EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: AlfBaz on April 11, 2014, 03:31:01 am
-
Just wondering if some of the more experienced members could shed some light on some of the things to look out for when using multiple power supplies in parallel or series
-
When running in parallel, ensure that there is a slight voltage difference between the two (or more) supplies. You need parallel supplies when you need more current, so we can assume that most or all will be (or may be at times) running in constant current mode rather than constant voltage. Regulation and ripple are usually worse in CC mode, and the small voltage differential helps ensure stability of the multiple supplies by ensuring that they transition from CV to CC mode one at a time. If they are set very closely in voltage, then they can interact with each other and oscillate. You may also need to add extra capacitance to the outputs of each supply to slow down their control loops.
When in series, you need to understand the implications of current limiting. The two supplies will always have at least a slight difference in current limit. This means that if you go into current limiting, then there may be a step (up) in the limit as the load resistance decreases and the lower-limited supply turns off. Note also that at that point (one one supply stops supplying current), the current from the higher-limited supply will be fed through the other supply through its output protection diode, a reverse biased diode across its outputs. Ensure that these diodes in your supplies are adequate; they are not always rated at the current limit of the supply (e.g. 1 A diode in a 3 A supply). Some supplies, especially older ones, may not have this diode at all. Any time you use supplies in series, you need to ensure that this diode can handle the full current being used. If this diode fails (due to being too small for example), then you could do considerable damage to the supply by applying reverse voltage to it. You can simply connect a large enough diode externally across the outputs (reverse biased!).
A while ago, I made a LTSpice model for an "ideal" current limited adjustable voltage power supply. You can easily connect these in series/parallel with various voltage/current settings to see what happens as the load changes. It won't simulate non-ideal behavior like instabilities. I can post this, just let me know.
-
Hi Macboy, sorry for the late reply and thanks for your insight.
The crossover between CV and CC hadn't even crossed my mind, not to mention circulating currents between devices and the effect on control loops
I'm still having a hard time "visualising" the problems you discuss and to that end I would indeed be interested in your ltspice models.
I've got an HP6632A on the way and was considering getting a second one. I've yet to have a detailed look at the manual and schematics for the unit but when I do I will keep your info in mind
-
Here is a simple (behavioral) model of a DC power supply with adjustable current and voltage. It has some output capacitance and a reverse-biased diode across its outputs to help make it act more like a real supply. You can adjust voltage and current by right-clicking on the V= and I= terms on the symbol in the schematic. You can control-right-click to adjust the capacitance, or you can edit the model.
To use this, save the DCPower.sub and DCPower.asy in the same directory as your schematic, then add the spice directive (press 's') ".inc dcpower.sub", then press F2 to add a component. At the top of the component picker, select the "Top Directory" to be your directory, and you will see the DCPower model listed. Connect wires to the + and - terminals, and note that the dials and meter are purely cosmetic. A sample schematic is included.
-
Thanks very much for that :-+
-
I've got an HP6632A on the way and was considering getting a second one.
Get on to the Agilent website and have a browse of their technical documentation for the 663x series - they have application notes that cover series and parallel connections. Also be aware that the 6632 current-control scheme is not a bog-standard current limit - it's designed to allow systems to be brought-up (and taken down) rapidly - it will allow current surges when the demand voltage is changed.
-
Thanks Andy, I'll start looking.
With regard to the 6632's output characteristics, it appears the Fast/Normal switch on the back plays a big part and there are stability, speed and noise trade offs depending on the selected mode. Apparently it switches output filtering caps in/out which can dump substantial current before CC kicks in
Tracking reveals the unit has arrived and cleared customs yesterday afternoon but alas the Easter long weekend will hold it up until Tuesday. Shame really I could've spent the next four days trying to blow it up :)