Running uC off 1v5 to 15v range

Go To Last Post
28 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I would like to be able to run an AVR at either 3v3 or 5v (don't mind either way but it probably makes no difference to the problem) with an input range to a regulator of 1v5 to 15v.

There are buck/boost regs but AFAIK they all have a max of 5v.

The best idea I can come up with so far is a linear or buck reg AND a boost reg, parallel them and diode OR the outputs. Another way is to buck down to ~2v and internally connect that to a separate ~2v input and boost back up to 3v3 or 5v.

I don't mind having two separate physical inputs, but one would be better.

Of course I want it to be free and use no components :)

Why? I'm designing a dongle that can connect to batteries in a battery bank (as in large solar power banks) and the three common battery sizes are 2, 6 and 12 volts.

6 and 12 are no problem, but 2 is a PITA and I don't want to have two versions of this gadget.

_____
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

If you can live with lower limit of 4V, one of the megaXXHVB series should do the trick. Rated for 4-25V vcc....

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Sorry, gotta be 1v9 at the most, 1v8 would be better. If the dongle is monitoring a 2v battery I need it to stay alive if the battery dips to that sort of level.

What I'm now thinking is a say 5v linear reg followed by a buck/boost reg to get back to 3v3/5v.

The REG711 buck/boost reg has an input range of 1v8 to 5v5 and produces 3v3 output.

When I have > 5v available the first reg limits the second reg's input to 5v.

As the input drops below the 5v + headroom I think most linear regs will follow the input for a long time, thus I effectively have an input that rises up to 5v then stops there.

If my thoughts about the linear regs doing that are correct then I effectively have what I need. The only catch is the voltage drop through the linear reg, that increases my affective input voltage.

I'll have a look at the HVBs anyway.

EDIT: Sorry kherseth, yes the processor can be any normal voltage, it's the input regulation I'm have a problem with.

_______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

The v or p versions can operate down to 1v8.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Could this or this be of any help?

-Pantelis

Professor of Applied Murphology, University of W.T.F.Justhappened.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks pnp, the tpic74100 looks hard to find but the tpic74100 has a reasonable price and is available. At first glance they both seem to need a lot of support components.

I'll have a good look at the data sheets, maybe they aren't all needed. For example the input L, C and diode, maybe that's just for filtering I don't care about. Given that they don't have values that might be the case.

Is there a 3v3 version, so far I can't find one.

I love the 1v5 to 40 input range, that's pretty impressive.

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I don't remember seing a 3v3 regulator with a similar, wide input range, but it is not impossible that there is one. My first step would be a quick look at the websites of the manufacturers that have a power product line, like TI above and Maxim and a few others. There I would search for buck/boost regulators with low Vin as the first sorting criterion.

Depending on your current demands, you could get away with a simple 3v3 3pin (or similar) LDO regulator as a post regulator. This way your power supply will also get "cleaner".

As for the parts, well, it is not *that* bad, I have seen worse! :) The LC parts without names are subject to this, from p.24 of the datasheet:

Quote:
B. The L and C component values are system application dependent for EMI consideration.
I would not suggest leaving them out.

-Pantelis

Professor of Applied Murphology, University of W.T.F.Justhappened.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A Linear Technology LT1615-1 boost voltage regulator followed by a LDO linear voltage regulator.
Linear Technology also makes a few other boost regulators that meet Rob's specs.

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
EMI consideration.

I thought that might be the case.

I've just spent some time looking (plus I looked before) and it seems any 3v3 versions don't have that input range, I found quite a few but they are are limited to ~5v input.

Quote:
Depending on your current demands

< 50mA, maybe 20-30.

5v output might be OK, I haven't settled on a processor yet.

Can you verify that all (most) linear regs just follow the input voltage when it gets below the set output. For example if you have a 5v reg and the input moves down to 4, 3, 2 etc does the reg output follow that (with allowance for some voltage drop)?

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

A LDO should follow its input, minus dropout, to its minimum input voltage (1.6v for one LDO).

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

@gchapman

I found a lot of boost regs (just grabbed the LT1615 data sheet, I'll look at that as well) which are good for the 1v5 end of the range. What I can't figure out is what happens when the input gets higher than the set output.

For example that LT1615 says it can handle 34V input, if you have it configured for say a 5v output (boosting from 1v5) what happens when the input goes to 34v?

If it just ramps up to follow the input I can live with that by adding a post regulator.

Every time I find something that states it can do X and Y I find that it can't do both at the same time (except those chips pnp found). Also all the data sheets assume you need a regulated output (I guess that's normally the point :)) so they don't state what happens outside those bounds.

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
A LDO should follow its input, minus dropout, to its minimum input voltage (1.6v for one LDO).

Thanks, then that's another option. Use a 5v LDO that can handle say 15v input then add a buck/boost reg like the Burr brown REG711.

With high voltage the REG711 sees 5v and bucks down to 3v3, as the voltage drops below ~5v the LDO starts to drop until it gets to ~3v3 at which point the REG711 starts to boost back to 3v3.

End result, ~1v? to 15v+ input, 3v3 output.

The only hitch is the dropout effectively adds to the lower input, some LDOs have very low dropout though, especially at < 50mA.
______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Graynomad wrote:
... what happens when the input goes to 34v?
Output follows the input minus diode drop and inductor's resistance drop. When the output voltage will exceed (come out of regulation), the boost regulator's switch pin will become inactive. Some boost regulators have an internal p-device to perform the diode's function; these may or may not have a pass-through mode (ex. LTC3525D-3.3, a 1v to 5v input boost regulator to 3.3v).

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Have you considered an ATMEGA128RFA1?
(it'll run full speed down to 1.8v spec)
Or, XMEGA is almost as fast down to 1.6v.
Or, AVR32 UC3L (core and I/O at 1.8v).
This assumes you can do your I/O at about 1.8v.

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
the boost regulator's switch pin will become inactive

Yes I read that in one of the data sheets and thought that might be the case, looking at the internals I figured that just leaves the inductor and diode in series with the power and they both conduct DC so the OP should follow the IP.

Looks like there are a few options, I'm just packing up to head out across the Nullarbor planes, 1200k of little or no phone so I'll be out of touch for a while.

I've downloaded a 1000 data sheets, hopefully I have enough information to do the design while I have some quite time in the desert.

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
Have you considered an ATMEGA128RFA1?...

Nope, more data sheets to download before I go :)

I still need the regulation though...I'll look at the IO but I think it's 3v3 min.

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Graynomad wrote:
With high voltage the REG711 sees 5v and bucks down to 3v3, as the voltage drops below ~5v the LDO starts to drop until it gets to ~3v3 at which point the REG711 starts to boost back to 3v3.
That means that you will be running the preregulator (the 5V one) below specs and you will have to do some trial and error, unless you design a way to switch the supply input to the REG711 directly when it's low.

Now, maybe a silly idea, but couldn't you use a Zener as a simple preregulator? It will limit the input voltage when it's above its Zener threshold and it will have no effect when the input it's below that, so that REG711 can take over. I have never played with Zeners, surprisingly, so I am not sure if something escapes my logic, but I don't see why not? Maybe someone with more experience could verify or correct me (please?).

-Pantelis

Professor of Applied Murphology, University of W.T.F.Justhappened.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Zeners at lower voltages are "very" leaky which is not good for sleep mode on a cell or most batteries.
Also the zener impedance makes it a so-so voltage regulator.
But your clamp idea can be implemented with a shunt voltage reference/regulator like LMV431/TLV431 for lower voltages and LM431/TL431 for higher voltages; there are a number of others besides those.
Edit: Oops! TLV431 needs 55 micro-amps to regulate. A lot of series (instead of shunt) regulators will quiescent and ground current at much lower than 55uA.

"Dare to be naïve." - Buckminster Fuller

Last Edited: Mon. Nov 21, 2011 - 01:30 AM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ah yes, your points make sence. I only meant it as a pre-regulator, but that can't help the leak problem. I haven't thought of those more specialized shunt references/regulators, so I will also have a look on them (for future reference). Thanks!

-Pantelis

Professor of Applied Murphology, University of W.T.F.Justhappened.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Not sure if someone has suggested it yet, but how about using a SEPIC converter (i.e. buck/boost converter)? However, this will require a few external components (including two inductors!), but would definitely be the most efficient way.

Links:
http://en.wikipedia.org/wiki/Sin...

As an example, here's one that handles Vin=1V to 10V: http://www.linear.com/product/LT...

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Linear also make buck/boost, non-SEPIC convertors.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

SEPIC is also likely more reliable (fewer semiconductors).
A cost is extra noise due to continual switching (may not be a problem).

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Many boost converters will do the job.

The trick is that you use a transformer instead a simple inductance and make the converter to work as flyback.

Two examples:
http://cds.linear.com/docs/Datas...
http://www.micrel.com/_PDF/mic25...

This inductor/transformer would do the trick
http://www.cooperindustries.com/...

I just picked the inductor, depending on the frequency used you can pick one more suitable. Would be good if you can find something with 1:2 ratio but 1:1 would be OK for a max 15V input.

Let me know if you want more details.

I doubt you can go without any inductance with a reasonable efficiency over a such wide input voltage.

George.

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

I'm back online for a couple of days, thanks for all the responses.

Quote:
that means that you will be running the preregulator (the 5V one) below specs and you will have to do some trial and error,

But once a particular part is found to work others of the same type should to the same thing I assume. However this does presumably add the dropout voltage to the mix which in turn adds to the minimum input voltage the circuit can handle.

Quote:
unless you design a way to switch the supply input to the REG711 directly when it's low.

Yeah I thought of that, too many components. One way to do it is have two physical inputs one for 2v and the other for 6 & 12v. But as soon as you do that someone will connect the 2v IP to a 12v battery.

Quote:
55 micro-amps to regulate

I'm not sure that's a problem, I'm talking 100-1000Ah batteries, although I admit less is good but if someone leaves a light on for 5 minutes longer than needed that will chew more than a week's worth of quiescent current for my gadgets I would think.

That said there could be 48 or more of these, so it all adds up.

Quote:
but how about using a SEPIC converter (i.e. buck/boost converter)?

That would be ideal but I can't find one that will do that range (except the TI ones pointed to by pnp), hence the talk of a pre regulator.

My current plan already uses a buck/boost reg (the REG711) but it can only handle 1v8 to 5v input.

Quote:
A cost is extra noise due to continual switching (may not be a problem).

Might be a problem as this is essentially an A/D circuit, proper PCB design may handle that.

@George

Quote:
MIC2570

I've been looking at that chip, despite the blurb saying it can handle 1v3 - 15v input not one of the suggested circuits shows an input voltage that low and at no point do the data sheets mention what happens if the input goes higher that the output. IE if you set it to boost to 5v and connect to a 6v battery.

gchapman said a few posts back that the output will follow the input (minus diode drop etc) and certainly that's the way I would see it working. In which case a post regulator is needed.

The way I see it so far there are three options

a) pass input through then clip at 5v -> boost to 5v
b) boost up to an input of ~6v then pass through -> linear regulate down to 5v
c) shit-hot boost/buck reg like the TI chips that can handle the input range

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Graynomad wrote:
I'm not sure that's a problem ...
(w.r.t. shunt regulator) You may be right especially when consider the batteries' leakage currents. But, vampire loads can be a problem in automotive use since recharge interval and duration vary. For a solar use may be OK as long as one has Sun.
Graynomad wrote:
That would be ideal but I can't find one that will do that range
(w.r.t. SEPIC) The LT1615's datasheet shows a SEPIC mode and its maximum input voltage is 15v. Linear Technology does have other parts that can go your required input range. Since automotive is a common use likely other chip makers have something.
Graynomad wrote:
Might be a problem as this is essentially an A/D circuit ...
Hopefully the power supply ripple rejection spec for the A/D will be OK else its reconsider a post-regulator and/or add a passive filter. At least the first regulator will reduce the power dissapation for the post-regulator and the post-regulator can add/fine-tune current limiting.

"Dare to be naïve." - Buckminster Fuller

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Quote:
I've been looking at that chip, despite the blurb saying it can handle 1v3 - 15v input not one of the suggested circuits shows an input voltage that low and at no point do the data sheets mention what happens if the input goes higher that the output. IE if you set it to boost to 5v and connect to a 6v battery.

As I said, many boost regulator can do the job. You just have to connect them in a SEPIC configuration.
I attached two pictures I took from the LT1615 and MIC2570. Both of them are connected in a SEPIC configuration and can handle up to 15V input and to deliver 3.3V output.
Also, flyback is another alternative.
I suggest to have the both inductances on the same core (as Micrel suggest) for space, price and energy transfer reason.
If noise is a problem, then a filter or a LDO (from 3.5 - 4 to 3.3V) is needed.
Quote:

gchapman said a few posts back that the output will follow the input (minus diode drop etc) and certainly that's the way I would see it working. In which case a post regulator is needed.

Actually he was referring to linear regulators, not SEPIC:
Quote:
A LDO should follow its input, minus dropout, to its minimum input voltage (1.6v for one LDO).

The drop voltage will be accomplished by the C2 (Micrel schematic, C3 LT schematic). You want a lower capacity for this capacitor because at power on (12V), will create a spike on the output voltage that will surpass 3.3V.
Since inductance do not allow a DC voltage at its leads, Uin = Ucap + Udiode(forward) + Uout. Since you may want to have Uout > Uin as well, a nonpolarized capacitor is needed as LT suggest.
Because LT one is working at higher frequency, a smaller capacitor and inductances are needed, but the current is lower and price higher.
George.

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Thanks George.

I guess the Micrel chip looks better to me at first, fewer components and as you say cheaper. With maybe 48 of these in an installation $1 saved makes maybe $150 difference at retail. OTOH with smaller components you may get some of the savings back and the 1615 is in a SOT23 package as opposed to a SOIC8 so all in all the LT option will be a lot smaller. I'll research the price difference.

I still haven't settled on a processor but all candidates run at 3v3 anyway so I think running this at 5v then post regulating down to 3v3 will be the trick.

Quote:
For a solar use may be OK as long as one has Sun.

The target is really solar and as such there is always a backup plan for sunless days. For auto use I'd have to beef up the input to handle load dump etc so I think I'll stick with solar as the target market.

Quote:
Hopefully the power supply ripple rejection spec for the A/D will be OK else its reconsider a post-regulator and/or add a passive filter.

Yep, see above, the post regulator should deal with this.
______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Looks like the LT1615 is a non-starter, it has an absolute max of 15v and in a solar charging scenario 15v or more will be a regular event with 12v batteries.

So it looks like the MIC2570 (18v max) will be the way to go if I use this approach.

______
Rob

Scattered showers my arse -- Noah, 2348BC.
Rob Gray, old fart, nature photographer, embedded hardware/software designer, and serial motorhome builder, www.robgray.com