Hi all!
Did someone tried to use (or is using) two modules like this in parallel mode?
They are rated at 5V 1A DC...and it is said they can be connected in parallel to further reduce noise, increase the current output and split heat.
I need about 2A current so for me would be best option since I have 2 of them unused.
But I have read it is not so easy to do it and in many cases it is not good.
So there are couple of methods like:
1. Running the output of the two regulators through a diode (one each), and then connect the outputs of the diodes together.This method has voltage drop of the diodes to contend with and the regulators are not well balanced.
2.Better method is "active load balancing". It still uses the OR-ing diodes, but this time a small circuit dynamically adjusts the output voltage of the regulators to keep the load balanced. In this case, the regulators can be kept very close to 50% of the total load. And thus, the max current can be double what any one regulator can achieve by itself. Doing this kind of circuit for me is rocket science since I am an amateur and new to DIY.
3.Method is quite simple and I have tried couple of times and it works without a problem with these modules.
I just simply wire up the two regulators in parallel.
But I have read that without knowing a great deal about the regulators and their behavior in could and up catastrophically.
The only thing I have noticed is that one of modules always gets hot on heatsink. Not to crazy hot (I can hold my finger on it for 2-3 sec).
So at the and I came back to 5V 2A switching power supply.
I assume this is happening because regulators are not "balanced" for supplying exactly 50% of the current and in this situation is more like one regulator takes 80% or more and the other one the rest. Right?
So...is it safe to use it like 3. method or should I do it somehow different?
Did someone tried to use (or is using) two modules like this in parallel mode?
They are rated at 5V 1A DC...and it is said they can be connected in parallel to further reduce noise, increase the current output and split heat.
I need about 2A current so for me would be best option since I have 2 of them unused.
But I have read it is not so easy to do it and in many cases it is not good.
So there are couple of methods like:
1. Running the output of the two regulators through a diode (one each), and then connect the outputs of the diodes together.This method has voltage drop of the diodes to contend with and the regulators are not well balanced.
2.Better method is "active load balancing". It still uses the OR-ing diodes, but this time a small circuit dynamically adjusts the output voltage of the regulators to keep the load balanced. In this case, the regulators can be kept very close to 50% of the total load. And thus, the max current can be double what any one regulator can achieve by itself. Doing this kind of circuit for me is rocket science since I am an amateur and new to DIY.
3.Method is quite simple and I have tried couple of times and it works without a problem with these modules.
I just simply wire up the two regulators in parallel.
But I have read that without knowing a great deal about the regulators and their behavior in could and up catastrophically.
The only thing I have noticed is that one of modules always gets hot on heatsink. Not to crazy hot (I can hold my finger on it for 2-3 sec).
So at the and I came back to 5V 2A switching power supply.
I assume this is happening because regulators are not "balanced" for supplying exactly 50% of the current and in this situation is more like one regulator takes 80% or more and the other one the rest. Right?
So...is it safe to use it like 3. method or should I do it somehow different?
Hi JonSnell Electronic.
I have messed up something while attaching pictures.
Hope I will do it correctly now.
I use two transformers (30W dual 8V) to power these modules.Power Module Linear LT3042 for XMOS
I have messed up something while attaching pictures.
Hope I will do it correctly now.
I use two transformers (30W dual 8V) to power these modules.Power Module Linear LT3042 for XMOS
Attachments
Paralleling regulated supplies will result in the one with the highest output voltage tending to try and do all the work. It only takes a few millivolts difference for that to occur.
Assuming the outputs are pretty close voltage wise then you could add low value series resistors to each output (few tenths of an ohm) to help balance the loading.
Assuming the outputs are pretty close voltage wise then you could add low value series resistors to each output (few tenths of an ohm) to help balance the loading.
At the and if I could find really good and stable finished 5V (min 2A) PSU board I would buy it to finish my story.
I have tried already with this one rated for 5V https://www.ebay.com/itm/Assembled-...ted-power-supply-board-LPS-PSU-/263151685639o but it does not work for me.
So if there are any suggestions to some other board...I would appreciate it.
Thank you all.
I have tried already with this one rated for 5V https://www.ebay.com/itm/Assembled-...ted-power-supply-board-LPS-PSU-/263151685639o but it does not work for me.
So if there are any suggestions to some other board...I would appreciate it.
Thank you all.
This is a bigger (known) problem than you would assume.
Worst case you risk that one supply sources the other(s) and that can harm those suddenly acting as load and overheat the supply acting as the sole source. Most likely, the single supply trying to pull the whole lot cannot supply the load correctly on top. A disaster.
Typically you have a poor load-sharing between more parallel-coupled supplies such that one supply heats more than the other(s) and the whole system has a poor reliability, poor efficiency and eventually insufficient capacity.
You risk that the supplies start oscillating because the regulation loop of one supply interferes with that of another supply.
That was a lot of misery but believe me, at the end of this posting I will hint a solution.
As Mooly explains, voltage regulated supplies (voltage sources) are difficult to put in parallel because they each INSIST on the common output voltage being the exact voltage they are set for. As they are all set for a slightly different output voltage, you can imagine the havoc.
With such power supplies you can use OR-ing diodes and at least they will not source one another but the load sharing can be expected to be (very) poor. You can use resistors in series with the outputs or implement electronic droop-characteristics (intentional positive output impedance) but the load sharing will still not be really good.
And now to a successful approach: Use a number of supplies designed as voltage controlled current sources. Currents can be added without trouble. If we design the voltage controlled current sources such that the output current is controlled by the control voltage in the same way for all supplies, good load-sharing is ensured. A single feedback loop provides the same control voltage to a plurality of supplies that in response supply a certain output current that is summed with other similar output currents. That is the general solution.
Now you are thinking "f... (Scottish expression), I cannot redesign the boards I already have". In your case the solution is to put the transformers in parallel BUT on the rectified side of individual bridge rectifiers. As the transformers are identical, the winding resistance will make the transformer loading rather equal and the buffer capacitors you can connect in parallel. In short, you put the two boards in parallel at the level of the buffer capacitors.
Then you have two regulators you cannot put in parallel without all the trouble described above. What to do? Abandon one regulator (eventually remove the main components) and load the other regulator to some 15% of its capacity (resistive loading). Then, take a solid NPN power transistor (preferably NOT Darlington) and connect the collector to the voltage at the buffer capacitors, the base to the output of the voltage regulator (with the resistor load) and use the emitter as the new output (with a 1uF connected to ground) This is a single buffered linear regulator design and it is MUCH easier than trying to connect two regulators in parallel. The buffer transistor will leave a 0.8V drop which you compensate for on the regulator.
NB: The power transistor will need a heatsink.
Worst case you risk that one supply sources the other(s) and that can harm those suddenly acting as load and overheat the supply acting as the sole source. Most likely, the single supply trying to pull the whole lot cannot supply the load correctly on top. A disaster.
Typically you have a poor load-sharing between more parallel-coupled supplies such that one supply heats more than the other(s) and the whole system has a poor reliability, poor efficiency and eventually insufficient capacity.
You risk that the supplies start oscillating because the regulation loop of one supply interferes with that of another supply.
That was a lot of misery but believe me, at the end of this posting I will hint a solution.
As Mooly explains, voltage regulated supplies (voltage sources) are difficult to put in parallel because they each INSIST on the common output voltage being the exact voltage they are set for. As they are all set for a slightly different output voltage, you can imagine the havoc.
With such power supplies you can use OR-ing diodes and at least they will not source one another but the load sharing can be expected to be (very) poor. You can use resistors in series with the outputs or implement electronic droop-characteristics (intentional positive output impedance) but the load sharing will still not be really good.
And now to a successful approach: Use a number of supplies designed as voltage controlled current sources. Currents can be added without trouble. If we design the voltage controlled current sources such that the output current is controlled by the control voltage in the same way for all supplies, good load-sharing is ensured. A single feedback loop provides the same control voltage to a plurality of supplies that in response supply a certain output current that is summed with other similar output currents. That is the general solution.
Now you are thinking "f... (Scottish expression), I cannot redesign the boards I already have". In your case the solution is to put the transformers in parallel BUT on the rectified side of individual bridge rectifiers. As the transformers are identical, the winding resistance will make the transformer loading rather equal and the buffer capacitors you can connect in parallel. In short, you put the two boards in parallel at the level of the buffer capacitors.
Then you have two regulators you cannot put in parallel without all the trouble described above. What to do? Abandon one regulator (eventually remove the main components) and load the other regulator to some 15% of its capacity (resistive loading). Then, take a solid NPN power transistor (preferably NOT Darlington) and connect the collector to the voltage at the buffer capacitors, the base to the output of the voltage regulator (with the resistor load) and use the emitter as the new output (with a 1uF connected to ground) This is a single buffered linear regulator design and it is MUCH easier than trying to connect two regulators in parallel. The buffer transistor will leave a 0.8V drop which you compensate for on the regulator.
NB: The power transistor will need a heatsink.
Last edited:
Paralleling power supplies can be a minefield.
Unless the transformer secondaries were on the same transformer I wouldn't parallel them. If the voltage on one is higher than the other there is pretty much a short circuit there.
I wouldnt parallel two regulators either due to them being slightly different voltage.
Answer is to get the right transformer and right regulator for the job.
Unless the transformer secondaries were on the same transformer I wouldn't parallel them. If the voltage on one is higher than the other there is pretty much a short circuit there.
I wouldnt parallel two regulators either due to them being slightly different voltage.
Answer is to get the right transformer and right regulator for the job.
True Nigel, not trivial at all.
Two identical transformers you can connect in parallel after the rectification. The individual bridge rectifiers ensure that one transformer cannot source the other. The rectified transformer voltage will drop some 10% up to nominal loading. That will let the other transformer take a good share of the job. This way you hardly get outside of 80%/120% sharing.
One transformer and one regulator is the right way - unless you already have bought two with insufficient capacity.
Two identical transformers you can connect in parallel after the rectification. The individual bridge rectifiers ensure that one transformer cannot source the other. The rectified transformer voltage will drop some 10% up to nominal loading. That will let the other transformer take a good share of the job. This way you hardly get outside of 80%/120% sharing.
One transformer and one regulator is the right way - unless you already have bought two with insufficient capacity.
I got caught out a few years ago with a SMPS transformer.
I paralleled the secondary windings due to skin effect.
The supply was pulling amps without a load !
I hadn't got the secondary windings windings exactly the same length.
So there was pretty much a short on output !
Once more very true. SMPS transformers have much less windings and accordingly much less winding resistance to reconcile two output voltages.
In the old days we only relied on bifilar wound windings to be connected in parallel on our SMPS.
In the old days we only relied on bifilar wound windings to be connected in parallel on our SMPS.
Last edited:
Thank you all very much for your information. Much appreciated!
I really liked FauxFrench detailed explanation.
Much more things are clear to me now...even though my knowledge could not let me understand really 100% what FauxFrench was suggesting me to do. 🙂
@gigigirl,
I was reading before about that STUDER900 "based"Linear power supply.
Here in P/S section is tread about MJE15034 TL072 Regulator based on STUDER 900...If you were thinking about that!?
I have read that one is rated for 5V 2A and some people have trouble feeding something current hungry.
Apparently circuit uses a current limiter set to 1.5A.
That is why I did not take this board into serious consideration. And that one that is on sale on audiophonics is also rated max output current 2A.
But now when you said it can do 4A...hmm?
They all look about the same on picture.
I will need to contact seller about more information and maybe the best solution would be to go for that finished board.
Again...many thanks to all of you!
Cheers
I really liked FauxFrench detailed explanation.
Much more things are clear to me now...even though my knowledge could not let me understand really 100% what FauxFrench was suggesting me to do. 🙂
@gigigirl,
I was reading before about that STUDER900 "based"Linear power supply.
Here in P/S section is tread about MJE15034 TL072 Regulator based on STUDER 900...If you were thinking about that!?
I have read that one is rated for 5V 2A and some people have trouble feeding something current hungry.
Apparently circuit uses a current limiter set to 1.5A.
That is why I did not take this board into serious consideration. And that one that is on sale on audiophonics is also rated max output current 2A.
But now when you said it can do 4A...hmm?
They all look about the same on picture.
I will need to contact seller about more information and maybe the best solution would be to go for that finished board.
Again...many thanks to all of you!
Cheers
It's easy enough to change the current limiting resistor, but I very much doubt if 4A would be possible. The eBay resellers won't know much about it, maybe Audiophonics would know more.
Last edited:
This is a smps but add some filtering and it might work fine.
2A 5V is the typical larger USB charger used for iPads and such.
BOLWEO AC 100-240V to DC 5V 2A Power Supply Adapter, 10W Adapter for LED Strip Lights,Audio/Video, Wireless Router,DC Connector Jack 5.5mmx2.1mm https://www.amazon.com/dp/B0719GY29M/ref=cm_sw_r_cp_api_i_-k1vCbCG6E76R
2A 5V is the typical larger USB charger used for iPads and such.
BOLWEO AC 100-240V to DC 5V 2A Power Supply Adapter, 10W Adapter for LED Strip Lights,Audio/Video, Wireless Router,DC Connector Jack 5.5mmx2.1mm https://www.amazon.com/dp/B0719GY29M/ref=cm_sw_r_cp_api_i_-k1vCbCG6E76R
What you already have is what I have included above the dashed line. What I suggest you to turn the existing two supply circuits into is found below the dashed line. The "regulator"-boxes include all components making up the regulator circuits, not only the regulator ICs.
R should have a value such that the regulator performs well but without too much idle loss and such that the base current of the power transistor at full load is well below the current in the resistor R.
The sketch shows my suggestion for how to turn two identical positive power supplies with linear regulators into a combined circuit with higher current capacity. For identical negative power supplies, a PNP power transistor is used instead. If the output current may be less than some 100mA, a dummy resistor should be connected between the output and ground for bias of the power transistor.
Attachments
Hi, And thank you all again!
I won't have time to work on this problem these days because I won't be at home this week.
All this trouble I got myself (and all of you...sorry)into it is that I need nice and stable way to power Rpi 3 B+. I use it just for audio with Moode Player taking music from SSD(with its own power) over one USB input.
What I use now for powering is Xiaomi's phone USB charger. It is smps rated 5V 2A and it works perfectly but I want it to be linear and to use another (free) secondary from transformer and finally pack all components in box and close it so it looks nice like normal Hi-fi component.
Since I bought new Rpi version I did not know it will consume more than 1A and this is where my problems started...because models before B+ work 1A supply with no problem.
So I could degrade to older Pi models but then I don't have some nice features that make my life easier...better heat dissipation,built in WiFi...etc. What I have find on net is that B+ probably won't go much more then 1.5 A in really high stress and when it is stable it should not go over 1A even with more CPU stress. My only problem is when Rpi starts to boot it consumes much more current and that is the point when 1A is not enough. When it sometimes manage to boot even with 1A and starts audio player it works. So because I can use my Pi 5V 2A smps I really think if I can find nice and stable board that does not go hot on heatsink and that can give me 5V with 2.5 A I could go live with that. No need for 3A or more I think.
I need to say it again I really appreciate FauxFrench effort but to do something like that I would need to find someone else to do it for me. My tools,technical skills and knowledge is just not on that level.
Cheers guys!
I won't have time to work on this problem these days because I won't be at home this week.
All this trouble I got myself (and all of you...sorry)into it is that I need nice and stable way to power Rpi 3 B+. I use it just for audio with Moode Player taking music from SSD(with its own power) over one USB input.
What I use now for powering is Xiaomi's phone USB charger. It is smps rated 5V 2A and it works perfectly but I want it to be linear and to use another (free) secondary from transformer and finally pack all components in box and close it so it looks nice like normal Hi-fi component.
Since I bought new Rpi version I did not know it will consume more than 1A and this is where my problems started...because models before B+ work 1A supply with no problem.
So I could degrade to older Pi models but then I don't have some nice features that make my life easier...better heat dissipation,built in WiFi...etc. What I have find on net is that B+ probably won't go much more then 1.5 A in really high stress and when it is stable it should not go over 1A even with more CPU stress. My only problem is when Rpi starts to boot it consumes much more current and that is the point when 1A is not enough. When it sometimes manage to boot even with 1A and starts audio player it works. So because I can use my Pi 5V 2A smps I really think if I can find nice and stable board that does not go hot on heatsink and that can give me 5V with 2.5 A I could go live with that. No need for 3A or more I think.
I need to say it again I really appreciate FauxFrench effort but to do something like that I would need to find someone else to do it for me. My tools,technical skills and knowledge is just not on that level.
Cheers guys!
Attachments
I know what you mean PRR. 😀
I didn't express myself correctly.
I already had in the beginning Rpi + Kali reclocker using both with 5V 2.5A smps no problem.
When I bought better DAC with I2S input I have decided to buy I2S to HDMI board and isolator board also and build myself better supply.
I knew exactly that 2.5A was recommended for RPi, 3A for Kali..etc.
So because I am total amateur and was first time going into something like DIY electronics I have asked the guy I got the DAC from (we don't live in same country and we speak different language)what components he uses in his DIY audio streamer and how much "A" I will need for each board.
I was told that 1A will be OK for every board including Rpi...only Kali reclocker will need more current.
Kali actually...if not powering anything else consumes less then 0.5A...about 100mA. And that was the board I was most afraid about high current demand.
The guy and his other friends have build more than 10 similar like streamers that all work without a problem with 1A supply.
Ok...they don't use Allo stuff...they are using Ian's boards and much more boards in it...but the catch was that they all build them much earlier...and on older version of Rpi...not the 3 B+. That is why for their Rpi's 5V 1A is sufficient.
At that time I did't know what version of RPi they have and he probably did't know that 3 B+ will consume much more current.
All I knew then was that Rpi is rated for 2.5 A...but 1A is sufficient.
Yes it can...but not latest version. 😀
Thank you for your input PRR!
Cheers
Last edited:
- Status
- Not open for further replies.