Results 1 to 10 of 10

Thread: Pixel Blades, Wire Gauge, and You!

  1. #1

    Default Pixel Blades, Wire Gauge, and You!

    Here is a very convenient tool to figure out if your wiring is appropriately sized in your high draw pixel blade saber. I've book marked a typical application.
    https://www.calculator.net/voltage-d...s=10&x=95&y=19

    Using Watt's law you can take these numbers and sort out the initial* power dissipation by your wiring.
    For example, using 6" of 28awg wire, at 3.7v, with a 12 amp load we would see a drop of about .78V which is a 21.05% drop. Our guideline is that we should not exceed 3% and we should really aim for less than that.

    P=IV, P = 7 x .78, P = 5.46 Watts

    We can calculate the initial* resistance imparted by the wiring using Ohm's Law.
    V=IR, 0.78v = 7a x R, R=0.111ohm

    As you will see below, this will put your wire somewhere between "melting" and "catching fire." The wire is insulated so the heat doesn't dissipate, and as the temperature increases the resistance of the wire will go up. This creates a feedback effect wherein as the wire gets hotter the resistance goes up and it gets even hotter until an equilibrium is reached with the heat it's able to dissipate, or the jacket of your wire melts and/or catches a flame. This is why I noted these as the "initial*" values.


    Strydur performed experiments to demonstrate this effect:
    Quote Originally Posted by Strydur
    I did a quick test using a power supply and a load generator. Keep in mind this is all tested with the wire we sell and could be different for the same thickness wire in a different brand. You would also get different results with longer or shorter wire, etc..

    3" long 24 gauge
    10A - Ok
    15A - Hot
    20A - Melt Casing

    3" long 26 gauge
    10A - Ok
    15A - Melt casing
    20A - Catch Fire

    3" long 28 gauge
    10A - Melt casing
    15A - Catch Fire
    source: http://forums.thecustomsabershop.com...l=1#post276869


    In Tim's example of 3" 28AWG at 10A & 15A the power dissipation was: 10A x .32Vd = 3.2 Watts; 15A x .49Vd = 7.35watts

    Keep in mind that if this was packed inside a saber, rather than a bench experiment, the effects would be worse. In any case a melting jacket will lead to a great sadness, and possible injury. Longer wires will also impart an increased resistance.

    There will be complicating factors like the contact resistance of your connector, the actual charge level of your battery, and the color of your blade, but with these examples and the mathematics I hope you have been convinced to appropriately size your hilt wiring for your pixel blade.
    Last edited by jbkuma; 01-18-2019 at 08:54 AM.

  2. #2

    Default

    As far as using multiple conductors to compensate for the drop:

    The calculator does include a calculation for drop over parallel conductors. In general my advice is to design your wiring to the saber's capabilities rather than your expected configuration.

    In the example I linked, it shows a fairly typical use scenario of 7A and about 5" of wire. With 22AWG wire you would have a voltage drop of about .13V or 3.5%. According to the guidelines for this calculator that is actually already a bit out of spec. For 2x 28AWG it would be about .26V per wire. I used multiple 28 AWG wires in the past, but I have long since replaced the wiring in all of my sabers with 22.

    The situation is actually a lot more complicated than the simply saying "it dissipates about 2.2 watts." While this is true, the dissipation is over the length of the wire. There are applications where this resistance is actually counted upon, such as some string lights which use a long conductor on one side that loops back to drop the voltage into the range of the LEDs and balance the paths. In other situations the resistance of the leads is problematic not because of heat, but because the drop degrades a relatively low power data signal. This is why we use switches, routers, repeaters, etc. for our data communication lines. (you may already realize this, but it's worth mentioning for those who don't)

    In this case we are using short wires though, and the overall dissipation can be treated as effectively as we would a resistor. Thinner wires will also have less mass and surface area for that heat. Multiple wires will have more surface area to dissipate heat. We will see losses in other parts of the circuit as well. Tinned vs bare copper (you do NOT want to use aluminum wire!), type of solder, quality of joints, jacket material, etc will also factor in, some things more than others. These calculated values are mostly useful describe the situation and use as a guideline.

    I would say, based on experience and experimentation, anything more than a calculated 1 Watt will not be a good idea. The guideline of a calculated 3% drop stated on the calculator page is probably a good mark to follow.

  3. #3

    Default

    I follow JBKuma's advice.

    "Mistakes are our greatest teacher."

  4. #4

    Default

    In my entry to the most recent contest I used 20 ga for the mains to and from the neopixel strips and accent neopixels, and 26 ga wire everywhere else (speaker wire, momentary switch, data line). The strips are Adafruit minis so if I recall the theoretical draw for each pixel is 35mA full white, x120 x2 = 8.4 A for my setup. The data lines come to a common point at the PCB but each strip has its own +/- wires to the PCB.

    It's tricky using 20 ga, especially connecting the PCB connector to the strips - 1" length 20 ga wire between PCB and strip becomes very stiff when soldered and the version 1 PCBs were not robust enough (pads from 2 PCBs were torn off by sparring 8 year-olds).
    Last edited by Kooky Mawnster; 01-25-2019 at 06:06 PM.

  5. #5

    Default

    At what voltage was the test below done? A single 18650 is about 4.2 V fully charged.

    --------------------------------------------------

    Quote Originally Posted by Strydur
    I did a quick test using a power supply and a load generator. Keep in mind this is all tested with the wire we sell and could be different for the same thickness wire in a different brand. You would also get different results with longer or shorter wire, etc..

    3" long 24 gauge
    10A - Ok
    15A - Hot
    20A - Melt Casing

    3" long 26 gauge
    10A - Ok
    15A - Melt casing
    20A - Catch Fire

    3" long 28 gauge
    10A - Melt casing
    15A - Catch Fire

  6. #6

    Default

    At 3.7v the pixels should still be pulling around 47ma each at full white. 35ma might be with the brightness capped in firmware (FX-SaberOS details to cap at 230/255) or could be the the result of losses and battery sag.

    You can read a full characterization report here: https://github.com/Protonerd/DIYino/...on_report1.pdf

    You can design to a lower standard, but as long as you saber has the ability to go into those higher regimes there is potential for disaster.

  7. #7

    Default

    Here is a quote from one of the pages of Adafruit under the heading "Sipping power with Neopixels" (not sure if I can post the link on this "forum associated with a store"). It is where I got the theoretical 35mA figure for mini-Neopixels. The 60mA is for the regular size Neopixel.

    Of course, for a definitive answer for my setup I could set fdrive=(1023,1023,1023) as a color and measure the current and voltage on the mains to get a real world figure.....

    ----------------------------------

    Some Finer Points of NeoPixels

    - The “60 milliamps” rule of thumb is just that…a rule of thumb, not hard science. Actual peak use is slightly less. But for napkin calculations and mental math, 60 is much easier than fifty-whatever. (The mini NeoPixels on Circuit Playground and “Mini Skinny” NeoPixel strips are more like 35 mA each, max. And RGBW NeoPixels, with a fourth LED element inside, may draw closer to 80 mA each.)

    - Also, 60 (or 35, or 80) milliamps is a worst case. When animating and mixing colors, current draw will be less. Sometimes much less. That’s what this guide is all about. Even when set to color 0 (no LED elements lit), the driver logic inside each NeoPixel still uses a tiny bit of current. It’s under 1 milliamp per pixel…but with lots of pixels this could add up.

  8. #8

    Default

    The information I referenced is all from either scientifically performed tests or the datasheets. This is not just random speculation. You are free to design for less than peak performance, the point is you should understand that if you operate outside of your intended parameters there may be consequences. When you design things to fail, they will.

    As you noted the idle current is not insignificant. In FX-SaberOS we have a feature that shuts down power to the blade to keep it from draining the battery.

    I build pixel related circuits and I have tested used and tested many more formats than just the common 5050 and 3535 ("mini skinny"). The 3535 pixels have the same rating as the 5050. I linked to a detailed report on the current draw from pixels at various voltages. Please review that for accurate information rather than speculating and giving false impressions.

  9. #9

    Default

    @jbkuma

    DISCLAIMER: I am not a person trying to make money selling complete sabers and parts.

    I read that article, and other research before I starting making my saber. I'd be careful accusing me of speculating and giving false impressions in your previous post. I've seen specs ranging from 38mA to 56.25mA per die.

    That being said, since you are experienced, may I ask of you: What do YOU consider a "safe" saber? In the sabers you have made, what are some hard numbers regarding:

    - type, current draw and number of neopixels
    - wiring and other electronics considerations
    - power source
    - intended uses

    that you consider safe?


    I still believe the 35mA "rough" figure for the strips I have. It would explain why this Panasonic NCR type 18650 rated 4A continuous I have would run the blade color, but the protection circuit would trip on FoC. The saber otherwise runs fine on the battery I have rated 10A continuous. I'm somewhere south of 4A for blade colors, but somewhere south (or slightly north) of 10A for full white, considering the ability for the batteries to handle an intermittent spike in current above the continuous rating, the values of which I don't know. This is with blade color at about 75% duty cycle for normal use, and as an experiment I tried running a blade color of drive=1023,1023,1023 with the 10A battery and it ran fine, though I didn't test for heat generation over time.

    Unless the 10A battery's protection circuit is not tripping.....but I would be seeing/smelling smoke by now if that were true.....

    I think the point is that all of these sabers being made, neopixel or otherwise, have the potential to be dangerous, but you run them (as part of the design) so they are not. No?
    Last edited by Kooky Mawnster; 01-27-2019 at 01:55 PM.

  10. #10
    Owner of the Custom Saber shop Strydur's Avatar
    Join Date
    Oct 2005
    Location
    USA
    Posts
    3,520

    Default

    IIRC I ran those test at 3.7v

    I would consider a safe saber one that works. Can you do the research and make it "safer" sure but if it works..it works. All of this while based on logic is not figured out based on easy calculations. Some plugs for instance will say they have a max amperage but if you go over that and the saber works fine then obviously they can handle more amperage than stated. I would take any data sheet with a grain of salt and if you have the proper equipment test it yourself.
    Tim
    The Custom Saber Shop

Bookmarks

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •