Excessive Power Supply Noise/Ripple

Goeytex

Senior Member
Earlier this morning I was experiencing some weird problems while testing some Picaxe I2C code, so after about 30 minutes or so I decided to check for noise./ripple on the +V rail of the breadboard. With the scope set for AC coupling, it reported a whopping 300mv of random and intermittent high frequency noise.

The only devices on the breadboard were a Picaxe 20X2 & a 14M2 connected via the I2C lines and a 74HC14 Hex inverter for inverting a signal. All of the ICs have a 100nf decoupling capacitor. There are 100nf and a 100uf filter caps where the supply comes into the breadboard. The Power Supply is a BK Precision 1621A bench supply.

So I turned off the nearby florescent lamp and everything else nearby but to no avail. I then disconnected the supply from the board and scoped the output with a 50 ohm load attached. It was fine, less than 5mv of noise/ ripple. So then I reconnected the supply and pulled the power jumpers from all of the ICs. The noise was gone. So I connected power back to the ICs one at a time and the noise reappeared when I supplied power to the 74HC14 Hex Inverter.

The problem? I was only using one of the 6 inverters and failed to ground the unused inputs. All five of the unused inverters were randomly going into oscillation. I was embarrassed at myself because I've been doing this stuff for > 40 years and know better than to leave an unused input on a logic chip floating.

I posted this to save others the same problem and emphasize the importance of following the recommendations of a device datasheet, which in this case clearly states that unused inputs MUST always be tied to ether ground or Vcc. This cost me 45 minutes of head scratching. It could cost someone without a scope a lot more time as well as the unnecessary frustration that come with weird problems. When frustrated enough, the temptation is to blame the "stupid Picaxe" or even unnecessarily abandon a project. Many frustrations can be avoided by thoroughly reading and following the datasheets of the devices we are using.

FWIW, a similar "recommendation" is given in Picaxe Manual 1 where it states:

"It is recommended, but not essential, to tie unused inputs low via a 10k resistor."

Admittedly , I do not generally follow this recommendation. I think this is more about saving a bit of power in sleep mode than anything else, but it may be something to keep in mind if experiencing weird / seemingly unexplainable problems.
 

srnet

Senior Member
Its a good reminder for sure.

Although I prefer to set unused pins to outputs and low, which saves the resistor ......
 
Last edited:

manuka

Senior Member
A great posting - a logic IC mindset of "it's just an inverter/NAND/NOR/XOR" may be so prevalent that one fails to appreciate their mischief potential!
 

vk6bgn

New Member
I had a similar problem with an older PICAXE project a few years ago. The little black wall wart switching power supply I was using was so "dirty" that the PICAXE was do weird thing. Process of elimination, replaced the power supply with a larger one twice the size with a small transformer in it.... all good now.


"HamRadioAddict"
<(^_~)>
 

John West

Senior Member
I too have been doing this stuff for > 40 years. Greater than 50, actually. Thanks for the reminder about floating inputs, Goeytex. I needed it, on account of the fact that I've been doing this stuff so long that I've grown old and am not as sharp as I was when I'd only been doing it for 20 years. That was prime time. The rest is history.
 

erco

Senior Member
Great info, Goeytex. I knew from my old Forrest Mims books about grounding the inputs of unused comparators, but not inverters. I guess when in doubt, ground 'em all.
 
Top