Voltage regulator trouble when powering a wireless router from a battery

Thread Starter

Zurn

Joined Mar 4, 2019
117
Hi everyone,

I'm trying to power a tp-link wireless router from a 15V battery system using a LM7805CT 5V voltage regulator and its has me a bit confused.

First off I should say that it's not the only device connected to the battery system. There's also a few other microcontrollers. The battery system itself is a fancy uC powered voltage regulator that has four 15V batteries hooked up, and always outputs around 15V. Without the router, the whole system draws just under about 300 mA.

As for the router, it works fine when I power it through a power supply, drawing a peak of 200 mA - I can connect to it on my computer and do all my business. It also works *sometimes* with the voltage regulator, but only when it's the only thing connected to the battery. Adding the rest of the devices means the router will power up, draw the correct amount of current, but not connect to my computer. Sometimes, however, with the router connected, the current from batteries becomes significantly larger, peaking at about 750 mA.

The original 5V power supply for the router is rated for 1A. All of my components should be able to handle that. I have a voltage regulator hooked up the a substantially sized heat sink as well. I've also tried adding bigger caps than the 7805 datasheet suggests (which are .33 uF and .1uF) - 10 uF on both the input and output.

Is there something I'm missing?

I'll try and get a schematic together ASAP to further illustrate my circuit.
 

MrSoftware

Joined Oct 29, 2013
2,188
Is the regulator getting hot? I'm guessing you've either got a sagging voltage issue, or a noise issue. Can you put a scope on the regulator output and see if you're getting any noise or voltage sags?
 

Thread Starter

Zurn

Joined Mar 4, 2019
117
The regulator is definitely getting very hot, hence the heat sink which seems to have resolved that issue.

Hooking it up to a scope doesn't seem to indicate much. The signal looks fairly clean - no noticeable spikes or dips, unless I zoom WAY in (like 10mv/div), and even then it's pretty consistent.
 
Last edited:

Thread Starter

Zurn

Joined Mar 4, 2019
117
You know what, I swapped the 10 uF input cap for a 100 uF one... and it's working (for now).

If anyone cares to lay down an explanation it'd be most appreciated, otherwise I'll count my blessings and see if I can reach some understanding on my own time.
 

MrSoftware

Joined Oct 29, 2013
2,188
You're probably getting a voltage dip or noise that happens very quickly, too quickly to see on the scope unless you set the trigger to catch it. And it may happen sporadically, not continually. That's my guess anyway without more info. Try setting trigger points on the scope slightly above and below your steady state voltage and see if you can catch it. Also measure the current being used, if you're anywhere near the max for the regulator, and/or if the heatsink isn't resolving the heat issue completely, consider adding a second regulator in parallel. This article will help you with multiple parallel regulators.
 

bertus

Joined Apr 5, 2008
22,270
Hello,

An other option is using an opamp to keep tracking the "main" regulator:

LM317_tracking_with_opamp.png

It is taken from the 2007 version of the LM317 datasheet.

Bertus
 

Attachments

Thread Starter

Zurn

Joined Mar 4, 2019
117
Unsurprisingly when I returned and fired things up this morning it didn't work again. I did find that voltage drop that was happening, probably right when it tries to connect.

Been messing with 7805s quite a bit today... I think its safe to conclude a buck converter is the way to go at this point. Thanks for the input, everyone.
 

MrSoftware

Joined Oct 29, 2013
2,188
There's nothing wrong with linear regulators, they're even preferred when you want clean power. The biggest drawback is that they're inefficient, and the bigger the difference between the input and output voltage, the less the efficiency. So if you're on 15V battery power and getting 5V output, the switcher will definitely give you longer battery life. Just for completeness, make sure it's not the supply side (battery) dropping voltage.

That said; note that different regulators react at different speeds. If that one reacts slow to a sudden load then this can also cause a voltage dip. Putting a large cap on the output side is one way to address it, or switching to a faster regulator is another. Before you switch to a new regulator, find the response time for your current regulator and make sure the new one is faster, or you may end up right back in the same situation.
 

Thread Starter

Zurn

Joined Mar 4, 2019
117
Just for completeness, make sure it's not the supply side (battery) dropping voltage.
Very good point on this - the battery IS dropping the voltage.

Not entirely sure how to address this. Is it at ripple filter I want to add to the regulator input? A negative clipper? Would the same thing happen with a switch regulator or is that what they're specifically made to handle?
 
Last edited:

MrSoftware

Joined Oct 29, 2013
2,188
How low is the battery dropping, and is it dropping below the 5V that you need? Trying to make up for a weak battery with a boost circuit is not the ideal way to go. If a beefier battery isn't an option the adding capacitance would be my next best suggestion. If you have to add a lot, then consider also adding an inrush current limiter (ICL) keep things under control when you first connect it to the battery.
 

ci139

Joined Jul 11, 2016
1,898
15V to 5V puts (worst case) 10V on 7805 . . . × 300mA = 3W (if i recall right the chip can take 1.5 or 2W . . . 1.5W Max.)
the std. measure is series drop resistor and/or parallel resistor if the output current draw is sufficient for the least ...
anyway that smokes the 2/3-rds of the power into the air (for emergency quick alternative it could do - but this also wears off the battery sooner)

. . . the less the output requirement the less voltage head room there should be for 7805 . . . about 2V for 300mA + an unfinished design option to compare against (adding functionality reduces efficiency . . . unless it's a fuse at the OUTP)
 

Attachments

Last edited:

Thread Starter

Zurn

Joined Mar 4, 2019
117
How low is the battery dropping, and is it dropping below the 5V that you need? Trying to make up for a weak battery with a boost circuit is not the ideal way to go. If a beefier battery isn't an option the adding capacitance would be my next best suggestion. If you have to add a lot, then consider also adding an inrush current limiter (ICL) keep things under control when you first connect it to the battery.
It's not dropping below 5V, more like down to 10V. I'll try adding more capacitance (a capacitance multiplier?) or the ICL (though not until monday!).
 

MrSoftware

Joined Oct 29, 2013
2,188
The poor regulator is getting a double whammy, a sudden big draw on the output side combined with a big sag on the input side. A capacitance multiplier is good for noise filtering, I'm doubtful it would help here where you're actually starving for electrons. It's possible a better or faster regulator could handle that drop better. Since even with the input sag you're still 5v above the output, you might go ahead and try the switcher like @dendad referenced above. Worst case you'll get better efficiency and still need to add some capacitance on the 5V side, best case you'll get better efficiency and it will handle the sag better.
 

bloguetronica

Joined Apr 27, 2007
1,541
Hi,

I think you should try either a DC-DC converter solution (or SMPS), or a mixed solution with a DC-DC converter as a pre-regulator, followed by an LDO. As ci139 mentioned, the dissipated power is just too big for the regulator. It you wish to use a linear solution, while keeping the same current ability, you have to drop the voltage at the regulator input. Enter the DC-DC converter.

What I would suggest: use a 6.5 to 8V DC-DC converter, followed by a 5V LDO. An LM1117 will do for the LDO part. You can find DC-DC converter modules on eBay. Or just use a fixed 5V DC-DC converter alone, since you are powering digital equipment anyway.

Kind regards, Samuel Lourenço
 

Thread Starter

Zurn

Joined Mar 4, 2019
117
Sadly neither of the DC-DC converters I ordered are doing the trick (this one and this one). With a 100 uF cap on the output, the line is certainly a lot smoother than w/the linear regulators and the router usually connects but unfortunately it can't maintain the connection. I'll try and track down a 5V LM1117 next (I have plenty of 3.3 V ones).

Tricky stuff but this has been very educational so far.
 

ci139

Joined Jul 11, 2016
1,898
"The excellent performance of these converters is based on minimal output capacitance, ... blah blah . . .
. . .
Use only enough output capacitance to achieve the energy storage and noise suppression you need and no more. "
 

Thread Starter

Zurn

Joined Mar 4, 2019
117
"The excellent performance of these converters is based on minimal output capacitance, ... blah blah . . .
. . .
Use only enough output capacitance to achieve the energy storage and noise suppression you need and no more. "
Ya, I read the datasheets more closely and only have 10 uF on the output now. Have tried adding more but it doesn't seem to make a difference. 10 uF on the in and 10 uF on the out seems to work best, but still only maintains connection for a few minutes max before cutting out.
 

MrSoftware

Joined Oct 29, 2013
2,188
Try replacing the battery with your bench power supply see if the problem persists. This will help verify that the battery is the issue. Scope the input and output side of the regulator to determine if there is still a significant voltage sag. The input should not sag at all with the bench supply, and ideally the output side will not either. If the input does not sag, but the output still sags, then that's a sign your regulator is not up to the job.

10uF reasonable for filtering, but is not a lot for energy storage. The proper fix is likely a more capable battery, but just for giggles you can try adding a pile of capacitance to the 5V side just to see what happens. Try a 1kuF or 10kuF electrolytic. Though you probably want to add an inrush current limiter. If nothing else it will be educational. ;)
 

crutschow

Joined Mar 14, 2008
34,282
You have to be careful adding capacitance on the output of a switching regulator as it can upset the control loop and lead to instability/oscillation at the output.
 
Top