Voltage Regulator Whine on QuickStart Rev B
Tintin
Posts: 37
I recently got a QuickStart board, and when power is applied to it, it makes this annoying high pitched sound. I've worked with other QuickStart boards, but never with a squealing unit. The sound is emitted no matter how I power the board, either via USB or Vin. I've also noticed that the sound changes frequency as the current draw changes, such as when turning on the blue onboard leds. Interestingly, the culprit seems to be the (3.3V) voltage regulator: I've found that putting a 4.7uf electrolytic capacitor between Vdd and ground gets rid of the sound.
Perhaps some of you can shed some light on this: Is my board less reliable because it makes this sound? Is my solution with the capacitor viable long term?
Perhaps some of you can shed some light on this: Is my board less reliable because it makes this sound? Is my solution with the capacitor viable long term?
Comments
It is common practice to accept the wide range of values in capacitors and to adjust your specification so you get the desire performance.
In large production runs, it is just a lottery as to who will get an occasional outlier. Providing capacitors with tighter specifications just makes everyone pay more for the end product.
If you desire an exchange, contact Parallax. If not, adding your own capacitor eliminated the 'sweet spot'. I suspect the board would work fine, but be a perpetual annoyance.
I guess it depends on your costing methodology.
Most serious designers I know, design to include the margins, and will happily pay another 0.5c to get a much more stable Cap.
They also avoid any design that even hints at the word 'lottery' to the user.
jmg is right. It's a carefully controlled lottery.
When you are making anything, say a capacitor, there will be variations in the resulting capacitance value from device to device, from batch to batch. There will be variations in the secondary parameters like inductance, series resistance, and so on. It's impossible to be 100% accurate.
The values you actually make will be spread out around the value you are trying to make. There is a distribution of values. As first approximation we might assume this is the "normal distribution" or bell curve (google it).
That distribution could be wide, values +/-10 or 20 percent. Or it might be narrow +/-0.1 percent.
When a designer puts such a component in his circuit he has two choices, design the circuit to be tolerant of wildly different values of component, or specify a very tight tolerance for the component.
Either way there is a chance that a particular instance of his design will get that capacitor, or whatever, that is out of working range.
You can spec tolerances, and allow variation as much as you like but there will always be units that fail, could be 10 percent off the production line, could be 1%, could be 0.1% could be a lot less than that now a days.
The question then is how much do you want to spend in reducing the failure rate, or the return rate from customers?
A serious designer of items made in huge quantities will not just happily pay another 0.5c to get a more stable cap. Over a run of a million units that's 50,000 dollars off the potential profit. Such a decision requires balancing the cost of that cap vs any potential improvement in reducing failure rate. If an engineer assesses that the cheaper cap is good enough then he has saved his whole years salary in a few hours!
A serious designer of one off and low volume products may well just throw in the the 5 cents. It's not worth his time to even think about. Just play it safe and move on.