Circuit to use the lowest available supply voltage

57 views
Skip to first unread message

Aaron Power

unread,
Apr 19, 2013, 9:39:39 AM4/19/13
to sydney-h...@googlegroups.com

I am trying to power an Arduino circuit from multiple possible power supplies : 12V, 24V or 48V. Any one or more of these supplies could be connected at the same time. I could wire the three supplies through diodes, as shown below, but that would mean that the highest of the available supply voltages will power the circuit. Which in turn means that my DC-DC step down voltage converter (which I have just shown as “LOAD”) would have to work harder and end up getting much hotter.

 

 

Alternatively, I could use three DC-DC converters in front of the diodes with their set points such that the lowest voltage would appear as the highest voltage to the diodes (e.g set the 12V step down to 9V; the 24V step down to 8.5V and the 48V step down to 8V). Now the 48V supply would only draw an appreciable current when neither the 24V nor 12V supplies were available. But that would then take three step down converters and more importantly, take up three times the space.

 

Does anyone know of a simpler way to achieve this?

 

Thanks,

Aaron.

 

image001.png

Andrew

unread,
Apr 19, 2013, 4:50:58 PM4/19/13
to sydney-h...@googlegroups.com

Voltage comparators controlling P-channel FETs?

--
You received this message because you are subscribed to the Google Groups "Robots & Dinosaurs" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sydney-hackspa...@googlegroups.com.
To post to this group, send email to sydney-h...@googlegroups.com.
Visit this group at http://groups.google.com/group/sydney-hackspace?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

image001.png

Kean Maizels

unread,
Apr 19, 2013, 6:30:36 PM4/19/13
to sydney-h...@googlegroups.com
That would be my suggestion also.  The trick is then having a power source for the comparators.
You could possibly just do some simple logic to do the priority selection by using diodes and/or transistors.

Kean

On 20/04/2013, at 6:50 AM, "Andrew" <and...@arcadius.com.au> wrote:

Voltage comparators controlling P-channel FETs?

 

From: sydney-h...@googlegroups.com [mailto:sydney-h...@googlegroups.com] On Behalf Of Aaron Power
Sent: Friday, 19 April 2013 11:40 PM
To: sydney-h...@googlegroups.com
Subject: [RnD] Circuit to use the lowest available supply voltage

 

I am trying to power an Arduino circuit from multiple possible power supplies : 12V, 24V or 48V. Any one or more of these supplies could be connected at the same time. I could wire the three supplies through diodes, as shown below, but that would mean that the highest of the available supply voltages will power the circuit. Which in turn means that my DC-DC step down voltage converter (which I have just shown as “LOAD”) would have to work harder and end up getting much hotter.

 

<image001.png>

 

Alternatively, I could use three DC-DC converters in front of the diodes with their set points such that the lowest voltage would appear as the highest voltage to the diodes (e.g set the 12V step down to 9V; the 24V step down to 8.5V and the 48V step down to 8V). Now the 48V supply would only draw an appreciable current when neither the 24V nor 12V supplies were available. But that would then take three step down converters and more importantly, take up three times the space.

 

Does anyone know of a simpler way to achieve this?

 

Thanks,

Aaron.

 

--
You received this message because you are subscribed to the Google Groups "Robots & Dinosaurs" group.
To unsubscribe from this group and stop receiving emails from it, send an email to sydney-hackspa...@googlegroups.com.
To post to this group, send email to sydney-h...@googlegroups.com.
Visit this group at http://groups.google.com/group/sydney-hackspace?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

Terry Dawson

unread,
Apr 19, 2013, 8:29:10 PM4/19/13
to Sydney Hackspace Mailing List
Aaron,
So long as your load is low current draw you might get away with three appropriately-valued Zener diodes in place of the 1N4148s.

Terry



--
image001.png

Kris

unread,
Apr 19, 2013, 9:30:49 PM4/19/13
to sydney-h...@googlegroups.com
I'm at the shops so apologise for shite drawing, but what about using a combination of normally open, normally closed relays? (or their solid state equiv) you can arrange them so that the 12 line will disconnect the other sources.

Kris

(the drawing isn't complete but you get the idea)
--
Sent from my Android phone with K-9 Mail. Please excuse my brevity.
20130420_112737.jpg

ada

unread,
Apr 20, 2013, 6:40:41 AM4/20/13
to Robots & Dinosaurs


On Apr 19, 11:39 pm, "Aaron Power" <Aaron.Po...@exemail.com.au> wrote:
> I am trying to power an Arduino circuit from multiple possible power
> supplies : 12V, 24V or 48V. Any one or more of these supplies could be
> connected at the same time. I could wire the three supplies through diodes,
> as shown below, but that would mean that the highest of the available supply
> voltages will power the circuit. Which in turn means that my DC-DC step down
> Does anyone know of a simpler way to achieve this?


Are you guaranteed that the supply labelled 12V is always the lowest
supply, 48V the highest, etc?

If so, you can turn this into a logic problem:

- always supply off the 12V
- only supply the 24V if the 12V is not connected
- supply the 48V if neither of the others is connected

You can do this with discrete PMOS NOT and OR gates.

ada

unread,
Apr 23, 2013, 8:38:29 AM4/23/13
to Robots & Dinosaurs


On Apr 20, 8:30 am, "Kean Maizels" <K...@kean.com.au> wrote:
> That would be my suggestion also.  The trick is then having a power source for the comparators.
> You could possibly just do some simple logic to do the priority selection by using diodes and/or transistors.

you can diode or the power source to the comparators, and then divide
down the voltages so that you're not comparing outside the supply
voltage range of the comparator.

On Apr 20, 10:29 am, Terry Dawson <vk2...@gmail.com> wrote:
> So long as your load is low current draw you might get away with three
> appropriately-valued Zener diodes in place of the 1N4148s.

Zener diodes don't work because they will conduct in the opposite
direction like normal diodes (Zener breakdown is a specialised case of
regular reverse breakdown).

Terry Dawson

unread,
Apr 23, 2013, 5:00:42 PM4/23/13
to Sydney Hackspace Mailing List

Ada,
I was suggssting using them in precisely that say, reversed, for the voltage drop. It's something that I've seen exploited in a number of circuits over time. It's not common because often current requirements make it impractical.

Terry

ada

unread,
Apr 23, 2013, 9:13:45 PM4/23/13
to Robots & Dinosaurs


On Apr 24, 7:00 am, Terry Dawson <vk2...@gmail.com> wrote:
> Ada,
> I was suggssting using them in precisely that say, reversed, for the
> voltage drop. It's something that I've seen exploited in a number of
> circuits over time. It's not common because often current requirements make
> it impractical.

What I am saying is that you can't use Zeners in a diode or
configuration because they will allow current to flow in the forward
direction, unless you size them so that the highest voltage one drops
to a voltage lower than any of the others will ever achieve.

Imagine your sources are three alkaline batteries of 8S, 16S, and 32S
configuration respectively.

You place a 1V zener on the 8S battery, a 13.5V Zener on the 16S
battery, and a 34V zener on the 32S battery, so that the voltages
after zener breakdown and drop are 11, 10.5 and 10V.

If the battery voltage drops to say 1.1V/cell (not unheard of for a
pretty flat alkaline), the 8S battery now produces 8.8V. Its zener is
now forward biased from the 16S battery (and the 32S battery), and
will start charging the flat battery.

Julian Sortland

unread,
Apr 23, 2013, 10:26:07 PM4/23/13
to sydney-h...@googlegroups.com
You could use a normal (1N400x) diode in series with each zener.  This would provide the "OR" function, and prevent backwards current flow.

I have used a similar set up to make a polarity and voltage checker, using SMD parts, and the board glued to the top of a lighter socket double adapter, and covered with more epoxy.  The green LED string was just a diode, R and LED, indicating correct polarity.  The reverse, in red, indicated negative, and the following, also in red, indicated correct polarity, but over about 18 volts, suggesting a 24 volts outlet.  The Zener was probably 15v.

+ o---|>|---|<|---www---|>|---o  -
        D    ZD    R     LED   

I also used a more complex circuit involving a bridge, and a common cathode RGB LED build into a Clipsal Extra Low Voltage "T" plug - designed for 32V farm power, but frequently used in places like police cars and ambulances for special equipment.  While some states follow a standard, Vic doesn't, but even following the standard (the bottom of the T ground, as per a 240v plug) would put negative 24v on the top of the T in an old positive chassis fire engine!  And ambos don't care, because they are running a heating element in a humidicrib.

73,

Julian VK2YJS.

Terry Dawson

unread,
Apr 24, 2013, 6:32:30 PM4/24/13
to Sydney Hackspace Mailing List

Yes, goid point, you're right, they'll still need conventional diodes in series.

Terry

Kris

unread,
Apr 24, 2013, 7:50:43 PM4/24/13
to sydney-h...@googlegroups.com
Are relays unsuitable because they draw too much current for the coil?
You can satisfy the brief with three components...
relayselect.GIF

Aaron Power

unread,
Apr 27, 2013, 10:31:02 AM4/27/13
to sydney-h...@googlegroups.com

Thanks to everyone that replied to this question. I know I have been very quiet with your responses, but I have been reading each one and looking at how I could make it work in my application.

 

Relays are a possibility, and it does solve the problem in only three components, but the power draw for the three coils is a reasonable amount of wasted energy in a battery only circuit. And finding small PCB mount relays in 12V, 24V and 48V would not be so easy. And even the relatively small ones still take up fair bit of circuit board space. What I do like about this solution is that the devices would be very forgiving of voltage variations

 

Using voltage comparators does leave the problem of how do you power the comparators when you don’t know which voltage supplies will be available. The final “load” on the circuit will be a micro-controller, and one of its jobs will be to monitor the various supplies. I see this as “boot strap” type problem, and really didn’t get very far with it. It might be possible to use a simple diode OR to initially power the micro-controller, and then have it turn off the higher voltages, if a lower one is also available. Maybe.

 

I have spent most of my time looking at the option of using discrete PMOS devices, and have come up with the following circuit (for the purpose of the simulation I just randomly chose a PMOS FET that LT Spice had in its library, and looked roughly suitable):-

 

If, for example, both the 12V and 24V supplies are available, then a current will flow from the 24V bus, through R5 and D1 to the 12V bus. This sets up a voltage drop across Vgs of PMOS4, turning it on. This in turn will short-out R3, turning off PMOS2. Thus the load only sees the lower voltage supply. In a similar way the 48V supply will be switched off through D2 if the 24V supply is also available, or through D3 if the 12V supply is available.

 

I have run the simulation with the following combinations of supplies, and found that the load is correctly supplied from the lowest voltage each time.

12V

24V

48V

V(vload):

PMOS1 Vgs

PMOS2 Vgs

PMOS3 Vgs

PMOS4 Vgs

PMOS5 Vgs

ü

û

û

12.0

-10.9

-8.0

-8.0

0.0

0.0

ü

ü

û

12.0

-10.9

0.0

-8.0

-11.4

0.0

ü

û

ü

12.0

-10.9

-8.0

0.0

0.0

-35.4

ü

ü

ü

12.0

-10.9

0.0

0.0

-11.4

-35.4

û

ü

û

22.6

-20.5

-1.0

-15.0

-0.9

0.0

û

ü

ü

24.0

-21.8

-16.0

0.0

0.0

-23.4

û

û

ü

46.6

-42.3

-31.1

-1.0

0.0

-0.9

 

In most instances, the circuit is very efficient. However, as you can see from the table above, when the circuit is only supplied from the 24V or only supplied from the 48V power source, the efficiency drops, but it is still not too bad. Some tweaking of the resistor values might be able to improve this performance.

 

But I am more concerned with the voltage drop across the Source / Gate. The small FETs I have won’t handle -40V or -30V. If I can solve this problem, then the next step will be to build up a prototype and see if reality agrees with the simulation.

 

Any thoughts or comments on what I have so far?

 

Aaron.

 

image002.jpg

ada

unread,
Apr 27, 2013, 11:44:37 AM4/27/13
to Robots & Dinosaurs


On Apr 28, 12:31 am, "Aaron Power" <Aaron.Po...@exemail.com.au> wrote:
> Using voltage comparators does leave the problem of how do you power the
> comparators when you don’t know which voltage supplies will be available.

sometimes it feels like I talk to a brick wall with this mailing list.

https://groups.google.com/group/sydney-hackspace/msg/4f5c88c1afb1dda3

> I have spent most of my time looking at the option of using discrete PMOS
> devices, and have come up with the following circuit (for the purpose of the
> simulation I just randomly chose a PMOS FET that LT Spice had in its
> library, and looked roughly suitable):-

Simulation is not a substitute for thought.

https://groups.google.com/group/sydney-hackspace/msg/601c5bdc72ed22c2

> In most instances, the circuit is very efficient. However, as you can see
> from the table above, when the circuit is only supplied from the 24V or only
> supplied from the 48V power source, the efficiency drops, but it is still
> not too bad. Some tweaking of the resistor values might be able to improve
> this performance.

`Pluralitas non est ponenda sine necessitas.'

I'm not sure how to create a voltage source in LTSpice that isn't also
a current sink, and that may be part of the problem, but:

- do you need the R1/R2/PMOS1 circuit? it's a mosfet biased to turn
itself on if the voltage is above a certain amount, but I don't see
what else it does at 12V where it's strongly biased on anyhow.
- you can pop a zener around R4 to protect the gate of PMOS2. With
that, you don't need the pullup (R3).
- again, a zener clamp around R8 and R7.
- what does D1 do?

I haven't thought thoroughly about transient response or stability,
but these might or might not be important.

Aaron Power

unread,
May 31, 2013, 11:19:23 AM5/31/13
to sydney-h...@googlegroups.com

This has been an interesting (and frustrating) problem. Just thought I would give an update on what I have found.

 

Ada's comment the last time I brought this up proved spot on. The problem with the circuits always came down to the fact that the power supplies did not also act as current sinks. The circuit relied on the lower voltage supplies to clamp the voltage at their terminals to that value. This only required a few uA of current, but if the power supply could not sink *any* current, and its output terminal was allowed to float above (but never below) its rated voltage then the circuit would fail under some scenarios. I could get it to work under most circumstances, but there was always at least one combination of supplies that failed.

 

I tried various different circuits to try and create the necessary logic, but every time I eventually came up against the same problem – I was trying to switch the high side of a higher voltage supply “off” or “on” based on the presence or absence of a lower voltage. Sooner or later this always ended up with the same problem.

 

Then one night I was looking at the circuits and realised that what was needed was some way to isolate the lower voltage from the higher voltage. I thought about using optocouplers, and came up with the following circuit:-

 

 

I simulated the fact that the power supplies were not current sinks by adding a series diode (and increasing the nominal voltage to compensate for the forward voltage drop). This circuit seemed to work under every scenario I tried in the simulation, so I ordered some optocouplers and waited for them to arrive. The package arrived earlier this week and tonight I built up a test circuit – and it works exactly as expected.

 

The only real problem with the circuit is the amount of wasted power in R2, R4 and R5. In order to saturate the optocoupler, I need around 5 to 10mA of current through the LED side. With a forward voltage drop of only around 1 to 2V, this leaves quite a large voltage to drop across the series resistors.

 

A better solution I did come across was an interesting circuit based around a depletion mode MOSFET that could effectively give you a normally closed “switch” which would turn off with any positive voltage applied. (see http://www.aldinc.com/pdf/ZeroPowerNormallyONSwitch.pdf - figure 2) However, the voltage ratings, and maximum current draw of these devices means that the circuit as shown would not be suitable for my application. I might see if I can get some of these devices and play around with them, but it is not a priority.

 

The reality is that setting up this circuit to automatically chose the lowest available voltage is probably more trouble than it is worth. I will probably just end up putting three pairs of jumper terminals on the board and use them to select one out of the three possible supply voltages.

 

Regards,

Aaron.

--

image002.jpg
Reply all
Reply to author
Forward
0 new messages