Cascading LDO voltage regulators

[andy1969]

Apologies if this is answered in another thread.
I am cascading a number of LDO voltage regulators, 7815, 7812, 7809 & 7805, to create a general power supply for my workbench. However, the output from the regulators seems to be all over the place. If I install the 15-volt reg, the output is correct; however, as soon as I add another regulator downstream, the output from the first reg reduces, and the output from the second reg is low. I have changed the voltage regulators for different units just in case I bought fake units from ebay. The unregulated supply seems to be good at 36v. I have reviewed the datasheets and the internet and can find details of any issues with cascading. Possibly I am caught in a "Trap for young players" or not so young in my case. Can anyone help and possibly provide a sketch if I am missing something?

[Siwastaja]

Cascading should be OK, can't think about any usual "trap". Can you show us the exact schematic (including capacitors etc.)?

(Also a terminology nitpick, the parts you listed (7815, 7812, 7809 & 7805) are not LDO (low dropout) regulators. The correct term is linear regulator.)

[mariush]

A lot of voltage regulators can not handle voltages higher than 30-32v. 36v is a bit high.  For example, LM7815's absolute maximum rating shows 35v input voltage, but other regulators may be capable of more.

You need some input capacitance and some output capacitance on the regulator, something like a 10uF electrolytic capacitor on input (doesn't have to be exactly 10uF, but anything higher than 100uF is unlikely to make a difference) and at minimum a 0.1uF ceramic capacitor on output is recommended. 
If the 12v and 9v and 5v regulators are very close to the 15v one, you'll be fine with just some 0.1-1uF ceramic capacitor on the input for these instead of the 10uF electrolytic.


Keep in mind linear regulators work by throwing out the difference between input voltage and output voltage as heat.
So for example, at 100mA (0.1A) current, the 7815 would produce (36v - 15v ) x 0.1A = 2.1 watts  - usually for more than 1 watt, a heatsink is recommended.

Each of these regulators consume around 5-10mA just to work (see quiescent current in datasheets), so keep in mind that it adds up if you cascade:

7815  - 20mA (5mA for own, 15mA consumed by 7812)
7812 -15 mA (5mA for own, 10mA consumed by 7809)
7809 - 10mA (5mA for own, 5mA consumed by 7805)
7805 - 5mA for own functionality

My advice would be finding a smaller power supply somewhere in the 18v and use an adjustable regulator like LM317 or LM1085 (or others with 1085 in name, LM is used by one manufacturer, there's other prefixes)

If you use a classic transformer, I'd also suggest to use one with two secondary windings, and switch between having the windings in series (more output voltage, less current) or in parallel (lower output voltage, 2x the current) so that you'll reduce the amount of power wasted on the regulator.

[srb1954]

Apologies if this is answered in another thread.
I am cascading a number of LDO voltage regulators, 7815, 7812, 7809 & 7805, to create a general power supply for my workbench. However, the output from the regulators seems to be all over the place. If I install the 15-volt reg, the output is correct; however, as soon as I add another regulator downstream, the output from the first reg reduces, and the output from the second reg is low. I have changed the voltage regulators for different units just in case I bought fake units from ebay. The unregulated supply seems to be good at 36v. I have reviewed the datasheets and the internet and can find details of any issues with cascading. Possibly I am caught in a "Trap for young players" or not so young in my case. Can anyone help and possibly provide a sketch if I am missing something?

You need to remember  that, with cascaded regulators, the current drawn from the lower voltage regulators will add to the current, and power dissipation, through all the higher voltage regulators up stream. I suspect you may be causing the 7815 regulator to go into thermal shutdown due to excessive power dissipation.

For example, if you draw 1A from the 7805 and 100mA each from the 7809, 7812 and 7815 the currents and power dissipations (ignoring each regulators own small bias current) will be:

7805   Iout = 1A        Pd = (Vin-Vout)*Iout = (9V-5V)*1A = 4W
7809   Iout = 1.1A     Pd = (12V-9V)*1.1A = 3.3W
7812   Iout = 1.2A     Pd = (15V-12V)*1.2A = 3.6W
7815   Iout = 1.3A     Pd = (36V-15V)*1.3A = 27.3W

All of these power dissipations are above the free air power dissipations limits of the 78xx devices so heat-sinks will be required to avoid the regulators shutting down due to thermal overload. In the case of the 7815 you will need an absolutely enormous heat-sink to handle this level of power dissipation. The problem is worsened by the 78xx regulators having a from of thermal overload protection which further limits the output current when operating with high input voltage to constrain the power dissipation to acceptable limits. In the case of the 7815 the input voltage you are applying exceeds the manufacturer's absolute maximum input voltage of 35V and so the the protection circuits will come into action for even lower output currents, causing the regulator to shut down.

The solution to the problem is to start with a much lower input voltage to the regulator chain. An input voltage of 20V would be more suitable for the regulators you are using and will significantly reduce the power dissipation although you will still need to pay close attention to heatsinking.

If you are stuck with the 36V input then you could use a 7824 regulator prior to the 7815 regulator. The 7824 is allowed to go up to 40V input and it will drop the excess input voltage to the 7815 to a more acceptable level and more than halve the 7815's power dissipation.

[james_s]

None of the regulators you list are LDO.

Cascading them is not really a great idea, because all of the current drawn from all of the outputs is going to load every regulator up the chain, if you pull 1A from the 5V output that will use up the full 1A budget that the 12V regulator can supply. You'll have to provide a schematic of the circuit you're using in order to get much help. How exactly are they wired up?

[wraper]

Apologies if this is answered in another thread.
I am cascading a number of LDO voltage regulators, 7815, 7812, 7809 & 7805, to create a general power supply for my workbench. However, the output from the regulators seems to be all over the place. If I install the 15-volt reg, the output is correct; however, as soon as I add another regulator downstream, the output from the first reg reduces, and the output from the second reg is low.

Most likely it overheats and reduces output current to protect itself. Many beginners think 1.5A current is the limitation but power dissipation is the actual limit. It can supply such current only if voltage drop across vreg is low enough and with sufficiently large heatsink.

I have changed the voltage regulators for different units just in case I bought fake units from ebay.

Most of 78xx sold on ebay are indeed fake. Most of them imitate ST but the vast majority actually have S7 logo, some proper looking logo but still fake. From what I've seen, they cannot supply rated current and are more like 78M05 (500mA max) but very unreliable.

[JustMeHere]

Get a bench power supply like this.  If not, a computer power supply will a handful of useful voltages.  However, a  current limit is essential for any bench power supply to help with magic smoke problems.

Tekpower TP3005P Programmable Linear Type Variable Regulated DC Power Supply, 0-30V at 0-5A https://a.co/d/jkxdGWg

[Terry Bites]

Dont chain.
Use high dropouts for the higher voltages fed from raw. Run your LDOs from them. If not the regulators nearest Raw will experience an unfair share of the total dissipation.
It'll be cheaper to build too. You might be able to get awya with lower current regulators for the higher voltages. Do the electric sums and do the costs too.

[wraper]

There is nothing wrong with chaining. It all depends on how you distribute power dissipation. For example you have 18V input and need 12V 0.2A and 5V 0.5A outputs. If chained 12V vreg will dissipate 6V at 0.7A = 4.2W, 5V vreg 7V at 0.5A = 3.5W. Which is quite reasonable with a heatsink. Then compare it with connecting 5V vreg directly to 18V. 12V vreg will dissipate 1.2W, while 5V vreg drop 13V at 0.5A = 6.5W. In this example chaining reduces stress on the Vreg by a lot. On other hand if 5V rail needed less current than 12V rail, it makes no sense to chain them. Of course these days it makes a lot of sense to use buck converters instead.

[james_s]

Either way, this whole design sounds pretty crappy. Starting with such a high voltage with linear regulators is silly, it's going to be massively inefficient. It's much better to use an input voltage only as high as needed, and better still to use a transformer with multiple windings so the lowest voltages are fed from a lower voltage winding than the highest. A variable design is also preferable in most cases to a bunch of fixed voltages.

[wizard69]

Either way, this whole design sounds pretty crappy. Starting with such a high voltage with linear regulators is silly, it's going to be massively inefficient. It's much better to use an input voltage only as high as needed, and better still to use a transformer with multiple windings so the lowest voltages are fed from a lower voltage winding than the highest. A variable design is also preferable in most cases to a bunch of fixed voltages.

this pretty much sums it up.   When I read the 36VDC value I immediately had concerns, it has been awhile but that seemed to be on an upper limit.    Cascading supplies is something that needs to be considered carefully, and probalby not a beginners avenue.

[james_s]

this pretty much sums it up.   When I read the 36VDC value I immediately had concerns, it has been awhile but that seemed to be on an upper limit.    Cascading supplies is something that needs to be considered carefully, and probalby not a beginners avenue.

I remember trying to use a 7805 on a device powered by rectified 24VAC a long time ago before I knew better. It was within the specified maximum but in practice it was hopeless. The regulator ran scorching hot and cooked everything in its vicinity. Later the Simple Switcher buck regulators came out and I rebuilt it around one of those and it worked great until I retired it for other reasons around 15 years later.

[Siwastaja]

Dont chain.

Bad advice. There is absolutely nothing wrong with chaining - do it whenever convenient. Chaining can help spread power dissipation over multiple devices and larger area, easing thermal design.

Of course, one needs to calculate if the regulators can dissipate the power required, but this calculation is done the same way regardless of chained or not.

P = (Vin - Vout) * Iout
Tj = Ta + P * RthJ-A

[CaptDon]

The part numbers you listed are not LDO devices. A 7812 cascaded from a 7815 probably isn't a good idea. If you want to go that route you need 'real' LDO devices.

[james_s]

3V of overhead ought to be adequate even for a conventional 78xx regulator. Especially when cascading so there won't be much ripple for the downstream regulator(s) to deal with.

[srb1954]

3V of overhead ought to be adequate even for a conventional 78xx regulator. Especially when cascading so there won't be much ripple for the downstream regulator(s) to deal with.

Most of the time you should be OK with a nominal 3V input-output differential but you might run into some problems if the upstream regulator is at the lower end of its voltage tolerance and the downstream regulator is at the upper end of its voltage tolerance. In that case you could end up with as little as 2V input-output differential.

Such a circuit would appear to work well enough at low output currents but as the current is increased the downstream regulator would start to drop out and you would see poor load regulation on that regulator output.

[David Hess]

You need some input capacitance and some output capacitance on the regulator, something like a 10uF electrolytic capacitor on input (doesn't have to be exactly 10uF, but anything higher than 100uF is unlikely to make a difference) and at minimum a 0.1uF ceramic capacitor on output is recommended.

Low dropout regulators, and negative standard regulators, require some bulk output capacitance for stability, however adding the typical 0.1 microfarad ceramic decoupling capacitor as well is both useless and can sometimes cause problems.  The small ceramic decoupling capacitors should be placed close to the load and not the source.

If the 12v and 9v and 5v regulators are very close to the 15v one, you'll be fine with just some 0.1-1uF ceramic capacitor on the input for these instead of the 10uF electrolytic.

That should be acceptable since these regulators require the ceramic decoupling capacitor at their input when the bulk capacitor is located at a distance.  By preference I would use the 10 microfarad aluminum electrolytic, or maybe a 1 to 2.2 microfarad solid tantalum capacitor.

[Siwastaja]

A 7812 cascaded from a 7815 probably isn't a good idea.

Nothing wrong with that. For example for LM78xx, 7815 is rated for minimum output of 14.40V. 7812 is rated for 2V typical dropout at Iout=1A. There is not much extra headroom beyond specifications, but it's fine. (It's a bit concerning that dropout is only specified "typical", of course.)