How much 12V power is needed?


 
Bryan Mayland

Bryan Mayland

TVWBB Hall of Fame
I did some testing this morning to determine just how much power is needed in our 12V power supply. The wiki states that you need a 12V 1A supply and I'm going to agree based on me using those almost exclusively. However, some users find that their Pi reboots for no apparent reason on a 1A supply but works fine on 1.5A or 2A supplies indicating that we're probably close to the current limit the 1A supplies can produce. I measured with a Pi 3 in wifi client mode streaming data to a web client to make sure the wifi was active. The output was a RotoDamper that uses the old slightly larger 50mm blower (rated at 230mA), not sure if RD3 still use these? All of these images are 1mV = 1mA with the same vertical scale,
however the time scale I adjust based on the frequency of the data. Here's my results:

First we will start with the Pi fully booted and streaming data, with the blower output at 100%. 360mA average, 512mA peak
eaZhJrA.png


Next is Pi + Blower 50% (Voltage Mode). 232mA average, 816mA peak
yUgN2g8.png


Pi + Blower 50% (Pulse Mode). 244mA average, 864mA peak
cWysBHW.png


Finally, I wanted to check the Pi power usage just by itself to see how big any spikes were in the power usage. 157mA average, 176mA peak. I set a trigger for 200mA and after a few minutes there was no events triggered.
wIebHFL.png


Note that our current use is way lower than you'd expect running a Pi 3. In fact, the official Pi power supply is a 2000mA 5.1 supply! Because we're putting 12V in and stepping down to 5V with 80% efficiency, my reported numbers are 1/3rd the 5V current. That means normal Pi 3 power I'm seeing about 471mA - 528mA in use. That leaves at least 450mA of 12V overhead to run a blower / servo with a 1A supply.

Looking through my suppliers I can probably stock a name brand 12V 1A power supply for $11-12, but looking at 1.5A that goes up to $14-17. These are with 1800mm to 6ft cords and are available in US and European plug versions. That's pretty steep compared to 6 to 9 dollars on Amazon. I think it might be worthwhile to offer them but I'm concerned with buying $2000 worth of power supplies if nobody would buy them. For folks who have bought a kit would you have also bought a 12V 1A power supply at the $12 price point? What about 1.5 amp for $17?
 
Last edited:
Hmmmm..... I've got a Kill-a-watt hiding somewhere at home. Might be interesting to put that inline, not sure if it's really sensitive enough to measure that low current at house voltage.

I tried to use a wall wart from a Linksys router, HM 4.2.4, with Tom's barrel damper & squirrel cage fan assembly. It had issues over the long term, but that might also be due to a power jack issue (looks like it's the right size, but a bit of a loose fit.)
 
Bryan, where are you measuring the current, @ the connector? Any way to measure @ the wall wart? For the voltage mode - those spikes are fairly short, and may not even make it back to the wall wart. Peak having ~15% margin to max anyways, and a ton on average overhead. For pulse - the spikes are tiny on the screen due to the scale chosen. But, they don't look like they would really matter either.

I think it comes down to quality of the wall wart. A "real" 1A supply should be able to handle this no problem. The problem is there are very few quality supplies being made and most are junk.

If you can ensure a quality 1A that works for the application for not much more than a crapshoot at Amazon I personally think it'd be worth it. I know I paid up for a 2A when I bought mine for this exact reason.
 
Power supplies are one of the items that pinching pennies on is a fools errand.
IMO finding a "cheap" 1A adapter is just asking for trouble, if your going cheap, get a 2A adapter minimum.

Bryan, rather than stocking power supplies in your store, would it be better to just make an "Approved PS List"?

Another problem with PS is that they come in different sizes and form factors. I personally like the "brick" style that come with laptops as the Wall Wort style has too many drawbacks for me.
 
I actually prefer the "wall wort" style, as I can snug it up under cover with a piece of tupperware if the weather gets inclement. I BBQ year round, and here in Montana, the weather can change on a dime.

I use a 2A Omnihil power adapter and it's proven true. I actually found it by googling "heatermeter" on ebay.

I bought a few as I sell complete HM units, and have had no "druthers" from buyers.
 
Kevin Akers said:
Bryan, where are you measuring the current, @ the connector? Any way to measure @ the wall wart? For the voltage mode - those spikes are fairly short, and may not even make it back to the wall wart. Peak having ~15% margin to max anyways, and a ton on average overhead. For pulse - the spikes are tiny on the screen due to the scale chosen. But, they don't look like they would really matter either.
Click to expand...
I measured it inline halfway between the source and the heatermeter unit. When my big box of power adapters comes tomorrow I'm going to do a whole lot more testing with the same test scenarios and also 50% output while servo is moving and 50% output while servo is trying to move except I am jamming it. Maybe add a second blower. I'll watch the current, 12VDC and 5VDC as well as the 12V or 5V line AC coupled (whichever turns out to be more interesting). I'll see if I can get the Pi to brown out or throttle due to undervoltage. We'll see which performs the best!
 
Well long story short, all of the legit adapters from Mouser had no problems driving a Pi 3 with active wifi transmitting, a USB keyboard, a second USB wifi adapter, a USB flash drive, a servo damper that I jammed, and two 250mA blowers all at the same time. The Pi never even reported an undervoltage condition. All of them produced more stable voltages than the 12V 2A adapter from Amazon that was used in the initial testing. Surprisingly, the 5V line is largely unaffected by any operation but the 12V line does dip on power up, when the blower comes on, and when the servo pulse signal fires. The first two draw over 1.3A for 150-200uS (1.3A is the max current I can measure in my setup with accuracy) and can pull down the 12V line by 1.6V as the adapters struggle with the current rush. The servo pulse interestingly can pull the 12V line down by up to 300mV. In all of these cases, the 5V line stays above 4.930V, well within the 5% tolerance of standard USB power.

Also all of these adapters are Class VI efficiency rated so they are more than 80% efficient. The cords on all the adapters are 24AWG and don't pretend to be anything else, almost all of the unbranded adapters I own have thicker cables, I assume to make them seem beefier. The barrel plugs on all of these adapters are 11mm long which is ~1mm longer than the couple of noname adapters I had lying on the bench.

The top performer for both voltage stability and noise was the Triad Magnetics WSU120-1000, a US-based company (although Taiwanese owned) that produces power adapters for medical equipment among other things. It comes in the wider profile (43x71mm) with an unpolarized plug but unfortunately is only made in US outlet prongs. Coming in a very close second was the SGA12U12-P1J from the well-known power supply manufacturer Mean Well (spoiler: Taiwanese). This one comes in a the slim but deeper profile (27x62mm), roughly the Triad design turned sideways, and is also unpolarized. They also make an EU Plug version. The Mean Well adapters are also 23% more expensive Q100 than the Triad.
 
The difference between the 12V and the 5V is the magic of having the regulator close by. It can "see" the change in load quicker and react accordingly. I think the diode may be involved too though. Its forward voltage changes a good amount with current, doubling from 100mA to 1A. So as the current spikes because the voltage sags the diode actually drops more across it - not very helpful! Try messing with additional capacitors near the input to the board. Adding one (100uf?) outside the diode will help with the noise from the blower. There really isn't any capacity available when the fet turns on to charge the inductor meaning it all has to come from the wall wart. That will probably knock down the ripple significantly. Which should increase stability on the weaker wall warts and reduce overall noise in the system. Secondary to that another cap in parallel with C4 (100uF inside the diode) wouldn't hurt either -another 100uF is probably overkill, ~1-10uF will probably do if you have one around.

I wouldn't worry about the inrush sag at power up. As long as it isn't enough to make the wall wart completely fold it should be fine.
 
I've personally never had a problem with a 1amp adapter other than their longevity (and dropping them in water doesn't help either), then again the MD runs at only 100 mA so I guess it's further from pushing the limits.
 
I have been running an old, I believe either a 1.2 or 1.5amp adapter that I had to open it up and replace one of the diodes that fried, a few times. I just replace it with one of the 4001 diodes and it works perfectly and it can handle a 30cfm blower. It works great as long as I don't short it out again.
 
The power supply testing data was good to see. One thing that sometimes gets overlooked during testing is the background noise of the supply. This component can cause all the issues you might have with the HM not booting correctly or strange readings, mysterious reboots and response issues with fans and dampers. I always test a wall supply prior to use. I look at the background noise component and do not use any supply that has over 150mv of noise. I usually do not even use one over 100mv of noise. Modern semiconductors switch at super low levels compared to the days of providing a .6 vdc level to turn a device on. Unfortunately a device does not care that the switching level is DC or a AC peak, it`s going to switch. So with that said, a good high quality 1A supply should work fine.
My rule is a 2x rule. I use a supply with at least 2A. This usually will give me a better regulating supply and almost always a lower background noise levels that do not go way up when you get to the tested manufactures current limit of the supply. Cooler is always better in electronics so with higher current ratings can come lower temp.
 
So, I run my HeaterMeter with just a damper. No fan. I successfully ran my pit for about 12 hours powered through the RasPi with a 3000mAh USB battery power supply. I didn't let it go until it died because I was done cooking and weather was crap. 12 hours is a good chunk of time. Next time I run it I will see just how long it will last and still move the servo.

This was a HM 4.3 with RasPi Zero W. Running my FlatDamper attached to a pit barrel cooker.
 
Just posting some thoughts here, I guess as a bit of a sounding board...
Does anyone see any issues with the thinking that I outline below?

John Bostwick has assembled a HM for me and it is currently in transit to Australia.
I was thinking that I may have had a suitable power supply/adaptor in my collection, but I don't.
So I'm looking to buy one on Ebay.
I think due to my situation the "Brick" type of power supply is going to be more suitable, rather than the "Wall Wart" types.

It's quite difficult to tell what sort of quality I'm going to be getting from Ebay, but the brick types seem to be more prolific on the Aussie Ebay.

I can get a Brick type rated at 12V 4, 5, or even 6 A for about $20.00 delivered.

So that is what I think I will get.
My thinking is that a power supply that is rated at 5 amps will cruise if there is only a couple of amps (or less) being drawn, even if it struggled to live up to its 5 Amps rating.
I guess there is still a potential to be concerned about how accurately it will supply the rated voltage,bearing in mind that I have no idea as to the quality (good or bad) of these but I guess all I can do is suck it and see.
The only test equipment I have available is a volt meter, but I can put it under load and check it out with that.
Not ideal, but the best I can do I guess.
Interestingly the connector that the Heatermeter uses appears to be not very common on adaptors that are available in Australia.
I think all of the ones mentioned above are supplied with a 5.5 X 2.5mm connector.
My local electronics store has a 5.5 X 2.5mm to 5.5 X 2.1mm Adapter, for just a couple of dollars so I plan to use one of those.
Can anyone see any flaws in my reasoning?
Any suggestions as to a better way to do it?
My plan is to use a stainless steel dogs bowl to connect the RD3 to my Webber Kettle.
I'm going to remove the original bottom Damper and fit my Dogs bowl over all 3 holes.
I'm going to have my local engineering shop weld a short piece of stainless steel tube into the side of the dogs bowl.
The RD3 will slide on to that.
I'll use use a stainless steel screw to fasten the bowl to the base of the kettle through the hole that the damper was originally held on with.
I'm not sure if I am able to post links... but if I can I'm thinking of something like this....
https://flic.kr/p/VdQSmF
If I find I need to seal my dogs bowl where it comes in contact with the the kettle, I'll use heat resistant silicone to make a gasket.
Not a lot I can do apart from plan untill the Heatermeter arrives in the post, to ensure I get the size correct for connector pipe.
But if I don't have anyone pass on any reasons why I shouldn't use one of those power supplies, then I will order one of them.
 
Bryan Mayland said:
Next is Pi + Blower 50% (Voltage Mode). 232mA average, 816mA peak


Pi + Blower 50% (Pulse Mode). 244mA average, 864mA peak
Click to expand...

Bryan Mayland said:
The first two draw over 1.3A for 150-200uS (1.3A is the max current I can measure in my setup with accuracy) and can pull down the 12V line by 1.6V as the adapters struggle with the current rush. The servo pulse interestingly can pull the 12V line down by up to 300mV.
Click to expand...


Great research!

If I'm reading all of this correctly it seems that a spike to 864mA or a draw of 1.3A with the use of my cheap 12V 1A power supply would have starved my Pi3 and caused reboots like I was seeing. Assuming my cheapo power supply was providing 80% efficiency 800mA would not have covered the 864mA spikes you were seeing.

If I'm looking at this incorrectly or missing something let me know. In any case using a 12V 3A power supply solved my reboot problem and it was still compact enough for me to fit into my project/controller box!

I know I'm running my unit to control an SSR for an electric smoker so it is not the most common setup, but a once size fits all power supply approach or an approved list of power supplies may make life easier. I look forward to seeing what you recommend in the future.
 
You must log in or register to reply here.

 

Back
Top