Wall-wart characteristics and measurements

[George Giblet]

We get a lot of posts here with people asking why the voltage they measure on their wall-wart is a lot higher than the nominal voltage on the label.  I've done a shirt load of wall-wart measurements and created some summary statistics so rough conclusions can be drawn about wall-warts. 

I've limited my tests to: wall-warts that are DC output, unregulated and filtered. They are constructed of a transformer, rectifier and filter capacitor.  This is the most common type. (Other types are: - DC output, regulated and unfiltered. - DC output, regulated.  - AC output.  Always unregulated.)

The first thing about this type of wall-wart is that it is unregulated.  When you measure the DC output voltage without a load the output voltage is usually much higher than the nominal voltage on the label.  As you increase the load current the output voltage drops.  When the load current matches the rated current on the label the voltage drops to the nominal voltage on the label .  In most cases the voltage turns out lower.   So where does the voltage on the label come into it?   It's only a rough ball-park figure, under full load the voltage will be close - that's all you can say.  Most people are measuring the output with no load and that's why they get the higher voltage.

The main reason the output voltage drops is the transformer.  Real transformers are'nt ideal.  Transformers have copper windings and the windings have resistance.  As the load current increases the winding resistance causes the the output voltage drop (basically I*R).

No amount of theory and explanation will tell you how these things behave in detail.   What I've done is made a stack of measurements on wall-warts.  The idea is this will expose the manufactures secret recipe for the nominal voltage and rated current. The table below shows these measurements.



The following conclusions can be made:
- The voltage with no load is always higher than the nominal voltage on the label.  It can range from 113% to 151% of the nominal voltage.  Most will be around 130%.
- When the maximum load is placed on the output the voltage is typically lower than the nomincal voltage.  It can range from 60%(!) to 100% of the nominal voltage.  Most units will be around 85% of the nominal voltage, ie. low 15%.
- So just how much current can we draw from these things to get the nominal voltage?  This varies widely from a minute 17% upto 93% of rated load current.  For most units you can only draw 56% of the rated current before the voltage drops to the nominal voltage.
- It appears the filter capacitor is 1000uF capacitor, independent of the current rating of the wall-wart.
- The ripple is roughly 1V p-p at 200mA.

As a final comment.  All the voltages I've given are the average, as measured by a DC multimeter.   When there is ripple present the voltage will be above the average by Vrip/2 and below the average Vrip/2.  If you are designing regulators you need to make sure the average voltage - Vip/2 is above the dropout voltage of the regulator.

The number given in this post do not apply to home made designs using transformers.

[Paul Perry (Frostwave)]

This is a very worthwhile exercise.
It shows why Roland etc. are always keen for people to use an approved Boss power supply - and why we can (usually) get away without doing so.

[brett]

Thanks for that.  It's an excellent piece of work.

I've got a couple of switchmode PSs in the shed, so if I've got the right resistors (or light bulbs) I'll do some measurements on those.

PS In my experience the VA rating for small transformers is approximately equal to the area of the core, which is closely related to weight.  There is little difference in the quality of the iron or windings.  If in doubt, choose a heavier transformer.

[George Giblet]

Transformer manufacturers roughly work to a recipe.  Using weight is fine provided you don't change the general core shape (such as going low profile or Toroidal), or the change operating temperature.  With those constraints weight will work as good as any - others are outer surface area (more or less, the size) and winding resistance.  I've done silly mass samplings of this type of thing in the past too!

As a means of getting a more indicative figure I've taken a new spin on the figures.   A low voltages the wall-warts start to be have differently so I decided to only use voltages >= 7.5VDC.  Also the PS499 unit appears to be a little odd ball , this is an oldish unit and it looks like the manufacturer tried to drop the voltage on purpose to get around the high non-load voltage issue.  With these constraints the summary figures become:

For >= 7.5V
         
   Voc/Vnom           131%
   V@Irated/Vnom   90%  ; 10% lower than nominal.
   I@Vnom/Irated   70%
   Voc/V@Irated     141%
   Capacitor           1000uF independent of rating
   Vripple              approximately 1Vp-p @ 200mA load

I'm happy to use these as ball parks with the assumption that there might be odball units out there (you can't expect to cover an uncontrolled world with a summary).  The V@Irated/Vnom figure is a little higher than before and I@Vnom/Irated is somewhat higher.

[R.G.]

As the resident theory-wonk, I guess I disagree a little bit with the statement that:

No amount of theory and explanation will tell you how these things behave in detail. 

On the contrary, the exact operation of transformer-rectifier-filter caps is pretty easy to model and work out. These kinds of circuits are one of the places that theory agrees pretty well with observation. It's just that you have to know something about transformers and ...  yep, Ohm's law... to work through it.

On the other hand, it does require some understanding, and not everyone will want to work through it, so for those folks, rules of thumb are useful.

This is not to say this isn't a valiant effort. It's good to get some real world measurements to get a feel for what happens.

Transformers have some hidden truths inside them. For instance, the thing that limits the power you can pull through one is how hot you let it get, and that is in turn limited by the insulation, not the iron and copper. Another one is that the transformer's power handling ability at a given temperature is proportional to the Area Product, the product of the copper winding window area and the magnetic cross section of the core.

Several of these mystical things combine with the square-cube law of proportions to force transformers to obey a size law: the bigger the transformer, the less it sags within its load rating. The transformer makers wrap this up in the "regulation" spec of their iron.

The regulation of a transformer (that is, the no-load to full-load voltage change) gets worse with small size. A 1VA power transformer is about the smallest size I've seen, and these things are the ones that have the horrible 40-50% regulation numbers. The primary wire has to be so fine and so many turns to withstand AC power line voltages that the resistance of the primary eats all the available voltage and causes horrible sagging.

As volt-amp rating gets bigger, regulation improves. You can get about 10% in doubled-fist size iron. You can get 5% in transformers you can hardly pick up. You can get better in transformers you can move only with a truck.

On top of that, rectifier-capacitor setups force high pulse currents from the transformer. A full wave bridge may force pulse currents as much as 10X the DC average current to flow in the transformer. That current goes through the transformer resistances and the voltage sags even more than you think the RMS or DC currents would make it. The bigger the capacitor, the worse this is. And the bigger the cap, the lower the ripple , so to fairly compare rectifer-capacitor circuits you have to compare ripple as a % of the output voltage and capitance per ampere of output.

The reason I'm muddying up something you started getting a handle on is that you need to add a few columns. You need to add the rated VA of the transformer, or at least the primary and secondary resistances, which let you calculate a lot of this. You need to add capacitance per ampere of output current, and ripple %.

It's easy to take the same transformer and make the output voltage sag more by putting a bigger capacitor on the filter if you are using a regulation limited transformer.

As my friend told me today, after all, this ain't rocket surgery... 

[George Giblet]

> On the contrary, the exact operation of transformer-rectifier-filter caps is pretty easy to model and work out.

I was coming at this from a different angle!

Sure, it is possible to *analyse* a transformer+rectifier+cap.  I've actually done a lot in depth analysis on capacitor filters and have derived formulas which are much more accurate than the text book stuff.  I've also got formulas for designing transformers, trading copper loss, iron losses and heating.  But...

My angle is, all the analysis in the world  won't tell you how wall-wart manufactures are making them and their general behaviour as a blackbox.  You can only analyse what is given to you - the problem is we don't know what wall-wart manufactures are giving us!

For example if you take a 12VDC 300mA wall-wart.   You know the regulation is bad, you know there will be ripple, but just how high will the unloaded voltage be and just how far will the voltage drop when you load it down?   If you were designing one from scratch then *you* have to decide the unloaded voltage and the voltage under load.  If I was an ultra-cautious engineer and I wanted to put 12VDC on the label I might decide that the voltage should be 12VDC at full load, but then my manager will say well the unloaded voltage is going to be too high and we don't want to get complaints about a 12VDC unit putting out 16VDC and blowing up people's equipment.

When wall-warts first came out a 12VDC wall-wart was designed using a 12.6Vrms transformer and the transformer was designed not to overheat at full load.  The result was the usually 16VDC to 18VDC unloaded voltage and quite often a severe drop in voltage at full load - in the old days wall-warts have low pwer ratings and the small transformers have inherently poor regulation.

The idea of the measurement/statistical approach was to observe the behaviour of these things from outside and derive the design decisions made by the engineers in this area.  It seems that the 130% of nominal voltage at no load and 90% of nominal voltage at full load is a good ball park (a least for the smaller units, larger units might be different due to the inherently better transformer regulation).   It is of course an over simplification there are no standards manufactures are bound to so we still could get anything!

[Paul Perry (Frostwave)]

While it is true that in the real world, power through a transformer is limited by one of the coils overheating & melting, it is also true that if the coil DIDN'T melt, then it would be the saturation of the core by the peak magnetic field that would limit the throughput.
That's where the power goes - it comes in thru the primary, magnetises the core back & forward, and as the core responds, the secondary generates the voltage.
If you ever look at avionics power supplies, you will be suprised how small the power supply transformers are for the VA rating - that's because they run at 400 Hz instead of 50 or 60, so they get to transfer about 7 times as much energy, per weight of core.
And the same goes in spades, for the ultrasonic frequencies that switchmode supplies run at.
Certainly, very small mains power transformers (an inch cube or so) are often weak points in equipment. (like the old EH stuff). I'm happier seeing toroids.