Linear regulator input voltage limitations.

[David_]

Hello.

I find my self very limited and troubled by the max input voltage of 3.3V and 5V regulators, a max 50mA output current is really on the edge to not be enough. 100mA would definitely suffice, is there any simple cheap way of using a regulator supplied by a for the device in question damaging input voltages?

Say that my regulator can't handle more than 30V input and I have a 40V rail, if it would so happen that I can't(for the sake of argument) get another regulator and have to find a way to use this 30V input range regulator. What to do?
Is there a solution?

Regards

- - - Updated - - -

I know that there are regulators out there that fit my need but I don't want to pay those prices, not if there are any other way.

- - - Updated - - -

Could one design any kind of transistor/zener or perhaps opamp/transistor/zener design that can rival or equal the function of say ordinary 7805/7833(3.3V) regulators?
Short circuit protection might not be really relevant in this case since it is a internal devise to the apparatus i'm designing, but over current protection would be a good idea I guess.

 

[D.A.(Tony)Stewart]

The proper way to design a power supply if define available input range, and required output V,A peak, avg and error tolerance.

 

[betwixt]

As you apply higher input voltages, the heat generated by the linear regulator will also increase. You need to take that into consideration. Even with a 50mA load and 30V dropped to 5V it will dissipate 1.25W and if you take the other extreme of 100mA, 40V and 3.3V output it will be 3.7W.

If you know the current will be drawn continuously, you could share the voltage drop and heat dissipation with a fixed resistor before the regulator. It will drop (I * R) volts.
Beware that if you stop drawing current (I = 0) the voltage dropped will also be zero and full voltage will arrive at the regulator.

Brian.

 

[KlausST]

Hi,

For a consrant voltage drop you could use a zener in series.

Also there are simple "resistor, zener, bjt" voltage regulator circuits.

But the additional cost fro a standard regulator to a 40V input regulator is not much either.

Klaus

 

[srizbf]

May be you can take the 723 based regulator Ic and construct a supply to your requirement.

if you have already ruled out 723 IC , then a discrete and zener diiode may be the only solution.

 

[asdf44]

There are lots of ways to do it but none of them avoid the main problem of higher heat dissipation. As Betwixt pointed out you're talking between 1.5-4W which ends up adding size and cost.

It's worth considering a switching supply which avoids thermal issues completely in this range. Here is a line that goes up to 72V. This especially makes sense if you actually have multiple low voltage rails, say 3.3 and 1.2. Then you use the switcher to drop the bulk of the voltage (say, 40->5) and use linear regulators from 5 down.

https://www.recom-power.com/?id=271&no_cache=1&user_recom_pi2[xseries]=R-78xx-0.5+R-78HBxx-0.5(L)+R-78HB24-0.3(L)+R-78Exx-0.5+ROF-78xx-0.5

 

[D.A.(Tony)Stewart]

The proper way to design a power supply if define available input range, and required output V,A peak, avg and error tolerance.

This is a MUST, so learn now, to write simple specs before design starts and gives us a better understanding, rather than waste time, in the dark

if relevant , include cost, qty , efficiency min, ease of design...

 

[dick_freebird]

There are higher input voltage range regulators, and if you
don't want to pay the extra pennies you're probably ignoring
the BOM cost of discrete Band-Aids to make a weaker one
survive.

There are app notes out there about increasing output
current by adding an external power transistor, for some
of the old school linears at least.

You could fairly simply create an input cascode "pre-
regulator" to maintain the linear's input voltage in a safe
range. But again you're probably adding a buck's worth
of loose components to get input range that would cost
you that same dollar, or less.

 

[David_]

I will check and see at what price I can implement a resistor, zener, bjt circuit.

I should make this clear, if I had been looking to buy regulators in the 1000 peaces quantity then the problems addressed by this thread would not exist.
There are lots of good high input voltage regulators that would be a perfect fit for me with a good price tag to but when I look for components on digikey I have to sift through pages and pages to find the cheapest parts while the minimum quantity is 1, or 10. And those suitable regulators are simply not available to me in that way or those that are are sold >1pc have a price that makes them not even possibly a candidate for my circuit. Okey they are I guess but it feels like a very sour apple for me and as pointed out even at low voltages and low currents there will be a power dissipation not to be over looked.
I am working on a project of my own and my economy is today so limited that it gets in the way of the progress of design.
I have not been doing electronics for very long and I had made out a budget to get me through the design but this is my very first "real" design which will be produced in China, and that brings problems that has blown my budget while I have learned some lessons. I am learning and when I had a finished PCB layout I had learned enough about switching converters that I feelt I could tackle a off-line flyback design in order to get rid of a big mains transformer.

Now I had come to think that I had wasted your time since given the flyback design I could just implement a low voltage auxiliary winding to supply all low voltage circuits, but that is not the case. The master output voltage will be adjusted(or that is the plan) to function as a pre-regulator for a linear regulator stage, which means that a low voltage auxiliary winding will not work.
The master output would idealy go between 55V down to max 5V as the low voltage limit and how then could a low voltage aux output be implemented...

I think I will have to have two identical windings and regulate the "un-regulated" output down to low levels,

I see it like this, first I need a circuits that has the following specs:

Input voltage range = 5-55V.
Output voltage = 5V(within ±6%)
Output peak current = 50mA.
Output avarage current = 29mA.

and also a circuit with these:
Input voltage range = feed from the rail specified above.
Output voltage = 3.3V(-1% / +10%)
Output peak current = 37mA
Output avarage current = 19mA

The 3.3V rail is supplying a voltage reference IC that outputs 3V so it needs a headroom, maybe I should consider routing the 5V to the reference to relax on the 3.3V tolerance...

 

[KlausST]

Hi,

If you could adjust the flyback transformer.
Then you have a primary winding and a secondary. Now you speak of an auxilary...

You know the flyback design. Often the primary is connected to +input DC voltage. The other primary to a switch.

Then during switch is ON the core gets charged.. during this time an about constant output voltage is at the secondary abd auxilary winding. Youcould use this constant voltage for your "fixed" low power supply.

When the primary is switched OFF, then the transfirmer output voltage is reversed and it charges the variable voltage high power output.

I think there is a good chance to get an almost stable output from the auxilary winding..

Klaus

 

[crutschow]

How about an LM317HV adjustable regulator.
They can handle up to 60V with the output voltage being determined by two resistors.
They are reasonably cheap in the TO-220 package.
As noted, you will need a suitable heat sink for it but this can be reduced by adding a power resistor in series with the input to absorb some of the power.

 

[David_]

I had thought that the meaning of auxiliary would enable me to use it to mean another word for a third winding whatever the purpose would be. But when searching google for auxiliary I find that my interpretation of that word was something of my own mind.
When I say auxiliary winding/output, what am I really then talking about?

The following image is what I had imagined, note that it is just a figment of my imagination so far and I have not even worked out how the feedback would work or if the low voltage stuff could work but it shows my concept as I had thought of it:


I'm not sure how you mean Klaus, do you say that I can produce a low voltage rail with the same winding as my main output during its usual "off" period caused by the reversed biased diode?

Those LM317 is not my friends, or more likely I'm not there's(I had probably damaged them in some way). I have had a coupe of experiences where for some reason the have acted very strange and I have come to feel much safer with fixed output regulators.

But I see it as I have two options, ether I find(or understands) a way to make my self a low voltage rail with the transformer or I have to go with the suggestion of LM317HV. There are others fixed <60V regulators a little cheaper than LM317HV but they come in SOIC8 and it would seem as with a SOIC8 you can't really go much further than 2W.
So that leaves LM317HV.

 

[KlausST]

Hi,

Why does the transformer have two "1" and two "4" pins?

In the current configuration the voltage of c14 depends on the regulated output voltage.

But when you interchange pin1 and pin4 of the upper "auxilary" winding, then the voltage of c14 will be more constant.
Is it a custom made transformer? If so then adjust winding count to get the desired output voltage.
Output voltage is about: V_aux = V_prim * n_aux / n_prim.

Maybe a current limiting resistor in series with the catch diode is necessary.

Klaus

 

[D.A.(Tony)Stewart]

Thanks for specs. For 8$ these are my 3 TI picks
https://www.ti.com/product/lm5160

For 3.3V 40mA https://www.digikey.com/product-search/en/integrated-circuits-ics/pmic-voltage-reference/2556223
For 3V adj zeners use ~1mA with 10 or 15 mA out
Both very accurate.

 

[crutschow]

......................
Those LM317 is not my friends, or more likely I'm not there's(I had probably damaged them in some way). I have had a coupe of experiences where for some reason the have acted very strange and I have come to feel much safer with fixed output regulators.

But I see it as I have two options, ether I find(or understands) a way to make my self a low voltage rail with the transformer or I have to go with the suggestion of LM317HV. There are others fixed <60V regulators a little cheaper than LM317HV but they come in SOIC8 and it would seem as with a SOIC8 you can't really go much further than 2W.
So that leaves LM317HV.

The LM317 is generally very tough and forgiving since it has built-in over-current and over-temperature protection, but can be damaged if you hook it up incorrectly.
Make sure you double-check the pinout of the device package you use.

And use the recommended decoupling capacitors at both the input and output pins directly to ground.
The device may oscillate or otherwise act up if you don't.

 

[asdf44]

I wouldn't avoid the LM317 which is a useful part.

Your specs above did you account for the 3.3V load in your 5V output numbers? A 37mA 3.3V regulator uses up most of your specified 50mA for the 5V?

Also, note that you probably can't get away with a resistor in series with your regulator without messing up operation at low voltage. So you probably need a fairly large heatsink (2.5W ambient isn't unreasonable, but isn't small either for a TO-220).

 

[David_]

One weird thing that a LM317 was doing was to change the output voltage when the input voltage changed, I had(with a pot) dialed in 3.3V while having two 9V batteries in series on the input then I added one battery to make three batteries in series on the input and I expected to LM317 to maintain 3.3V but no. Now the output was 4.1V and in that manner was the output changing... anyway.
But if I would just use them in a design again, calculating needed resistors and experience it working I'm sure I would be happy with them again. While having trouble with two LM317 I did blow up quite a lot of silicon which is what I now associate with LM317 even though they where not the culprits.

As for the transformer, I wanted to illustrate how I thought about is so I quickly opened the transformer symbol, "cut" the secondary side and used that to make a new 'symbol/same package/device' so I could draw longer lines in the schematic to fake-extend the core and then make it appear as one device but its really two.

But yes I will wind my own transformer as soon as I have succeeded in calculating what core size I need, There are obviously quite a lot about flybacks that I don't get but seeing as the master output(lets call the two(possible) outputs master / slave) is regulated by the PWM controller which somehow will be set by the "linear control stage"(I hope) then the slave output will follow the master given that they have an equal number of turns on the transformer. And since I need >5V on the slave at all times and the master will go between 5V to 55V then the slave would need to be at least equal to the master.

But do I understand you right in that the above does not need to be true if I flip the winding polarity so as to supply the slave during the switch ON-time?
The name flyback converter comes from the winding voltage polarity reversal action(flyback action) when the switch turns off does it not?

Anyway here's two variations:

 

[KlausST]

Hi,

But do I understand you right in that the above does not need to be true if I flip the winding polarity so as to supply the slave during the switch ON-time?

Yes, that's what i have in mind.

Therefore i recommend to use variation1. Then you don't have the in a wide range changing input voltage to the 317 regulator, and thus not the huge power dissipation. Try winding count W3 = W1 / 30.

*********
If you simulate the circuit, then simulate:
* Max Vout with max Iout and min Vin. To check if the core size meets your power requirement.
And
* Min Vout with max Iout and max Vin.
* Min Vout with min Iout and max Vin. To check that core does not saturate. In those cases the controller needs a very low duty cycle or pulse skipping,

Klaus

 

[David_]

I see that you already was very clear about taking the voltage during the 'ON' time Klaus, then I'll guess this is quite solved.
It took a while to get through to me though

I will think it through before I mark th thread as solved though.
Thanks

- - - Updated - - -

Is this something that LTspice could handle or do I need another simulator?

- - - Updated - - -

I don't get how to do the transformer/coupled inductor in LTspice but I have found a model that I will try to adapt.

 

[KlausST]

Hi,

Sorry i never did a simulation on flyback circuits, therefore i can't recommend a simulation tool.

Klaus