I am trying to come up with a simple, yet efficient way to power a nominal 12 volt device (35watts) from a 24 volt battery bank.

This device needs to power the load with 1/2 of the supply voltage, not supply "regulated" 12v output.

It is important that as supply voltage "24v battery" decreases, the output voltage decreases as well.

I have tried a "darlington pair" as well as other transistor based voltage converting circuits, but the waste too much power, dissipated as heat.

Ideas?

 

How about something like this? http://www.amazon.com/dp/B003P1D82S...&hvpone=&hvptwo=&hvqmt=&ref=asc_df_B003P1D82S

Doubt if you could build one yourself for that price

 

You could use a buck switching regulator but instead of using a fixed reference voltage as is typically done to regulate the output voltage, you would use the input voltage through a voltage divider. That way the output would track the input.

 

I am trying to come up with a simple, yet efficient way to power a nominal 12 volt device (35watts) from a 24 volt battery bank.

This device needs to power the load with 1/2 of the supply voltage, not supply "regulated" 12v output.

It is important that as supply voltage "24v battery" decreases, the output voltage decreases as well.

I have tried a "darlington pair" as well as other transistor based voltage converting circuits, but the waste too much power, dissipated as heat.

Ideas?

Simple switcher. LM2576 or any of the newer variants.

Use an external op amp with an R-R divider from the input for the reference to the op amp whose output controls the FB pin of the LM2576.

 

Expanding on Crutschow's solution, it can be even easier: Use a buck converter topology and drive the switch with a constant 50% duty cycle. The output voltage will be 50% of battery voltage less IR drops and switching losses.

 

Thanks for the input, I have some LM2576-12 regulators on hand , and was thinking of using them. How could I set it up so that the output tracks according to the input?
The ones that I have are 12V fixed, not the adjustable ones.

 

Thanks for the input, I have some LM2576-12 regulators on hand , and was thinking of using them. How could I set it up so that the output tracks according to the input?
The ones that I have are 12V fixed, not the adjustable ones.

Use an external op amp with an R-R divider from the input for the reference to the op amp whose output controls the FB pin of the LM2576.

Use an external op amp for the error amp and it's output goes to the FB pin of the 2576. The ref voltage for the amp input comes from a divider off the input voltage. The other input comes from the output of the converter.

 

I dont think it will work,as when the supply voltage drops the buck ic will kick in and try to up the output voltage to maintain the ref voltage of 1.23V (internal voltage on pin 4) as per page 2 datasheet,
http://www.ti.com/lit/ds/symlink/lm2576.pdf

so you need to increase the ref voltage feed back to lower the output.
So your circuit needs a way of upping the ref voltage as the supply drops to lower the output voltage.

What is the application used for as there may be another solution?

 

addition to thread

you could use this op amp circuit to connect to your feedback pin

so as the supply drops the output of the op amp rises and and the regulator will lower its output in track, you may need to tweek the resistors across the 24v supply to compensate for losses.

 

so... this should be simple, but has proven to be difficult.
The application is a 24VDC system that powers a (12VDC) computer, 60 or so watts.
The computer is designed so that as its 12v (nominal) battery drops in voltage, the software shuts down in stages, allowing it to re-start systems as voltage increases.

I just need a switching converter that will take 18-29VDC input, has a fixed duty cycle, and outputs 1/2 of its input (9-14.5VDC), without wasting a lot of power.
A typical "regulated" supply is no good, because if the system voltage gets low enough to shut down the converter, the 12V is turned off without "warning" the computer.

 

so... this should be simple, but has proven to be difficult.
The application is a 24VDC system that powers a (12VDC) computer, 60 or so watts.
The computer is designed so that as its 12v (nominal) battery drops in voltage, the software shuts down in stages, allowing it to re-start systems as voltage increases.

I just need a switching converter that will take 18-29VDC input, has a fixed duty cycle, and outputs 1/2 of its input (9-14.5VDC), without wasting a lot of power.
A typical "regulated" supply is no good, because if the system voltage gets low enough to shut down the converter, the 12V is turned off without "warning" the computer.

Can you not just use a LM358 op amp regulator and a TIP3055 transistor, that way your output would be half of the supply all the time?

 

Like this circuit the output will be half of the supply at all times

 

Thanks Dodgydave for the example.
The linear regulator you mentioned will disipate too much power (heat) for my application, a low speed switching regulator would do a better job, I just havent been able to define the component, or circuit, yet

 

My original solution is still the best one:

Use an external op amp for the error amp and it's output goes to the FB pin of the 2576. The ref voltage for the amp input comes from a divider off the input voltage. The other input comes from the output of the converter.

The LM2576 works down to 7V input. If you insert an external error amp in the feedback loop whose reference is derived from an R-R divider off the input voltage, the LM2576 buck converter will regulate to exactly half of VIN all the way down to an input voltage of 7V.

Picture the basic circuit with the feedback line opened with an error amplifier inserted. One input to that E/A comes from the switcher output, the other input comes from the center of the R-R divider from input voltage to ground, the output of the E/A goes to the FB pin of the LM2576. The E/A would need loop compensation, but it would work.

 

@bountyhunter
Can we use the internal compensation with - LM2576 Adjustable version
A simple adder with the positive connected to (Vin/2)+Vref, and negative input connected to Vout, and the adder's output connected to the FB Pin of 2576?

 

@bountyhunter
Can we use the internal compensation with - LM2576 Adjustable version
A simple adder with the positive connected to (Vin/2)+Vref, and negative input connected to Vout, and the adder's output connected to the FB Pin of 2576?

If you add an op amp into the FB loop of the 2576, it may not be stable unless you put compensation on the op amp to reduce it's gain. I would do it on the breadboard to get best values.

 

@bountyhunter
Can we use the internal compensation with - LM2576 Adjustable version
A simple adder with the positive connected to (Vin/2)+Vref, and negative input connected to Vout, and the adder's output connected to the FB Pin of 2576?

If the adder has a gain of 1 from all inputs to the output, then you should be able to get by without any additional compensation.

But you need Vout going to the positve adder input and (Vin/2) + Vref connected to the negative input since the FB pin already has an inversion internally (a higher voltage on the FB pin reduces the output voltage).

 

If the adder has a gain of 1 from all inputs to the output, then you should be able to get by without any additional compensation.

But if it is to be an error amplifier, it can't be unity gain.

 

But you need Vout going to the positve adder input and (Vin/2) + Vref connected to the negative input since the FB pin already has an inversion internally (a higher voltage on the FB pin reduces the output voltage).

Thanks for pointing out! I realized that I have interchanged these!

But if it is to be an error amplifier, it can't be unity gain.

I guess in this case the internal EA of the LM2576 is being used, and the external opamp is only an adder which is not the error amplifier. Wont that be alright?
@shawnbeachside
LM2576 is limited to 3A only and cannot supply the 5A as required, (60W output at 12V), and a higher output current rated buck regulator would be required.

 

LM2576 is limited to 3A only and cannot supply the 5A as required

There are 5A simple switcher versions, you can find them easily.

I guess in this case the internal EA of the LM2576 is being used, and the external opamp is only an adder which is not the error amplifier. Wont that be alright?

I don't think so.

The schematic below shows how I did a similar function: opened the feedback loop of an LM2636 switcher and inserted a second error amp so the regulated output became half the core voltage. U2B is the error amp. U2A is just a gain block. All of the compensation is on U2B.