low voltage regulator bypass - is there a better way?

[m12lrpv]

I'm working on a battery powered circuit that I've tested down to 2.5v without a problem.
The whole circuit is made of components that can handle up to 5v but I will be running it at 3.3v via a 1117-3.3 linear regulator.
While the general current draw of the main circuit is very low. It can rise to as much as 30mA when the LCD which forms part of the project is switched on.
There are no issues with the main circuit.

What I am trying to do that has me asking for help is to bypass the regulator when the battery voltage drops to a certain level so as to extract every last bit of energy out of the batteries.

I've tested the circuit attached and it works and in fact works better than I initially expected when controlled by a micro (which it will be).


However I really want to reduce the 2mA to 3mA of wasted current through the base. If I could drop it down to 1mA or less I'd be happy


I looked at mosfets but they don't seem to be suitable for what i'm doing (I don't understand them well though).

Is there a better way or is the simple pnp bypass that i've been playing with the only way to go?

[free_electron]

so you are concerned about a little bit of drop in the transistor, but at the same time you burn all this energy in the regulator...
you're solving the wrong problem... chuck the linear regulator in the bin and move to a switching regulator. if you go for a sepic construction you can really suck that battery dry ...

[m12lrpv]

so you are concerned about a little bit of drop in the transistor, but at the same time you burn all this energy in the regulator...
you're solving the wrong problem... chuck the linear regulator in the bin and move to a switching regulator. if you go for a sepic construction you can really suck that battery dry ...

Regulator losses were going to be the next test just to see if I can live with it.

I took one look at all the extra components required to put in a switching regulator and really baulked at it.

[Rufus]

Is there a better way or is the simple pnp bypass that i've been playing with the only way to go?

Not using such a junk LDO would be a start. The Microchip TC1014 has about 85mV drop out at 50mA rather than about 1v from an LT1117 for example.

[Psi]

I took one look at all the extra components required to put in a switching regulator and really baulked at it.

It's not that bad, all you need is
- Switchmode IC (probably with built in driver) 
- Schottky diode
- Inductor
plus maybe a few caps and resistors to set the voltage and frequency

For example, quick google image search found this simple one.
The IC is even available in a 3.3v version (5v version shown)

[m12lrpv]

Not using such a junk LDO would be a start. The Microchip TC1014 has about 85mV drop out at 50mA rather than about 1v from an LT1117 for example.

Selecting parts I can actually get and use is a contributing factor.

I should also say that this is all through hole.

If I was experienced enough to be using SMD I wouldn't be asking these questions.

[amspire]

If you want to stick to linear regulators, you want one with a lower dropout voltage.

The LP2950ACZ-3.3 is pretty nice cheap regulator. TO92 package micropower regulator with about 0.28V dropout at 30mA. Available in SMD package. You would have to try one to see what happens when the input voltage goes down to 3V on the input - the data sheets are not clear, but hopefully, it is at minimum voltage drop. You could lower the circuit voltage a bit to 3V with the LP2950ACZ-3.0 or the lp2951 to give a bit more regulated headroom.

There the LM2940 regulators have 0.1V dropout at 30mA, but they only go down to 5V. I think there may be similar regulators that can do 0,1V at 3.3V, but I would have to search for them.

The 1117 family of regulators are really great regulators but they have a dropout closer to 1v at 30mA, so it is just not the right regulator for you to use.

Richard

[amspire]

You have to read the specs for switching regulators closely, as a lot of the down converters will switch off when the input voltage gets too low. The LM2674 suggested above will shut down when the input gets below about 3.8V (for the 3.3V output version) so it is no use at all.

Richard.

[m12lrpv]

If you want to stick to linear regulators, you want one with a lower dropout voltage.

The LP2950ACZ-3.3 is pretty nice cheap regulator. TO92 package micropower regulator with about 0.28V dropout at 30mA. Available in SMD package. You would have to try one to see what happens when the input voltage goes down to 3V on the input - the data sheets are not clear, but hopefully, it is at minimum voltage drop. You could lower the circuit voltage a bit to 3V with the LP2950ACZ-3.0 or the lp2951 to give a bit more regulated headroom.

There the LM2940 regulators have 0.1V dropout at 30mA, but they only go down to 5V. I think there may be similar regulators that can do 0,1V at 3.3V, but I would have to search for them.

The 1117 family of regulators are really great regulators but they have a dropout closer to 1v at 30mA, so it is just not the right regulator for you to use.

Richard

Thanks. I think I will stick with the linear's because looking at the others they just aren't practical as it seems I'd need one to step up then another to step down as the battery level passed the operating voltage.
Ha! You spotted that too.

The LP2950A is nicer than what I have with less than half the dropout voltage ( I measured 1117 at 0.86V) so thanks for that. I've only just discovered RS's free postage which has opened up my component options a lot.

[free_electron]

guys, low drop regulators don't do $#!%. you are still burning power in the regulator.
Regulator lesson :

if your battery is at 5 volts, you need 3.3 volt out and draw 30 milliampere then the power dissipated in the regulator is (5-3.3)*0.03 or 51 milliwatts.
What is low drop ? most regulators need about 2 volts across them in order to work correctly. A Low drop regulator can live with less. ( down to 0.5 volts ) so , if your battery has fallen to 3.8 volts you are still burning (3.8-3.3) * 0.03 = 15 milliwats.
and this is on a circuit that consumes 3.3 volts * 30mA = 100 milliwatts.

With a full battery you are wasting 50% energy ! ( 50milliwat burned, 100 used ). with a battery at 3.8 volt you are still burnig 15% energy.
And this is not even looking at the regulators own consumption  (called the quiescent current) most stupid linear regulaters eat around 2mA. there goes another 5 miliwat... ( baatery voltage * quisecent current.)

not very efficient....

There are very effective regulator chips that use a switching mode. usage is imple. they are 5 pin devices. all you need is a tiny inductor and two caps. they are even smaller than an ld1117..
if you want low quiescent current regulators take a look at the LK or LF series form STmicroelectronics.
if you want a high performance SEPIC ( buck boost : these can run on higher and lower voltage in than out and are designer to run off batteries. they can really suck batteries dry ... ) take a look at things like an LT1615 or LT1610. these things eat anything between 1 volt and 15 volt and produce whatever you want at the output. SOT23 package.
need 1 inductor , 1 diode and two little caps. ( 1uf) you can take 1uf in 0603 package. the inductor can be a ferrocore chip inductor in 0805 or 0603 package for such small currents. the diode can be in sod123 package. if you place the passives on the backside of the board you need even less place than an ld1117 ...

and these chips have a shutdown pin, draw less than 1 microamp in shutdown and quiescent current is like 20uA... perfect solution.

SMd is not something to be afraid of !

[amspire]

if you want a high performance SEPIC ( buck boost : these can run on higher and lower voltage in than out and are designer to run off batteries. they can really suck batteries dry ... ) take a look at things like an LT1615 or LT1610. these things eat anything between 1 volt and 15 volt and produce whatever you want at the output. SOT23 package.
need 1 inductor , 1 diode and two little caps. ( 1uf) you can take 1uf in 0603 package. the inductor can be a ferrocore chip inductor in 0805 or 0603 package for such small currents. the diode can be in sod123 package. if you place the passives on the backside of the board you need even less place than an ld1117 ...

It is not quite that simple.

You are describing a step up or step down converter.

The SEPIC converter circuit for the LT1610 involves 11 components including two inductors, compared to two components for a linear regulator.  Then if you get the layout or part selection wrong, you get a very noisy output which could affect the load, or radiate noise to nearby electronics. The right inductor can cost much more then a regulator IC in one off quantities.

Linear regulators are less efficient, but are as close as you can get to foolproof.

Technically you are totally correct, but I can understand wanting to get the best out of a simple and dirt cheap linear regulator instead.

Richard

[free_electron]

DRAT ! foiled by the marketing page of a datasheet. ( the first page of a datahseet is the typical marketing bullshit. you need to readon to see the real limits... they always show a minimal part case that turns out unusable)

ok here is some others : those are all 2 caps 1 inductor ...
TI TPS6300x series.
TI TPS6303x series
ST STBB1xx

google for 'single inductor buck boost' and you will find plenty. looks like 'sepic' is 'stone age technology' .....

[m12lrpv]

Unexpected noise from switching supplies is an issue for someone that doesn't have an oscilloscope.

SMD is an issue for a beginner that is making a one off item for personal use and not a production item that justifies the additional time, effort and cost of producing the necessary board to solder all the SMD stuff to not to mention the rework cost and effort if I screw it up which as a beginner I undoubtedly will.

Ruling out SMD is not a fear issue. It's a practical issue. 

[Mechatrommer]

Ruling out SMD is not a fear issue. It's a practical issue. 

but, but... with current "technology of diy", i believe smd is now viable. doesnt mean ruling out TH, but we have both option. now i much prefer soic version due to compactness and no need to punch hole on my diy pcb. ordering pcb from fab house for 1-off is another dreading experience, but thats maybe just me. 2cnts.

[Kremmen]

[...] that justifies the additional time, effort and cost of producing the necessary board to solder all the SMD stuff to not to mention the rework cost and effort if I screw it up which as a beginner I undoubtedly will.

[...]

What effort and cost would that be then? Producing the board is no more difficult for smd and one could argue that it is actually easier (no drilling for through hole). Of course there will be drilling if you make a 2 sided board but use of surface mount devices should not be an issue. It is different but if you start easy, keeping to 1206 size and larger, SOIC packages and so on, everything will be doable with regular through hole soldering tools. I made the switch to predominantly smd some years ago and never looked back.

[free_electron]

Exactly. There is no additional cost , apart from a magnifier and a fine tip for your soldering station.
There is a whole world waiting out there. And most new and cool parts are SMD only.. DIL is going the way of the vacuum tube and the dinosaur ....

if you stick to 1206 or 0805 they are perfectly manageable.
If you invest in a stereo inspection microscope you can go all the way down to 0402 and even 0201.

boards can be made cheap at iteadstudio / seeedstudio and other outfits.
no drilling needed. i just made a circuit about half an hour ago. i have a pulsarpro PCB-fab in a box kit. print layout on special paper , send through laminator soak in water.
soak a spunge in some Fe3Cl , put pcb and spunge in ziplock bag and 'dab' spunge on pcb. 3 minutes later :board. done.
this afternoon i will aseemble and test it. if it works : push to real pcb  ( i need multiple channels. this is just a quick proof-of-concept slapped together in half a day ) A little digitally controllable load. 0 .. 5 ampere driven by a 12 bit dac. the real system needs 12 of those channels. before i design the big board ( 45 cm x 25 cm ) with a bunch of other stuff on it i will test this little slap-together on bench. according to LTspice it works ... but the proof of the pudding is in the eating. i always have a weary eye on simulation that look too good.. there may be something mising in the model ...

http://www.pcbfx.com/main_site/pages/start_here/overview.html

[m12lrpv]

I can do a through hole project right now with what I have. A basic 25W cheap soldering iron from the local electronics store.

Given all this advice that there's no additional cost for smd can anyone promoting that please instead advise where beginners can get all the FREE soldering stations, hot air rework stations, board etching supplies and endless supply of boards and components for when they need to be scrapped for the usual beginners mistakes and changes.

I just made a power supply box that i'm really happy with (I posted about it before) and with the changes I made to it after soldering up original I would have had to made it 4 times over at 4 times the cost if it was in SMD.

How that reality isn't completely obvious to some people is astounding.

I even love people posting that it doesn't cost extra, then go on and list the soldering station, magnifier and link to a  $75 pcb etching kit. 

[T4P]

Would you need to spend on a pcb etching kit ? You still do on a through hole pcb

I can do a through hole project right now with what I have. A basic 25W cheap soldering iron from the local electronics store.

Given all this advice that there's no additional cost for smd can anyone promoting that please instead advise where beginners can get all the FREE soldering stations, hot air rework stations, board etching supplies and endless supply of boards and components for when they need to be scrapped for the usual beginners mistakes and changes.

I just made a power supply box that i'm really happy with (I posted about it before) and with the changes I made to it after soldering up original I would have had to made it 4 times over at 4 times the cost if it was in SMD.

How that reality isn't completely obvious to some people is astounding.

I even love people posting that it doesn't cost extra, then go on and list the soldering station, magnifier and link to a  $75 pcb etching kit. 

.
A 25W firestick only gives you shitty joints no matter your skill,
You need to drill a through hole board and that introduces extra costs,

Given all this advice that there's no additional cost for smd can anyone promoting that please instead advise where beginners can get all the FREE soldering stations, hot air rework stations, board etching supplies and endless supply of boards and components for when they need to be scrapped for the usual beginners mistakes and changes.

Free soldering stations ? If anyone was a proper beginner he would have a proper soldering station.
And SMD parts are cheaper actually, on top of that, it's all part of the fun to make your own pcb.
There's not much experience to be had using protoboard or vero or whatever.

[free_electron]

Given all this advice that there's no additional cost for smd can anyone promoting that please instead advise where beginners can get all the FREE soldering stations, hot air rework stations, board etching supplies and endless supply of boards and components for when they need to be scrapped for the usual beginners mistakes and changes.

You don't need soldering stations or hot air rework stations, not until you start messing with really big componentns with hundreds of pins. I solder tqfp240 pins by hand using the rolling ball method. no hot air rework station needed for that.

You have a soldering iron you dsay .. get a 2 $ fine tip for it and start off with that. the fancy pants stations can wait for later.

As for parts : where do you get parts now ? SMD parts are typically cheaper than through hole parts. unless you are restricted to the local 'parts-farmer' at the corner.... those guys specialize in valves and 20 watt resistors... any decent parts store is well stocked on SMD these days. If you order parts by mail order you can have a ball ...

I even love people posting that it doesn't cost extra, then go on and list the soldering station, magnifier and link to a  $75 pcb etching kit. 

you can get away with a simple magnifying benchlamp/loupe costing a few dollars. there are even magnifying  'glasses' you wear like normal glasses. dealextreme and others sell these for a few dollars. ideal to get started.
http://www.dealextreme.com/p/eyewear-style-eye-glasses-1-5-2-5-3-5-magnifier-with-led-light-3-1130-batteries-35607

as for pcb kit : how do you make boards now ? you can use the same technology. of course if you are only using predrilled array-of-holes (breadboard , although breadboard for me is this white plastic things you just plug parts in ) boards then SMD is a bit of a problem.... although there are such arrays for smd parts as well.

[m12lrpv]

If anyone was a proper beginner...

Seriously... is there any continuing after this???

[baljemmett]

as for pcb kit : how do you make boards now ? you can use the same technology. of course if you are only using predrilled array-of-holes (breadboard , although breadboard for me is this white plastic things you just plug parts in ) boards then SMD is a bit of a problem.... although there are such arrays for smd parts as well.

Yep, or depending on the packages in question even a bog-standard piece of stripboard might do the job in a pinch.  Plus there's always dead-bug style construction; I think Dave did a video on that a while back...  For something like the REG710 buck-boost regulator in a SOT-23-6 package I wouldn't have thought that was too onerous, and you're out, what, a buck or two for some parts to experiment with and half an hour's playtime.

[DavidMenting]

And otherwise you can always use a FlowerPad proto board. It's specifically made for mixing up through hole and smd components.
See: http://www.elecfreaks.com/store/flowerpad-noplated-protoboard-225x68mm-p-388.html


I found smd soldering quite easy. Knowing how to also allows you to repair more modern electronics. The mental barrier seems the largest. Just get some 0.5mm solder, a pair of tweezers and an optional medium-sized chisel tip for your soldering iron and you're set. No need for the fancy magnifiers, fluxes, rework stations etc. until you really ramp it up.