Optimizing Non-Adjustable Voltage Regulators with Simple Resistor Configurations

[GregValcourt]

I've played with adjustable voltage regulators already (such as nte956). My question is about "non-adjustable" voltage regulators (78xx).

What is the easy way to adjust these? And by easy, I mean just using one or two resistors. I've seen some solutions using zener diodes, but I would like to use parts at hand.

I would think that changing the "reference ground" for the 7805 up by 0.25 volts, I should get 5.25 (in theory, forgetting about margins of error). So why not just use two resistors: one between real ground and the ground pin, and another between input positive and the ground pin? Shouldn't this work?

We could easily calculate a ratio for those resistors. Say input is 12V. 0.25 / 12 = ~0.02. I would think to use higher values (1MΩ on the positive and 220KΩ on the -ve side) to use less power. I'm wondering if I'd be hampering the ground pins ability to discharge current (i don't know if it's an issue, just a thought, if it's a pure reference voltage, it shouldn't matter).

What about using the output voltage to create a reference voltage (resistor to the output instead of the input +ve)? Would that be better (since it's regulated)? Or would that be worse (feedback)?

 

[berkeman]

I've played with adjustable voltage regulators already (such as nte956). My question is about "non-adjustable" voltage regulators (78xx).

What is the easy way to adjust these? And by easy, I mean just using one or two resistors. I've seen some solutions using zener diodes, but I would like to use parts at hand.

I would think that changing the "reference ground" for the 7805 up by 0.25 volts, I should get 5.25 (in theory, forgetting about margins of error). So why not just use two resistors: one between real ground and the ground pin, and another between input positive and the ground pin? Shouldn't this work?

We could easily calculate a ratio for those resistors. Say input is 12V. 0.25 / 12 = ~0.02. I would think to use higher values (1MΩ on the positive and 220KΩ on the -ve side) to use less power. I'm wondering if I'd be hampering the ground pins ability to discharge current (i don't know if it's an issue, just a thought, if it's a pure reference voltage, it shouldn't matter).

What about using the output voltage to create a reference voltage (resistor to the output instead of the input +ve)? Would that be better (since it's regulated)? Or would that be worse (feedback)?

I'm not familiar with the 956, but the LM317 is the traditional way to do a positive adjustable linear voltage regulator. Trying to adjust a fixed linear voltage regulator is going to have lots of other issues...

Why are you reluctant to just use an LM317?

 

[CWatters]

Greg, See Figure 3 on page 7 of this data sheet. It shows exactly what you are talking about. An adjustable voltage regulator using two resistors..

https://www.sparkfun.com/datasheets/Components/LM7805.pdf

 

[CWatters]

PS. They connect one from the output (rather than the input as you suggest) because the output voltage is regulated where as the input isn't.

 

[sophiecentaur]

If you happen to have a drawer full of 5V regulators then why go to the trouble of buying a special 12V reg if all you need is a 7V Zener in the ground connection - or a simple pot-down? (Maybe not such good regulation but that's not usually a serious problem).

 

[GregValcourt]

I'm not familiar with the 956, but the LM317 is the traditional way to do a positive adjustable linear voltage regulator. Trying to adjust a fixed linear voltage regulator is going to have lots of other issues...

Why are you reluctant to just use an LM317?

I'm not reluctant use an adjustable voltage regulator (the 956 is an adjustable switching regulator). I'm just curious and experimenting.

 

[Averagesupernova]

If you keep in mind that the three terminal regulator only worries about keeping the voltage between the output and the ground pin (or adj pin in the case of adjustable regulators like the 317) constant then you can do a lot with them. There are several threads on PF that talk about this. I recall posting in one. Maybe more.

 

[meBigGuy]

A zener diode in the ground pin is a bad idea since the ground current is generally not enough to accurately produce the zener voltage.

The resistive divider shown in the datasheet Cwatters posted is the best for modern regulators.

Better yet is an LM317 or equiv.

 

[sophiecentaur]

A zener diode in the ground pin is a bad idea since the ground current is generally not enough to accurately produce the zener voltage.

I would, of course, use a bleed resistor to make the zener work right.
If you wanted 10V, an even better way would be to use two 5V supplies, piggyback.

 

[Averagesupernova]

I see no reason to ever use a zener with a three terminal regulator. The function of a zener is as a voltage reference. The voltage reference is a function already built into a 3 terminal regulator.

 

[sophiecentaur]

Perhaps not. It should / could improve regulation, though, if the voltage reference is not a function of the 'stabilised' output voltage of the regulator. Perhaps a fat decoupling capacitor in the resistor chain might help with that.

 

[Averagesupernova]

I will say this: Take a 317 with a 1.25 volt reference between the adj and output. For the sake of argument let's say it has a +/- spec of 5%. Yes, that is high, but for the sake of the discussion, we will set it at that. Set the resistors for an output of 20 volts. The output will be 20 volts +/- 1 volt. If we were to just use the regulator with a 1.25 volt output by grounding the adj pin then it would be +/- .0625 volts. So, if we were to set a precision reference in between the adj pin and ground we would then get a higher voltage out with a lot better precision than using the resistors. I haven't found an actual zener diode that is as stable and accurate as a precision reference although they are often connected in a very similar manner. So your point definitely stands sophie, but I would never choose a zener.

 

[sophiecentaur]

Fair enough. Your resistor chain would be the obvious solution in a practical situation. Most electronics should be designed to be pretty well impervious to the quality of the Power supply, whenever possible.

 

[uart]

Sorry if I'm covering the obvious here, but I haven't seen this explicitly stated yet in this thread.

The salient difference between a 3 terminal fixed regulator and a 3 terminal adjustable regulator is the amount of current (and it's variability) in the common (ground or adjustment) terminal.

Yes there's the other difference of using a lower voltage on the adjustable type so as to get a greater adjustable range, but it's the previous consideration that's most important. It's bias/quiescent current that is the limiting factor on using a fixed regulator in the same manner as an adjustable type.

For example the LM317 specifies a typical adj pin current of 50uA and max of 100uA, with typical operational variability of 5uA. The LM7805 in contrast has specs of 4.2mA typical, 8mA max and 1.3 mA typical operational variability. As you can see, much larger currents and a variability over 200 times larger.

Consider trying to use the 7805 in an adjustable configuration, with a (approx R/2R) voltage divider to get 15 volts output. To keep the variation in the output (as caused by typical operation variation in bias current) to say 0.13 volts would require the 2R resistor in the common lead to be 100 ohms or less. This requires >1 Watt power dissipation and therefore the use of a power resistor rated at several watts. This is inconvenient and wastes power, and I generally loath to have very hot running components unnecessarily incorporated in anything I design.

And remember that this (0.13 volts) is just the typical operation variation when using 2R=100 ohms. The fixed offset could be 0.4 to 0.8 volts unless you individually tweak the nominal 2R/R for each individual part. I hope this makes it clear why a fixed regulator is not an adjustable regulator.

 

[sophiecentaur]

That's the sort of input we should have had about ten posts ago. It needs someone who 'knows' about theses things to put us right about how to do it properly.Thanks uart.

 

[uart]

BTW, referring to the usage of a zener in place of the grounded resistor in the voltage divider (when using a fixed regulator for other voltages). This is not nearly as bad of an idea as some people here have suggested. Sure the accuracy and stability of a zener is typically going to be far worse than that of the basic regulated voltage, but appropriately biased (* see note) the zener will have far less operational variation than that of anything you can do with resistors under 1 watt.

* Note. We'd still include the R from the R/2R divider in the above given 15 volt example. A 10 volt zener would simply replace the 2R resistor.