Designing an Opensource DC Load

[OpenWare]

I am designing Open source DC load with multi channel /Modular design, It would have a touch screen interface along with encoders knobs for fast selection and toggle for menu.

Suggestions are welcome
HMI interface - STM32H750 With 5" or 7" Touch Panel
DAC - 16 bit DAC
Resolution- 0.1mA @ 6.55A OR 1 mA is 10A.
Working Voltage 150v
Working Current 6.5 AMP -10 Amp
I need input from the community on the consideration taken to build a precise and accurate DC load

[Roehrenonkel]

Hi OpenWare,
 
welcome to the forum.
Make it understand SCPI,
over  RS232, LAN, GPIB or so.

Good luck

[Doctorandus_P]

Building a DC load is a common project, there are probably already 100+ of them floating around the internet. You can look at several of them to get some inspiration on how to build one yourself.

https://html.duckduckgo.com/html?q=open+source+dc+load

[OpenWare]

Agreed with you, there are literally tons of project but accuracy and precision is lacking in nearly all of them.

[RoGeorge]

You may want to make it work in all four quadrants, those are rather rare.  Search for SMU.  Some models have service manuals and schematics available online, for inspiration.

[Andy Chee]

Agreed with you, there are literally tons of project but accuracy and precision is lacking in nearly all of them.

Interesting that you mention accuracy and precision.

Have you characterised, tested, and analysed the performance of existing DC loads?  Or are you just guessing they lack accuracy and precision?

Or maybe someone else has already benchmarked DC loads, in which case, you might want to give examples of bad performing loads.

[Hawaka]

Features are quite important, mostly software. It should display directly the wattage, Wh, Ah and so on.  A battery test mode is also nice, where you can set a current and a low voltage limit. It will then draws current until the voltage is reach and display how much energy it was. Graphing and data logging is also important.

Every functions shall be accessible from the GUI. Most of the people don't want to use a dedicated software for setting a DC load.

150V @ 10A is 1.5kW, would be nice if the load was regenerative.

Good luck with your project!

[Doctorandus_P]

Agreed with you, there are literally tons of project but accuracy and precision is lacking in nearly all of them.

Well, you only need one don't you?

And I agree that most of the designs are pretty simple / basic, but there are also a bunch of (nearly) metrology lab worthy open source projects. But he more "quality" you wish, the more and more expensive the parts get. There is always some compromise to be made.

Have you built one of the simpler projects? It's a good way to start as a learning experience, and it may be good enough to be actually useful too.

And talking about SMU, there is a quite nice project floating around which has a row of 20 or so 7 (or 16?) segments displays in a row.

[schmitt trigger]

As others have mentioned, a 1.5kW supply is *a lot* of power to attempt to regulate with a straight linear circuit. The heat management challenge will be significant.
You should be giving some thought along the lines of a SMPS pre-regulator, or at very least a transformer tap changer.

There are older preregulator techniques utilizing SCR phase control, and many classic supplies from Sorensen and Hewlett-Packard used it.

I ignore whether regulatory agencies would require some sort of power factor correction and/or EMI suppression.

[thm_w]

As others have mentioned, a 1.5kW supply is *a lot* of power to attempt to regulate with a straight linear circuit. The heat management challenge will be significant.
You should be giving some thought along the lines of a SMPS pre-regulator, or at very least a transformer tap changer.

There are older preregulator techniques utilizing SCR phase control, and many classic supplies from Sorensen and Hewlett-Packard used it.

I ignore whether regulatory agencies would require some sort of power factor correction and/or EMI suppression.

Load not a supply.
Load will have a power limit, say 200W, you can do 150V or 10A but not both.

[schmitt trigger]

ooops, that is what happens when one has many windows open and is replying to several threads simultaneously.

You are correct, it is a load, not a supply. 

[Darkover]

I need input from the community on the consideration taken to build a precise and accurate DC load

The biggest question you have to find an answer: What bandwith did you need and how can you made it stable with unusual input. 

Olaf

[nctnico]

Agreed with you, there are literally tons of project but accuracy and precision is lacking in nearly all of them.

Interesting that you mention accuracy and precision.

Have you characterised, tested, and analysed the performance of existing DC loads?  Or are you just guessing they lack accuracy and precision?

Or maybe someone else has already benchmarked DC loads, in which case, you might want to give examples of bad performing loads.

Late to the game but having looked around at DC loads quite a bit myself, it turns out many are not very accurate at all. Mains ripple due to the magnetic field of the transformer gets into the regulation circuit quite easely causing a 100Hz ripple in the tens of mA range. Most of the low to medium cost DC loads seem to suffer from this problem due to compact build.

IMHO a nice feature would be dual range with a 0 to 1A range which is accurate to 100uA or better. And a 0 to 10A range with 1mA accuracy.

[Attenbach]

IMHO a nice feature would be dual range with a 0 to 1A range which is accurate to 100uA or better. And a 0 to 10A range with 1mA accuracy.

Equally mentally invested.

I was wondering the same and thought, whether it is possible to have two or more shunt resistors, say 1mOhm and 10 mOhm (and 100...) parallel to each other with each of them getting connected to common ground through a LV low Rdson NFET. Instead of a single current shunt between the load transistor and ground, you split up the net into three different paths and you only engage the path that gives you the maximum dynamic range.

Edit, I just realised you cannot wire the kelvin connections of all shunts in parallel, as the inactive shunts still provide a current path for the potential across the kelvin connection, hence instead of measuring the voltage drop through the active shunt, the connections would carry current themselves! Stupid me.. so yeah you'd need a separate adc or a multiplexer to deal with this.