not (take into account the offset which may affect)-

Using the same voltage at two points with different generator is a useless idea, it increases the cost and the PCB layout gets worse too.

I don't know if you want to learn to design circuits, but it's not making them work but making them easy, simple, reliable and cheap to develop. If you're working on a personal project you can use innecesary parts but it's good to reduce that quantity. If you've ever designed a PCB you will understand when I say that the less components you use, the less tracks on the PCB and the more easy design of it.

 

[David_]

I'm not sure where I got the idea with two references from, but as the data converters use external references I need very stable and clean sources.
I sure want to learn but this will be my first PCB that is this complex, though that is not very complex.
In the system I'll use the actuall voltage of the DAC is not related to the output of the supply which is what the ADC will read, I'm not entirely sure that it will be the same voltages but for now it looks like it will work.

It depends on the final ADC choice, and now I might have to change some things as I plan to go down to 3,3V system voltage.
I have been struggling with the weighing of cost/preformens and errors, from the start I was setting out to simply produce the best supply I could, and not think about cost and fines of implementation, but that has changed quite a bit and I am now going over it all again to cut the unnecessary corners and make more accurate notations of cost and board occupation area.

- - - Updated - - -

I would like to reply sooner but I have not been able.

 

[KlausST]

Hi,

usually it is not wrong to use different voltage references.
Personally i try to avoid it. Thiis is becaus it saves money and it ensures more stable readings (not necessary more precise), because often errors in the ADC ref compensate with the DAC ref.
but this depends on circuit.

BTW.
The same is with frequency sources. If possible i only use only one frequency source and generate all other needed frequencies from this one source.
This prevents from alias frequencies caused by "about equal" frequencies or their overtones. This can avoid additional noise in high precision ADC/DAC applications.
The ADC errors caused by a "single" frequency source often cause DC offset that can be easily compensated.
Also two frequency generators influence each other so phase jitter of every frequency generator increases.


Klaus

 

[BrunoARG]

If you use a voltage as reference and you have a reference generator of that value, then you should use the same device to control both ADC and DAC. Otherwise, you will have two similar devices doing exactly the same which is not useful since you spend twice the money, twice the space and you'll still having the same result.

If you board had very high frequencies and long distances to cover from the ADC to the DAC, having into account the board and ambient noise then you should consider using 2 reference generators to get the accurate value on each device. If not, just make what I said before.

Hope being useful.

 

[David_]

Hello.

This has been standing still for some time and now I have done as previously advices, which is to re-design the power supply and base it on a SMPS(Switch-Mode Power Supply) with post-regulation by linear regulator.

I did not want to go that way when I first got that view but I knew that it was the way to go since a linear power supply wastes a lot of power and efficiency, from the start I did not want to even think about SMPS at all with the understanding that a SMPS would bring lots and lots of more noise at the final output. But since then I have read that if designed properly there should be no noticeable difference between a linear/switch-mode power supply?

I know in principle how the work but I have just begun learning about buck-converters and I have a couple of thoughts that I need some guidance with.

I started looking through Linear Technology's(linear.com) switch mode regulators and I found that there are quite a few that can output as much as 60V, but many more that can go to 36V which will suffice. But some have a max output current of ex, 20mA or 100mA and others a max output current of tens of Amps. What kind of applications does these very different power levels tend to?

In my last design I had a linear regulator who's output was driving a NPN darlington transistor in part to increase the output current but that also gave the opportunity to without any great pain apply quite aggressive filtering at the regulator input since so very little current are going through there. I have never built the complete circuit and I have no measurements to share or discuss about that but I'm not totally wrong it seemed lie a good idea.

But now its a new story where that will not be suitable, I think not at least.
As I see it now this is what I need to find out about:

How to treat the noise from a SMPS, where to apply what kind of filter/filters.

How to tie together the DAC controlled Feedback resistor divider with the switch-mode step-down converter and to have it also controlling the post-regulation regulator...

I need to find a suitable SMPS regulator, I see there are some with very complex external support circuits while some has next to no external circuits. I need a simple one and I need to check what kind of technique is used to adjust the output voltage, I hope I can still use the excellent low voltage DAC circuit that Klaus provided

If anyone have something to add to this please do, in any case I will post again when I have made any progress.


About the previous post, I have been going for two reference IC's before not because any real board constraints as such but the layout proved to make it a problem with only one. The DAC and the ADC is used in very different places in the circuit and it was very difficult and it mead a mess when trying to do it as you suggests. Difficult to to level that I was happy to pay for one more reference circuit.

Thanks for all responses, really appreciate it.
Regards