ORING power supplies CIRCUIT problem

[ghassan]

Hello,

I have been working on this project for few month. I stopped on many problems and i really need assistance with this project .

- first problem     : if the AND logic gate has any time delay then the mosfet driver goes on even if the control does not turn pmosfet.

- second problem : according to the Lt-spice simulation file , aux power supply may be always on and if the aux ps is on and main power supply goes off or on then i can see the label net which is called "aux_out" gets reversed power even if "M4" is OFF as an ideal diode .. Why it is happening ?.
- third problem    : I am able to give priority to main power supply to power on the output even if the aux power supply is on as well by using XOR Logic gate , but unable to success with the cross conduction inhibition for a voltage that ranges between 21 VDC reaching 30 VDC .
- fourth problem : I was unable to speedup Q1 and Q6 the switching time due to the maximum absolute rating of the LOGIC Gate 25mA or 50mA!. C1 and C4 are taking about 800mA of current from LOGIC according to the simulation.

I've been asking these question somewhere else , but no one answers me . I really hope that i get some help.

I appreciate your help .

Thank you

[hermit]

First problem.  I see no schematic. 

[ghassan]

Would you like to see my full schematic ? . I can upload the schematic , but its a little big that's why i was trying to lower the design as soon as i can .

[ghassan]

Please see the attached full schematic design .. It is not completed yet.

[hermit]

You expect us to buy/download programs till we find one that works?  I have no idea what program created that file.  It is common to post a pgn file of the schematic here.

[ghassan]

Please try to check "ORING_PMosfet.asc" with Spice Simulator. You will see "A1" and "A2" AND LOGIC GATE . Open these logic components with the "Component Attribute Editor" and you will see that these values are inside the logic gate (Vhigh=5 td=100n trise=0n tfall=0n) . Then simulate the schematic and zoom the plot V(main_out) between 200.00us till 200.07us . I wanted to know if the problem from the schematic or from the simulator program .  I am talking about first problem .

[ghassan]

I'm sorry I didn't mention that i am using LTSPICE simulator.

[ghassan]

I uploaded the small design

[ghassan]

I updated the picture with comparison between main power supply and "main_out" net .

[Zero999]

I uploaded the small design

M3 and M4 are just being used as diodes, so why not replace them with proper diodes? They're much cheaper and have better characteristics too. Just using a MOSFET for its body diode is pointless.

[ghassan]

I uploaded the small design

M3 and M4 are just being used as diodes, so why not replace them with proper diodes? They're much cheaper and have better characteristics too. Just using a MOSFET for its body diode is pointless.

I chose to use Mosfets to have a very low power dissipation and very low drop voltage . I also wish to use this project to carry about 10 Amps @24vdc .

Regarding M3 , it should turn On when its power supply is allowed to be passed to the output , therefore it turns on and if the aux power supply is  turning off or turned on . Then M3 will act as an ideal diode. I kept M3 and M4 turned off just to lower the schematic to be able to share it with you . I already made the project , but i still need some help to finish it and i really hope you help me with it .

Would you see the whole project in an ASC (LTSPICE ) file ?.

I will try to explain the project as possible as i can :
- each power supply has a window comparator to check if the input at its power supply has a low voltage input or high voltage connected both comparators to an AND Logic gate . The AND Logic gate is responsible to check if there's no over-voltage and no under-voltage then AND logic gate will connect to the XOR Gate .

My main problem is when all these components such Logic Gates and comparators , we will have few nano seconds as a delay so if the main power supply turns on at a specific time and the p channel mosfet driivers has a delay or no delay , then the pmosfet turns on by itself and then turn it off again because its driver told it to stay off even though power mosfet turns on for few nano seconds then turns off and then turn on when its driver order to turn on!.

I would like to know what is happening here ?. Is this problem from the program or the power mosfet works like that .. I even tried to simulate new circuit .. it had the same problem as always .

 

[ghassan]

I simplified the problem with a simple schematic . I hope it is explained well to you.

[Zero999]

Yes, please post the full .asc file, including the part which turns on M3 and M4.

[Jay_Diddy_B]

Hi,

Are looking to OR two supplies? In this configuration the highest voltage supply will always deliver the load current.
For this situation look at a solution based on the LTC4355.



Link: http://www.linear.com/product/LTC4355




If you want to assign priority to one supply over another, independent of voltage. You might want to favour a wall supply over a battery even if the wall supply is less than the battery. If this is case look at the LTC4418.



Link: http://www.linear.com/product/LTC4418

Regards,

Jay_Diddy_B

[ghassan]

I wish if i can buy a chip , but unfortunately i am not interested to buy a chip . I thought i would build it by myself.

My real and important question is really easy .
This is a very simple schematic that shows a simple power mosfet that takes an input voltage of 24vdc and output its input when the voltage gate is about 12vdc which means it should turn on. If i decided to delay the gate alittle and tell  it to wait extra time even if the power supply comes online!!! .


Please see this one . What do you think this might be caused ?. Is it from the LTSPICE PROGRAM or it does work like this way in reality as well!.



this is the full schematic :


Full schematic in LTSPICE simulation file :

[Jay_Diddy_B]

Hi,

Have a look at these models:



And the modeling results:



And then look at the 'bonus' model:
The MOSFET has been replaced by a capacitor, representing Coss



The results are:



Coss comes from this graph on the datasheet:



You have no output capacitor, If you do, Coss and the output capacitor form a (capacitive) divider.

In addition the dv/dt, 24V in 10ns is unrealistic, this is contributing the output spike.

So the answer is that LTspice is correct, if you have the conditions that you are modelling.

I have attached my model.

Regards,

Jay_Diddy_B

[ghassan]

Thank you so much @Jay_Diddy_B. It was a good explaination.

I understand that a mosfet has internally capacitance and if we add at its drain a capacitor in parallel then it will form a capacitive divider .

Is there any trick that we can eliminate this capacity or discharge it quickly without adding more latency . As i understand is by adding more capacitance , there will be more latency to the switching time . 

I added "C7" with a value of 10nF at node "main_out" to minimize the problem and not to keep the latency lowered as possible as i can. It works 


Results:

[ghassan]

Moving to the second problem :

According to the below schematic , node "aux power" may be always on and if the Auxiliary Power Supply is on while the Main Power Supply goes Off or On then i can see the label net which is called "aux_out" gets reversed power even if "M4" is OFF as an ideal diode . I also replaced "M4" with a diode and it had the same results .

Before adding "R11" to "aux_out": Common Drains which is called "aux_out" should switch between 0V and 30V .



After adding "R11" with a value 1 kilo ohm:



Why this is happening ?.

I appreciate your help

[ghassan]

Does anyone know any body who can finish my project ?.

I really need some assistance to fix all problems.

I will pay to anyone and if you are not interested , then would suggest a website or some engineers who are willing to get paid for this project ?


I appreciate your assistance.

[IanMacdonald]

I might be misunderstanding what you're trying to do, but seems like an incredibly complicated way of performing a simple task.

If the switching speed is insufficient, then the probable reason is charge storage on the base of Q1/Q6, assuming these saturate.  Placing a schottky diode from b to c such that the collector cannot fall below base voltage is the standard remedy for this. Q4/Q7 seem to be there to 'crowbar' Q1/Q6 off, for this reason. Preventing saturation will be a better approach. Or, use a mosfet.

Though, I have to ask, if you are going to have diodes in the lines, why do you need the rest anyway? Since the output voltage cannot be that accurate with series diodes anyway, that seems to defeat any point in having this very sophisticated switching.

If you want to ensure no drain on the standby supply when the main supply is on, you just need a 'power good' signal from the main supply to turn on a p-channel in the aux supply when it goes low. No need to switch the main supply, if it's dead it's dead and the diodes prevent backflow anyway.

[ghassan]

Helllo,

First , I would like to thank you so much for your kind reply .

Second , I didn't think that someone would reply to my post as i have been posting the same project at many places on internet.

Third , I am trying to build a fast switching dual power supply that can handle at least 5 Amps up to 10 Amps , maybe later i can replace p-channel mosfets with higher current as an upgrade to the project , but i really need this project to end and i thought i might find someone who can assist or work it out the problems that i am getting and to explain it to me .

I will try to explain to you as much as i can ...

I assume main supply as wallet power supply or dc source that goes off ad on 2 times during the day , and i assume auxiliary power supply as a battery bank 24vdc . What i am trying to achieve is i don't want to discharge the battery at all if i am using main power supply . so all components that are connected to main supply such as window comparator , xor logic gate , and logic gate , main supply driver .. and for the auxiliary should be supplied through its auxiliary power supply . The auxiliary power supply should output when the main supply is completely off ..

I already replaced the ideal diode (mosfet) with a  schottky diode . I chose to use mosfet instead of a diode just to have a very low power dissipation and lower the heat since i am going to carry a load of 5 amps up to 10 amps @24vdc. i tried to add a dead time between both power supplies and it works but not 100% . I already attached the full schematic . I can re attach it again if you would help .

regarding putting a schottky diode between base to collector of Q1 and Q6 is causing the p-mosfet to have a delay of 1 micro seconds. removing the schottky diode improves the latency from 1 microseconds to 23 nano seconds. That is why i am aiming to have a fast switching within nano seconds.

Another problem is C2 and C4 which is damaging the gate due to the high capacitance , because the Logic gate has a maximum output current rating value of 50 milli Amps .

Cross Conduction is my main problem for now.

I attached my full LTSPICE schematic . That is what i came through so far .

I use LtSpice to simulate this schematic.

I really appreciate your help.


 

[hermit]

I assume main supply as wallet power supply or dc source that goes off ad on 2 times during the day , and i assume auxiliary power supply as a battery bank 24vdc . What i am trying to achieve is i don't want to discharge the battery at all if i am using main power supply . so all components that are connected to main supply such as window comparator , xor logic gate , and logic gate , main supply driver .. and for the auxiliary should be supplied through its auxiliary power supply . The auxiliary power supply should output when the main supply is completely off ..

So basically a UPS?  (Uninterpretable Power Supply)

[ghassan]

Hello hermit,

yes it should act as a UPS for my wireless repeaters and routers . I am living in a country with unstable electricity . The electricity goes off every 4 hours up to 6 hours during the day so i should keep all devices up 24/7 . I am really interested to build this project.

Thank you

[ghassan]

I already finished almost parts in this project such as :
- Over Voltage [each power supply] by adding window comparator
- Under Voltage [each power supply] by adding window comparator
- Reverse Polarity [each power supply] by adding N-Mosfet to the input GND for each power supply.
- Protection to Main supply & Aux Supply by Back to Back Mosfets!.
- Fast Switching Time between power supplies are in few nano seconds .

- Trying to Inhibit Cross Conduction: still not 100%.

I will upload more pictures to show the switching problems.