Multiple PSUs share load
In some projects, we need to deliver more power than a single power supply can provide, and in this situation we can use steering Schottky diodes to provide load balancing (Figure 1). In this schematic, we combine the output currents to provide some simple load sharing. Note that this is different from power redundancy, but instead the case where total output power cannot be delivered by a single power supply. This circuit is simple enough, and will work in ideal conditions where VPS2 = VPS1. What happens on the production floor is much more interesting, and invalidates this approach.
Two power supplies of equal voltage drive the load in current-share mode.
To evaluate the circuit, we can use the formula for Schottky diode forward voltage calculation at different currents to analyze the circuit performance at different loads and power supply deviations.
The problem is that this formula is a good approximation only, and you need to use appropriate n to get similar results to diode manufacturers’ V-I graph (in this case, n was selected to be 10). The analysis proved a little harder than expected since we have to consider two different power supplies, and calculate currents in an iterative manner. To solve this, we used multiple iterations using C (download available below) to calculate currents and voltages for this circuit.
The results were disappointing, since they show that in the case of ±1% voltage deviation, 90% of the power is supplied by a single supply. Basically, this circuit is not a good solution for power supplies with more than a few tens of millivolts difference. The problem is that not all off-the-shelf power supplies have output voltage adjustments – especially not the sealed ones. To solve this issue, a circuit was developed to ensure load sharing using off-the-shelf power supplies and components (Figure 2).
High-side current monitors U1 and U2 sense each supply’s current, and cross-coupled control of Q1 and Q2 equalizes the current sourced by each supply.
J1 and J2 are connected to input power supplies, and the load is connected to VIN. As shown, in addition to the original power steering diodes, we now also have Q1 and Q2 MOSFETs shunting the diodes to regulate load sharing. The MOSFETs are driven by U3B and U3A op-amps, each configured to compare the current of its own supply with that of the other. The circuit does not have any stringent requirements, but R1, R11, R2, and R12 should have 1% tolerance. The op-amps are driven from simple RC low pass filters to smooth out any transient response. We use U1 and U2 to measure the current at each power supply output, and use the RC filter-amplifier-MOSFET combination to equalize the currents. This solution has been proven to work for 12V-19V input voltages (common laptop power supplies), and used to provide 10A current to the load. Load sharing efficiency is good enough to allow cascading of these circuits to combine four power supplies. See the tables below for circuit performance results.
Links:
Schottky Diodes (the source of the diode formula, and other Schottky information)
ZIP file containing the C program and analysis spreadsheet (use Save As and remove the ".txt" extension)
About the author
Vardan Antonyan was born in Iran, grew up in Armenia and is now living in US. First realization for need to learn electronics came during teenage years when we were building a rocket to take us into space and suddenly realized that we need some means of communication. This is how I got involved in Ham Radio and was introduced into this wonderful world of electronics. We built 2KW transmitter with 10KV power supply using vacuum tube for power stage, the biggest I have ever seen. Our leader was Scientist that was coming up with all great projects and captivated our minds with his igneous approach of simplifying everything. I was hooked for life and by some chance I got my hands on Russian version of "Art of Electronics" in High school. I ended up reading that book many times and most of parts in the book ware impossible to get in Armenia, so it was almost reading Sci-Fi about those weird parts and imaginative solutions. I got first thrill of EE design when my friend complained that it is very costly to replace motor controller for automotive tape recorder because they were overheating and sometime blowing output transistors, his dad was running electronics repair shop. He asked me to help and I designed two transistor based constant current shunt regulator that was temperature compensated and had very low requirements on transistor characteristics. He was so excited and amazed that my design outperformed commercial controller, he run small production batch of the design. My carrier progressed and I had my ups and downs but there was always room in my life for electronics designs. While in collage I worked as electrician, electronics technician and after army I was installing/maintaining huge CNC machines and supporting their programming. After emigrating to US I worked some odd jobs while learning in NRE to become computer technician. After finishing the school and at my first interview there was sudden interest in my previous experience. I ended up taking an exam and starting my first normal job as electronics technician. Then it progressed to junior Hardware engineer, Hardware engineer and then Senior Hardware engineer. During my carrier span of over 15 years I was involved with all aspects of Analog, Digital and programming in Verilog, VHDL, c etc. In few words I am enjoying my job and cannot believe they are paying for this.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
