Testing boost converters, any recommendations?

[Circuitous]

I've been working on a small wireless device that spends most of its time asleep.  I'm looking for recommendations for a boost converter to provide about 20 uA steady, with bursts of 15 mA from one or two AA cells. 
(edit) I need a steady output of 3.3V  +/- 0.1V.  The device will spend about 99.5% of the time in sleep mode.

What are your favorite boost converters?

I tried out a few in the (exceedingly long) video below. 
(edit) I tested the NCP1402, AAT1217, and TLV61220.  The NCP1402 worked the best, but still only got about 45% efficiency when providing 22 uA, I'm hoping someone knows of one with better efficiency in the microamp range.

I challenge you to stay awake until the end... no surprises, just wondering if anyone can make it all the way through.

[Circuitous]

I've been browsing Digikey for some other options, and found a few more:

• Exar SP6641A, modest price, but requires an external schottky diode.
• Intersil  ISL9111A, expensive, looks like a good option.

Does anyone have experience with these chips that they can share?

Thanks!

[PeterFW]

but still only got about 45% efficiency when providing 22 uA, I'm hoping someone knows of one with better efficiency in the microamp range.

I was searching for something like that too and dit abandon the search.
The one i am using for now is a MCP1640 to step up 2AA cells to 3.3V,
It runs at about the same efficiency in PFM mode at 20µA (IIRC) and is very cheap.

The LTC3440 may be a bit more efficient but you need to enable the PFM mode with the MCU, i have it running of a single LiIo cell as buck/boost set a 3.3V.
But it is expensive.

Both might be a bit out of the +-100mV in PFM mode, they will only properly regulate the output voltage in PWM mode.

For future test i have the TPS630-something on my list, can not find were i put the datasheet right now...

[paulie]

Hard to beat the ready-to-go one cell Ebay USB booster modules that, with a little shopping, can be had 2 or three for a dollar. Adding a single chip resistor brings it down to 3.3v and if super low noise is a concern a cheap 220uf cap fixes that.  I suspect this solution may not be labor intensive enough or costly enough for many of the players here.

[PeterFW]

Hard to beat the ready-to-go one cell Ebay USB booster modules that, with a little shopping

You are joking, right? I will eat my hat if they run under 100 micro amps IQ and have a decent efficiency under 1mA.
The last chep china USB converter i tested was a 100kHz triangle generator, calling that a voltage regulator is a joke.

[paulie]

LOL.

I suspect this solution may not be labor intensive enough or costly enough for many of the players here.

[Circuitous]

@peterfw - thanks for the suggestions.  The MCP1640 looks like a good option, and it is cheap.  When I searched on it, another one came up, the MCP1625 which also looks like a good contender.  I didn't really like the Exar that I found.

@paulie - I had looked at the ebay boosters, but it always looked like potluck when ordering.  I really didn't know what I would get, and it is likely not a consistent product.

My plan is to build a simple integrated board with the boost converter, microcontroller and radio or headers for a plug-in radio module like the nFR24L01 (profiled in an earlier video).

I'll order up a few more chips and try them out.

[PeterFW]

Quote from: Circuitous on Today at 04:35:36 AM

@peterfw - thanks for the suggestions.  The MCP1640 looks like a good option, and it is cheap.  When I searched on it, another one came up, the MCP1625 which also looks like a good contender.  I didn't really like the Exar that I found.

When ordering the MCP1640 be shure to get the right variant, there are four or them.
Can you link me the MCP1625? I am unable to find it.

Btw... that was the first and only chinese usb boost converter i ever tried:

[Circuitous]

Wow!  That ripple is a bit high!  And, that's why I haven't used any of those ebay specials.

Here's the other microchip device, MCP16251/2 (I left off the last digit.)
http://ww1.microchip.com/downloads/en/DeviceDoc/25173A.pdf

[PeterFW]

Here's the other microchip device, MCP16251/2 (I left off the last digit.)
http://ww1.microchip.com/downloads/en/DeviceDoc/25173A.pdf

Oh, neat!
Pin compatible, layout compatible and vastly better efficiency.
I should give that a try too  only 64 Cent in single quantity, main difference seems to be the PFM/PWM threshold.
The MCP16251 will switch to PWM at a higher current, you may want to use the MCP1640 if you need PWM regulation at 15mA load.

[NiHaoMike]

Have you considered using a charge pump?

[Evil Lurker]

Have you considered using a charge pump?

Yeah for the max load I think that would be entirely do-able.

[PeterFW]

Yeah for the max load I think that would be entirely do-able.

They tend to have a horrible idle currents and only work of fairly high input voltages.
But TI has a single part that looks good:
http://www.ti.com/product/tps60310

But the numbers do not look that good at first glance compared to a inductor step up converter.

[Circuitous]

I had looked at a couple of charge pumps for a different project sometime ago, and they didn't look good.  But, after a much more thorough search today for this specific project, and the link from peterFW, I have found a few candidates.
Some can provide output currents as high at 100 ma, which will provide flexibility on any sensors I add to the wireless node.

I just ordered some other inductor based converters yesterday, now it's time for an order of charge pumps. I selected a couple from TI and one from Diodes Inc.

I'll just have to try them out to see how they work during the sleep phase.

Thanks for the suggestions.

[paulie]

Btw... that was the first and only chinese usb boost converter i ever tried:

Coupled with your previous comments implying more experience than this I suggest you might be giving an unfair impression. First you seem to be confusing the charger modules with the buck converters. They use completely different chips.

In my day job at 3rd party QC/QA agency several thousand of both types purchased and tested for clients in addition to a few dozen for personal use. A rare DOA but none showing nearly a volt of ripple on the 3.3v like in your photo. Not even 1/10th that. And as mentioned before a 2 cent cap will fix what remains for those 1% applications that need close to zero ripple:



I've built dozens of similar projects with no hint of ripple on the output. At least not visible on a scope let alone 3/4 volt. I think you may not be using gear properly or just got a bum unit. I can say in my experience defects were in fraction of a percent so IMO you are not giving these a fair shake whatever the reason.

[peter.mitchell]

I had looked at a couple of charge pumps for a different project sometime ago, and they didn't look good.  But, after a much more thorough search today for this specific project, and the link from peterFW, I have found a few candidates.
Some can provide output currents as high at 100 ma, which will provide flexibility on any sensors I add to the wireless node.

I just ordered some other inductor based converters yesterday, now it's time for an order of charge pumps. I selected a couple from TI and one from Diodes Inc.

I'll just have to try them out to see how they work during the sleep phase.

Thanks for the suggestions.

If you are really trying to eek out idle efficiency, i suggest using tantalum caps instead of ceramics for lower leakage, slightly oversizing the inductance to reduce ripple current may also help.

[PeterFW]

First you seem to be confusing the charger modules with the buck converters. They use completely different chips.

Mine looks a bit like the one in your picture.
No confusion, you might be right that i may have been sold a defective unit but i dit not confuse annything.
Were are speaking about boost / step up regulators, something that takes <5V in and will output 5V.
I ordered two of those of ebay sold as "5V DC boost Converter", i am not the brightest bulb in the chandelier but i can assure you i can measure a simple dc voltage.
Both of them had the same sawtooth output, maybe they are not 5V regulators but something different but they are sold as boost converters and i tested them under various loads.

The threadstarter has a input voltage from 2 to 3V and wants a 3.3V output.
You suggested modifying the feedback network of one of these chinese 5V boost regulators to achive 3.3V output.
Nothing wrong with that, but...

You do not know what you buy, you have no datasheet of these devices, you do not know what idle current they have, no idea about the efficiency, you do not know annything about them.
Shure you can buy one of each wich are currently available, test them, evaluate them and maybe out of 10 modules one has the specs you need.
Then you set out to buy a crate of them and the seller no longer has them in stock, there may be another one on alibaba wich looks the same but you do not know if it is.

We are speking about a battery powered device with runs in a MCU controlled power down mode that draws <50µA. Something wich needs to run extendet periods of time on a battery.
The best regulators from name brand companys have idle currents (quiescent current) lower then that and a efficiency above 50% in that load range.

Why should a random boost converter you pluck of ebay, alibaba, ali express or what not meet those narrow parameters?
And... one other thing, if you build something you may want everything on one neat PCB, neat and tidy, low profile, known dimensions. No chance dooing that with the china regulators.

[Fungus]

Combine a microcontroller (eg. Tiny85) and a MOSFET with one of those cheapo eBay converters. Job done.

The Tiny85 can wake up, power up the boost converter via its MOSFET, then wake up the main chip when the voltage stabilizes.

I built a system like that two years ago that runs off a 3V CR2450 coin cell battery.

I measured the battery voltage last week and it said "3.03V". I kid you not. It's been running for two whole years and the voltage hasn't dropped by a measurable amount. Sure, it doesn't do much work when it powers up, but still...

Edit: A Tiny85 will use under 10uA in sleep mode using an internal timer to wake up periodically.

[paulie]

Combine a microcontroller (eg. Tiny85) and a MOSFET with one of those cheapo eBay converters. Job done.

I measured ripple but must admit to not checking on idle current for the cheapie modules. That battery in my photo on previous page sat in the toolbox for over a year and still read 1.25v and after couple weeks of use hasn't needed recharge yet. Being LSD type helps and of course there is a power switch but even with my constantly powered projects those modules do an excellent job. No MOSFET needed in my case because the switcher chips in those do have a shutoff pin. Mega8 don't quite get down to Tiny85 idle but not far from it. And no coin cell but some of my lithium AAA powered circuits still running unattended after 3 years.

[Circuitous]

@peter.mitchell - thanks for the suggestion on the tantalums, I'll throw a few in on my next digikey order (today or tomorrow).  I just got another stack of various inductors, so I'll try bumping up that value as well on the next tests.

@fungus - Impressive run time with your ATtiny.  The approach of running a low voltage micro directly from the battery and having it turn on the boost if needed for other parts of the circuit has been an option.  I was planning to use a PIC 18, but had deferred this approach.  I had through about a watchdog chip to perform the sleep/wakeup... but hadn't thought about an ATTiny... maybe a PIC10?

I was hoping to initially pull this off with just using the Arduino and a boost circuit, mostly to share with others wanting to create similar wireless nodes that have a long battery life.  My final intent has been to build a board with an integrated boost and PIC micro, and the header for the nRF24L01, along with some other sensors and I/O.

Thanks! Keep the suggestions coming.

[PeterFW]

@fungus - Impressive run time with your ATtiny.  The approach of running a low voltage micro directly from the battery
(...)
Thanks! Keep the suggestions coming.

There may be a simpler solution to this aproach, a regulator with bypass mode.
You can swith them between two operation modes, boost or bypass.
In boost mode it works a a regular boost regulator, in bypass mode it just switches the input voltage through.
If you MCU can work with low voltages you could try that.

Code: [Select]

4.1.4 INPUT BYPASS MODE OPTION
The MCP1640C/D devices incorporate the Input
Bypass shutdown option. With the EN input pulled low,
the output is connected to the input using the internal
P-Channel MOSFET. In this mode, the current draw
from the input (battery) is less than 1 ?A with no load.
The Input Bypass mode is used when the input voltage
range is high enough for the load to operate in Sleep or
low IQ mode. When a higher regulated output voltage is
necessary to operate the application, the EN input is
pulled high, enabling the boost converter

Thats only 1µA during sleep.

[peter.mitchell]

@fungus - Impressive run time with your ATtiny.  The approach of running a low voltage micro directly from the battery
(...)
Thanks! Keep the suggestions coming.

There may be a simpler solution to this aproach, a regulator with bypass mode.
You can swith them between two operation modes, boost or bypass.
In boost mode it works a a regular boost regulator, in bypass mode it just switches the input voltage through.
If you MCU can work with low voltages you could try that.

Code: [Select]

4.1.4 INPUT BYPASS MODE OPTION
The MCP1640C/D devices incorporate the Input
Bypass shutdown option. With the EN input pulled low,
the output is connected to the input using the internal
P-Channel MOSFET. In this mode, the current draw
from the input (battery) is less than 1 ?A with no load.
The Input Bypass mode is used when the input voltage
range is high enough for the load to operate in Sleep or
low IQ mode. When a higher regulated output voltage is
necessary to operate the application, the EN input is
pulled high, enabling the boost converter

Thats only 1µA during sleep.

You still may need to use a switch because of leakage from the wireless module with the pass through mode, and it may have undefined start-up behaviours going 1.8~3v -> 3.3v, check that out first.

Also, sometimes the wireless module may have internal linear regulation, from 3v3 down to 2.5 or 1.8, you may save additional battery life by removing the internal regulator, running both the micro and wireless from a small buck-boost and using a p fat to turn the wireless module off - this obviously isn't practical if you're making 100+ units though.