power supply fluctuating

Discussion in 'General Electronics Chat' started by skk, Nov 26, 2011.

  1. skk

    Thread Starter Member

    Mar 15, 2011
    31
    0
    I started a new thread for this question. Related threads and the schematic are here:

    http://forum.allaboutcircuits.com/showthread.php?t=61894

    I just breadboarded this thing including all the recommended changes ( thanks guys ) including .33uf / .1uf caps on the 7805 voltage regulator plus decoupling caps on every IC. I also have 100uF caps across VIN/GND and VOUT/GND on the voltage regulator. I am using 51 Ohm resistors on the LEDs and 220 Ohm resistors on the base of each transistor.

    I was building this up slowly testing as I went when I added the 2nd TPIC6C595. I started to get what looked like flicker. I figured something wrong with my multiplex code, but that is not it.

    The problem turns out to be a fluctuation in the voltage when all the LEDs are on at one time. I don't have an oscilloscope or any good tools, but on the voltmeter it looked like the voltage was swinging between 5.0 and 4.8 volts. When the flicker goes away the fluctuation also goes away.

    I can make the fluctuation go away by increasing the supply voltage to the voltage regulator or drastically increasing the capacitance between Vcc and Gnd. I was using a 9V supply, going to 12V makes the problem go away, or adding 3300uF will let me run at 7.5V with no visible flicker.

    What is going on here? Can I counter this problem without increasing the supply voltage or adding such large caps?
     
  2. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    Sure, cut down the current you're allowing to flow through the LEDs - or get a more capable regulator.

    Have you looked at the datasheet for a 7805 regulator?
    Here's one: http://www.fairchildsemi.com/ds/LM/LM7805.pdf
    Here's another: http://www.onsemi.com/pub_link/Collateral/MC7800-D.PDF
    Here's another: http://www.national.com/ds/LM/LM7512C.pdf

    Did you see the great big "1A" at the top of the 1st datasheet?
    Did you see the great big "1.0 A" at the top of the 2nd datasheet?
    That means 1 Ampere.
    The National Semiconductor datasheet requires a bit more reading. On the first page, it mentions output "in excess of 1A", but if you look at the electrical specifications on page 3, you'll see that they only give guaranteed specifications for when the regulator temp is 25°C, and 5mA <= output current <= 1A. So, don't buy the "greater than 1A" stuff.

    Let's see, you have a couple of TPIC6C595 shift registers, and you're using them to sink current from 2x8=16 LEDs, and each LED gets 100mA current when it's on.

    16*100mA = 1.6 Amperes.
    Do you see any problem with trying to get 1.6 Amperes from a regulator that is rated for 1 Ampere?

    [eta]
    I might have caught this before, but you have so many threads going on about this same project, and your schematic is so confusing, that I completely missed it.

    You're really not doing yourself any favors by starting so many threads; one has to hunt all over the place to track the pieces down - and once the pieces are found, there's a big puzzle to unravel.

    If you want to get away from the 7805 problem, use a switching regulator that's capable of at least 2A. If you have overcapacity, you won't exceed your power budget.

    I don't know how many projects I've seen where people try to "get by" with the minimums on their power, and then overrun it by a substantial margin.

    You have a load that varies all over the place, from practically nothing to 1.6A, and seemingly at random. You'll need some good-sized caps to filter out those transients, and a capable regulator to keep up.

    Meanwhile, if you are going to stay with a linear regulator, you are going to wind up dissipating a good bit of power in the regulator instead of the load.

    With all the LEDs on and a 7.5v supply, you are dissipating 7.5v *1.6A = 12 Watts total, and (7.5v-5v) *1.6A = 4 Watts in the regulator, 8 Watts in the LEDs.
    So, your effiency is ~ 66%, and your regulator is mighty toasty.
     
    Last edited: Nov 26, 2011
  3. skk

    Thread Starter Member

    Mar 15, 2011
    31
    0
  4. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    That's what ST says.
    Why should their regulator be good for 500mA more than everyone elses', 7805, when the circuitry is the same for all of them?

    Did you look at all of the specifications?
    Vo; output voltage, Io=5mA to 1A, Vi=8v to 20v; min 4.65v, typ 5v, max 5.35v.
    Delta Vo; load regulation, Io=5mA to 1.5mA, Tj=25°C, 100mV
    Plots start on page 37.

    I don't know. I don't have enough information.

    You have bits and pieces of your circuit scattered over 8 or 10 threads, and I don't know how it's configured.

    Why don't you fix up your schematic so that it's readable, and put in the values that you're actually using at this point, and then post it?
     
  5. skk

    Thread Starter Member

    Mar 15, 2011
    31
    0
    Attached is the updated schematic.

    I found a number of different manufacturers who claim to guarantee 1.5A:

    http://61.222.192.61/mccsemi/up_pdf/MC7805CT(TO-220).pdf
    http://semicon.njr.co.jp/njr/hp/fileDownloadMedia.do?_mediaId=5803
    http://www.st.com/internet/com/TECHNICAL_RESOURCES/TECHNICAL_LITERATURE/DATASHEET/CD00000444.pdf
    http://www.ti.com/lit/ds/symlink/ua7805.pdf
    http://www.national.com/ds/LM/LM340.pdf

    However if this is just the exact same circuit, then this is just marketing, right, or did these guys do something slightly different to get better performance? I am having difficulty in determining the truth from the different datasheets.
     
  6. SgtWookie

    Expert

    Jul 17, 2007
    22,182
    1,728
    You need to look at the electrical specifications, starting with "Output voltage", or "Vo".

    Just about all of them will say that the Vo spec is when Vin is 2.5v or higher than the output voltage, and 5mA <= Iout <= 1.0A.

    The load regulation will vary. If you want to be certain that your design will work, believe the worst-case specifications, along with reviewing the data in the plots over temp. If a datasheet doesn't have the plots, move on to one that does have them.

    78xx series regulators are commodity parts; everyone and their dog makes them. It is a very competitive market, and as a result, Marketing will try to "punch up" the cover page to "show you" how much more wonderful their product is than everyone else's product. That's the hype. Then you look at the actual electrical specifications, and make certain to figure out what the worst case scenario will be over the temperature and full load range.

    Your mileage will vary.
    I can't recall offhand how the 220 Ohm base resistor was calculated offhand, but it likely has to do with the I/O current limits of the 74HC595. I'm not going to spend a half-hour trying to dig that information out of a half-dozen different threads.

    220 Ohms would probably get somewhere around 20mA base current. That's usually good for ~200mA collector current, keeping the transistor saturated, using the standard Ib=Ic/10 formula. As collector current increases, Vce(sat) increases, until the transistor comes out of saturation.

    As Vce increases, power dissipation in the transistor increases. Since Vce is on the rise, that takes away from the voltage available to drop across the load; so power dissipation in the load decreases.

    Your actual results will vary when you get below the Ib=Ic/10 calculation. As Ib remains constant for an increasing Ic, Vce(sat) depends more and more upon the gain of the transistor; which varies from transistor to transistor, and also varies over temp.

    Going back to your input voltage supply - I have no idea what you are using, or what it's rated for. If it is something like an unregulated "wall wart" aka plug supply, it's no wonder you are having difficulties. The output of a wall wart will vary significantly with load current. Also, plug supplies usually have very minimal filtering in the supply itself. You may find that adding a large amount of capacitance on the input side of the regulator helps a lot.

    If you have long wires from the regulator output to various points in your circuit, you will have a good bit of parasitic inductance. You will then need to add bypass capacitors at various points in your circuit, along with having your 0.1uF caps at every IC.

    Since your heavy loads occur at your TPIC6 shift registers and the 2N2907 transistors, you really should have caps from the 2N2907 emitter to the TPIC6 ground. Without that, you will have relatively large surge currents from the regulator output cap through the power rails to the abovementioned components, creating good-sized spikes.
     
Loading...