Shop OBEX P1 Docs P2 Docs Learn Events
Since its always the power supply... — Parallax Forums

Since its always the power supply...

prof_brainoprof_braino Posts: 4,313
edited 2013-06-20 07:35 in General Discussion
An EE said, "if you can't find the problem, it's usually the power supply."

Recently I've had issues with variance from USB, alkaline, LiPo, etc; just a bit bit off can make a problem. What can be done?

One solution is bite the bullet and make a proper power supply, but sometimes cost and skill are an issue. Like now with me.

Is there simple way to monitor 3.3 and 5 volts? I'm thinking a separate module so its reusable, we only need it when testing.

Test 3.3 and 5 volt, when these are in spec, the LED is green, when low, red, when hi, yellow. Maybe red lo, green hi, yellow ok, so we know the LED is working too.

What would be the simplest, cheapest way to do this?

Thanks!

Comments

  • CuriousOneCuriousOne Posts: 931
    edited 2013-06-18 08:52
    A simple TL431 IC, resistor divider and led will do the magic I believe.
  • PliersPliers Posts: 280
    edited 2013-06-18 08:56
    I would guess that the EE was referring to noise causing issues with the circuit. Not something as simple as low or high voltage. An led would not show you the micro second drop outs or surges.
    The monitoring circuit t should detect un acceptable noise, surges and drop outs. I'm a fan of the 74LS123 for detecting fast pulses.
  • prof_brainoprof_braino Posts: 4,313
    edited 2013-06-18 19:20
    Pliers wrote: »
    Not something as simple as low or high voltage.

    In this case, yes, just the simple voltage was the problem. 4 * AA batteries vs. USB: USB is 5.0V; alkaline is 6.0v; NiMH is 4.8V; a USB phone charger can be all over the place. The alkaline batteries were making at least one component WAY over volted, and pushed another way out of spec such that it no longer functioned.

    I will look at the 74LS123 as I have not delved into noise, surge and drop os yet, but I guess its about time to start.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2013-06-19 02:55
    I use a good 5VDC wall wart, and a tiny LDO 3.3 VDC linear regulator to convert to 3.3V.

    I strongly dislike powering from a USB port... has anyone figured that out yet? Batteries are an unnecessary expense and all of them have problems with fitting into the +5V and +3.3V world of chips.

    I only use batteries when I have to test isolation from ground. I use an 8VDC gel cell that is 9AH as I can just recharge with a 9VDC wall wart, and 12VDC is too high.. while 6VDC is too low .. and both need actual chargers.
  • Dr_AculaDr_Acula Posts: 5,484
    edited 2013-06-19 06:25
    Recently I've had issues with variance from USB, alkaline, LiPo, etc; just a bit bit off can make a problem. What can be done?

    1) Alkaline - they go flat, errors are unpredictable.
    2) Lipo - they go flat - errors are unpredictable.
    3) USB - overload it and it will drop out and/or the PC will shut down the supply briefly - errors are unpredictable.

    I use a 1 amp old-school wall wart. In fact, some years ago, I got hundreds from a Commodore 16 http://en.wikipedia.org/wiki/Commodore_16 for the princely sum of $1 each (erco would be proud!), specs 1A, 9V@1A (13V unloaded) and they have been serving me for several decades.

    For years I used the 7805 and equivalent TO220 3V3 linear regulators. Also 1A, so nicely matched to the wall wart. No problems at all.

    More recently, I have been using the LM2575 for 5V and the LM2574 for 3V3 - switchers, total cost ends up slightly cheaper as no heatsink needed, inductor, capacitor, fast diode. Works every time. Total cost maybe $4. 1A is more than enough for a propeller. If I want to power servos and motors, I run those of a separate 4x NiMH supply.

    When debugging, need to eliminate all variables and power supply variables are the most unpredictable :)
  • PliersPliers Posts: 280
    edited 2013-06-19 06:53
    My thoughts were, An EE should not be stumped by an under voltage power supply. It is the first thing checked when trouble shooting.
    A power supply with intermittent drop outs or spikes is difficult to spot. Unless the power supply is monitored and the data recorded. A storage scope set to trigger once during an event is one way to catch such an intermittent problem.

    For your problem, under voltage power supply....
    I would use a voltage comparator to sense the low voltage condition and a flip flop to make the LED to stay on even if the condition goes a way.
  • LoopyBytelooseLoopyByteloose Posts: 12,537
    edited 2013-06-19 07:06
    I'd really use prefer using a 0-10 VDC analog voltmeter over building a complex tangle of LEDs. Very simple, very reliable, and very retro.

    Hmm. I think I will actually do that! Maybe a pair to be used together - one for +5 and another for +3.3.
  • prof_brainoprof_braino Posts: 4,313
    edited 2013-06-19 07:24
    I guess I need more detail on the application :)

    This a cheapest possible robot.
    It must not require a tether during normal operation.
    It must allow power over USB (no batteries) during test.
    The uC requires 3.3v (Vdd)
    Some peripherals require 3.3v
    Some peripherals require 5.0v
    Some peripherals can run off whatever the battery pack is providing 4.5 to 12volts (Vin)

    The EE had no problems with the power supply, he made a custom design specifically for this application. I switched to batteries, and used what ever I had on hand (to see if it would cause issues, as we expect users to do just this). The EE had problems figuring out via email what was different about my rig, and why it did not work when his works perfectly.

    So, in the meantime, I'm going to set up a bench power supply, and three volt meters, and an amp meter, and check every point with every combination (until it looks like it works). I'm thinking I might try to make a test fixture rather than tie up all my equipment.
  • ercoerco Posts: 20,256
    edited 2013-06-19 09:45
    @Prof_braino:

    I get cheap. I know cheap. I love cheap.

    That said, cheapest is rarely the the best nor simplest, it often leads to workarounds & compromises. IIRC, this is for your gradeschool robot project. Whether you're talking about power supplies, sensors, or processors, sometimes a few extra cents to save some grief is worth it, especially given the brief attention span of today's youts. :)

    My two cents.
  • skylightskylight Posts: 1,915
    edited 2013-06-19 11:12
    erco wrote: »

    I get cheap. I know cheap. I love cheap.
    Reminds me of a song....

    http://www.youtube.com/watch?v=HSNSTerj2Kc
  • PliersPliers Posts: 280
    edited 2013-06-19 14:12
    Voltage Monitor.jpg


    Adjust R1 to set the sensitivity as needed.
    During normal operation Q1 is on. This keeps Q2 off.
    If the voltage drops below the set point, Q1 turns off allowing Q2 to come on and light the LED (D1).
    969 x 804 - 109K
  • CuriousOneCuriousOne Posts: 931
    edited 2013-06-19 22:10
    Another solution may be to build SEPIC power supply, which will deliver required output even when input is below or above it. Both texas and linear have apropriate ICs. You'll need a specific coil, too. But Coilcraft is also providing free samples.
  • prof_brainoprof_braino Posts: 4,313
    edited 2013-06-20 07:35
    Ok, I definitely have lots of options if I get to a custom board. at post 13, my brain began to melt. I'm going to have to draft a real EE and assign this as a task.
Sign In or Register to comment.