We often find that some rules or principles we take for granted may actually contain errors. Electronic engineers face such examples in circuit design as well. Below is a summary of eight common misconceptions compiled by an engineer.
Misconception 1: "This board's PCB design requirements are not high, so I'll use thinner traces and let the auto-router do the job."
Comment: Auto‑routing inevitably occupies more PCB area and creates many more vias than manual routing. For high‑volume production, PCB manufacturers consider not only commercial factors but also line width and via count when reducing prices – these affect the board's yield and the consumption of drill bits. Saving costs for the supplier gives them reason to lower the price for you.
Misconception 2: "I'll pull all these bus signals up or down with resistors, just to feel more at ease."
Comment: There are many reasons for adding pull‑up or pull‑down resistors, but not every signal needs one. A pull‑up resistor on a pure input signal draws only a few tens of microamps, but on a driven signal it can draw milliamps. In modern systems with 32‑bit address and data buses, plus buffered buses (via 244/245) and other signals, pulling everything up can waste several watts of power on these resistors alone.
Misconception 3: "What should I do with the unused I/O pins on the CPU and FPGA? Just leave them floating for now – we'll deal with it later."
Comment: If unused I/O pins are left floating, even a tiny amount of external interference can cause them to oscillate as input signals. The power consumption of MOS devices largely depends on the toggling frequency of the gates. If you pull them up, each pin will draw microamps of current. The best solution is to configure them as outputs (provided they are not connected to other driven signals externally).
Misconception 4: "This FPGA still has so many gates left unused – I can use them as I please."
Comment: The power consumption of an FPGA is proportional to the number of flip‑flops used and their toggling rates. Therefore, the same FPGA model can consume up to 100 times more power in different circuits or at different times. Minimising the number of high‑speed toggling flip‑flops is the fundamental way to reduce FPGA power consumption.
Misconception 5: "These small chips consume very little power – no need to worry."
Comment: For chips with relatively simple internal structures, power consumption is difficult to determine and is mainly determined by the current on the pins. For example, an ABT16244 with no load draws less than 1 mA, but its specification allows each pin to drive 60 mA into a load (such as a few tens of ohms). At full load, the maximum current can be 60 mA × 16 = 960 mA – of course, that is the supply current, and most of the heat is dissipated in the load itself.
Misconception 6: "This memory has so many control signals. On my board, I only need OE and WE; I'll just tie the chip select to ground. That way, data will come out faster during read operations."
Comment: Most memories consume more than 100 times the power when chip select is active (regardless of OE and WE) than when it is inactive. Therefore, you should use CS to control the chip as much as possible, and keep the CS pulse width as short as allowed by other requirements.
Misconception 7: "All these signals have overshoot! As long as I terminate them properly, it will be eliminated."
Comment: Except for a few specific signals (such as 100BASE‑T and CML), some overshoot is normal. As long as it is not excessive, termination is not always necessary; and even when it is, perfect matching is not required. For example, TTL output impedance is less than 50 Ω, sometimes even 20 Ω. If you use a matching resistor of that size, the current would be very large, making the power consumption unacceptable, and the signal amplitude would become too small to use. Moreover, the output impedance of a signal is often different between high and low states, so perfect matching is impossible anyway. For signals like TTL, LVDS, and 422, it is sufficient to terminate just enough to make the overshoot acceptable.
Misconception 8: "Reducing power consumption is the hardware engineer's job – software has nothing to do with it."
Comment: Hardware is just the stage; software puts on the show. Almost every chip access and every signal toggle on the bus is controlled by software. If the software can reduce the number of external memory accesses (by using register variables, internal cache, etc.), respond to interrupts promptly (interrupts are often active‑low with pull‑up resistors), and adopt other board‑specific measures, it can make a significant contribution to reducing power consumption.
About Maxipcb