Ac to dc adapters in series or parallel

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
multiple 220V to 5v,(2A max output) ac to dc adapters connected to the same single phase ac terminals
input voltage is 220v rms.
is it safe to join output in series to obtain 10v /15v/25v etc. ?
 
Last edited:

Bernard

Joined Aug 7, 2008
5,784
I have not seen AC to DC adapters that were not isolated outputs, so see no reason that outputs could not be put in series.
 

AnalogKid

Joined Aug 1, 2013
10,989
The noise will sum as well as the DC voltages, so the 15 V output will need extra decoupling at whatever circuit it is powering. Three identical 5 V adapters will oscillate internally at nearly but not exactly the same frequency, so the total output noise will have both a random and a vector component, and probably some beat frequencies.

ak
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
a related query-
cause ive a bunch of 5v , 2amp adapters lying around
could they be stacked in parallel to get 5v, 4amps out of identical (same model) (5v , 2amp max ) adapters?
and will it be safe if different product , same ratings (5v , 2amp max ) adapters are used in parallel?
 
Last edited:

AnalogKid

Joined Aug 1, 2013
10,989
1. Yes, but...
2. Yes, but...

Since you don't have access to the internal control circuits, the only way to share and add the output currents is with what is called "droop sharing". Basically, a small resistance in series with each output gives the inevitable small voltage differences somewhere to go. This decreases the available output voltage by Ohm's Law, but that is the cost of bending single-output designs into a shared output environment. You also can use a diode (one per adapter) as the droop device, basically diode-ORing them together into the load. With ideal diodes, the adapter with the highest output (even by a few millivolts) will supply 100% of the load until its output sags down to the level of the next highest output. then that adapter's diode will conduct, and the two will share the load current until their outputs sag down to the level of the third-highest output. then that one joins the party, etc. Of course with real-world diodes the transitions are much softer, but that's the overall idea.

As for mixing different current capabilities, if you follow the logic of what is happening above, you will see that a low-current adapter could hit its current limit before a higher-current adapter with a lower output voltage kicks in. How the adapter reacts depends on its internal design. It could fail, or shut down to protect itself, but it won't burst into flame.

ak
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
As for mixing different current capabilities, if you follow the logic of what is happening above, you will see that a low-current adapter could hit its current limit before a higher-current adapter with a lower output voltage kicks in. How the adapter reacts depends on its internal design. It could fail, or shut down to protect itself, but it won't burst into flame.
thatll burn a lot many plugpacks
are there any 5v adapters of higher amp rating(higher than 2amp)?
 

tcmtech

Joined Nov 4, 2013
2,867
Try it and find out. Theoretically ideal Vs real life function tend to be considerably different wher real life tens to work just fine opposed to what theoretically ideal says should happen.

I for one have never had problems with using two or more units of the same output ratings in parallel to get more current capacity or two or more of similar current rating in series to get a higher voltage either.
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
Since you don't have access to the internal control circuits, the only way to share and add the output currents is with what is called "droop sharing". Basically, a small resistance in series with each output gives the inevitable small voltage differences somewhere to go. This decreases the available output voltage by Ohm's Law, but that is the cost of bending single-output designs into a shared output environment. You also can use a diode (one per adapter) as the droop device, basically diode-ORing them together into the load. With ideal diodes, the adapter with the highest output (even by a few millivolts) will supply 100% of the load until its output sags down to the level of the next highest output. then that adapter's diode will conduct, and the two will share the load current until their outputs sag down to the level of the third-highest output. then that one joins the party, etc. Of course with real-world diodes the transitions are much softer, but that's the overall idea.
the power supply already contains the diode
Regulated supplies will be problematic, but not for the lack of diodes to prevent back-flow. The problem will always be that the supply set to the higher voltage will drive as much current as it can to the output; the problem is whether they "fail" gracefully when attempting to exceed their rating; if they current limit and "sag" to meet the other supply, that is OK, but if they continue to drive more current than rated, they will of course, overload.

Balancing resistors are of no value as a regulated supply with a resistor in series becomes - ipso facto - an un-regulated supply
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
I would consider powering part of the circuit from one power supply unit and the other part of the circuit from the other power supply unit. I still need to consider the interface between the two circuits when one power supply is on and the other is not.
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
in series should I have a reverse diode across each supply output to prevent back-driving on overload ?
if the load fails (short circuit or over-current), one of the supplies gives up before the other, then
one supply back-drives the other as with cells in a battery, or cells in a PV-panel.
 

AnalogKid

Joined Aug 1, 2013
10,989
the power supply already contains the diode
Yes, but that diode is inside the internal control loop, so it cannot be used as a ballast impedance, which is what the external diode or small-value resistor is for in a droop-share arrangement. Your question was can it be done, not will it be perfect.
Regulated supplies will be problematic, but not for the lack of diodes to prevent back-flow. The problem will always be that the supply set to the higher voltage will drive as much current as it can to the output; the problem is whether they "fail" gracefully when attempting to exceed their rating; if they current limit and "sag" to meet the other supply, that is OK, but if they continue to drive more current than rated, they will of course, overload.

Balancing resistors are of no value as a regulated supply with a resistor in series becomes - ipso facto - an un-regulated supply
Well - *slightly* unregulated, depending on the values of the ballast impedance and the output impedance of the supply; but that is the cost of any current sharing scheme that does not have a feedback mechanism such as a third-wire current share bus. Again, possible and safe (your questions), but not perfect.

I've deployed droop-share arrangements with multiple 200 W and 300 W supplies. It was a successful trade-off between a little extra heat and a little less efficiency and regulation versus either other supplies that were larger and more expensive or a custom development.

ak
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
i would have to hack the internal control circuits
the solution to paralleling individual supplies would be to tie together the feedback control of two supplies so that the natural internal resistance of the secondary (and in fact, primary) circuits caused them to share evenly in the same manner as unregulated supplies.
 

MrAl

Joined Jun 17, 2014
11,396
the power supply already contains the diode
Regulated supplies will be problematic, but not for the lack of diodes to prevent back-flow. The problem will always be that the supply set to the higher voltage will drive as much current as it can to the output; the problem is whether they "fail" gracefully when attempting to exceed their rating; if they current limit and "sag" to meet the other supply, that is OK, but if they continue to drive more current than rated, they will of course, overload.

Balancing resistors are of no value as a regulated supply with a resistor in series becomes - ipso facto - an un-regulated supply
Hi,

Yes a parallel connection partly depends on what kind of current limiting the single supply pack uses. It may just throttle the voltage down slightly or go into foldback current limit. If it goes into foldback current limit then it shuts down, but if not it will supply the max current at whatever voltage the output happens to be at the time.

However, you seem to be classing the supply as either regulated or unregulated, with no possibility of being pseudo regulated, when that possibility may really exist. In other words, when you connect a resistor in series with the output of one supply source you do ipso facto turn it into a LESS well regulated supply, but it was never a perfect regulator anyway as all voltage regulators have some droop associated with them even without a resistor in series. Thus, when you connect a resistor in series, if it is a small value, you only degrade the regulation performance, it does not all of a sudden turn a perfect regulator into a non regulator. The degree of regulation will go down somewhat, but it depends on the size of the series resistor...some small value of resistance may work good enough and still provide a decent level of voltage regulation.

That said, the size of the resistor(s) is going to depend on the degree of voltage mismatch. If the mismatch is large, then the resistor(s) will have to be larger too. For example with a 0.2v max difference a larger resistor would be required than with a 0.1v max difference in order to ensure the two supply sources put out roughly the same current.

The model for this is fairly simple:
Start with two ideal voltage sources that have a small difference in output, say 0.1v, and then determine what value series resistances would be required to get roughly the same current from both supplies.
There are two basic scenarios that can come up:
1. Two unknown sources.
2. Two known sources.

In #1 it's harder to choose because you cant know which one source is the highest one. In #2 however you KNOW through a measurement which one is higher so you can choose the resistors(s) based on that and get the two better matched. You can do a simulation and see what you can come up with to get the current draw roughly the same from each supply source without loosing too much regulation performance.

It's not going to be perfect, but it may still be good enough. The deciding factor is how different are the two output voltages because that determines the value of each resistor and the value of each resistor determines the degree of voltage regulation that is lost, and too much loss may be too detrimental to the application.
 

AnalogKid

Joined Aug 1, 2013
10,989
i would have to hack the internal control circuits
the solution to paralleling individual supplies would be to tie together the feedback control of two supplies so that the natural internal resistance of the secondary (and in fact, primary) circuits caused them to share evenly in the same manner as unregulated supplies.
Actually, no. That way has too many complications running two fast control loops from a single point that both can completely dominate. Also, it relies on magnetics characteristics that are not ... reliable.

The "standard" way is to derive a signal current based on the load current (in each supply), sum these currents into a medium-resistance analog bus, and then sense that bus against a reference to see if everyone needs to increase or decrease their output currents to meet the load requirement. The internal current sense is scaled to the max output of each supply, so different power-level supplies share proportionally. Astec high density DC/DC converter bricks do this, and both Unitrode (TI) and Linear Tech make controller chips to do the work. SynQor and other newer DC/DC converters use a digital bus to communicate, but I've found that to be much less reliable. Vicor, of course, does something completely different.

ak
 

Thread Starter

Sumit Aich

Joined Dec 3, 2016
100
Yes, but that diode is inside the internal control loop, so it cannot be used as a ballast impedance, which is what the external diode or small-value resistor is for in a droop-share arrangement
I was just pointing out that there is no advantage to using a diode where back-flow protection is not required and a good diode should have minimal actual resistance (and if its "knee" voltage is different, it may even make the disparity in "set" voltages worse)!
 

AnalogKid

Joined Aug 1, 2013
10,989
I was just pointing out that there is no advantage to using a diode where back-flow protection is not required)
Yes, there is. The near-constant voltage characteristic of a diode means it dissipates less power for approximately the same "ballasting" effect as a small fixed resistance. Because the voltage drop is nearly constant, power dissipation increases linearly (almost) with current. For a droop resistor, power increases as the square of the current.

ak
 
Top