AVR Freaks

Level: Hangaround

Joined: Fri. Aug 5, 2005

Posts: 373 View posts

Location: Tacoma, WA

Posted by jimlake: Sun. Jul 10, 2016 - 09:57 PM

My devices are portable, battery operated, and I have typically used 9v alkaline or 2S, 3S lithium batteries and linear regulators. On the latest device, I decided to try 2-AA alkaline batteries and a 1.5A TPS61252 boost regulator producing a 3.3v output. Is is small, inexpensive (though a lot more than a LM317) and very efficient. I used TI's (Nat Semi) automated workbench to design the regulator, and I was careful to lay it out on the pcb exactly according to the suggested layout in the data sheet. I always liberally place bypass caps on all power pins. Although I used a 2 layer pcb, I used 25 mil power traces and kept them as short as possible, and I used a grounded copper pour on the bottom layer. It should have been successful, but it wasn't.

It appeared to work as advertised until I powered up the device with the incorporated Xbee radio, which draws as much as 500mA in short pulses. Then, noticed a lot of the strange intermittent malfunctions typical of power problems, all of which disappeared when I powered the 3.3v rail with my lab power supply. I spent many hours trying to track down the source of the problems, without success. Finally, I gave up and redesigned the pcb with a buck regulator and a 9v battery. This works fine, but I'm back with the 9v, which I don't like because of high cost and low power density.

I am sure some of you are using boost regulators and 2 alkaline batteries for portable devices, and I wonder whether you have encountered similar problems. Does anyone have any boost regulator recommendations. I am especially interested in using 2- Alkaline AA cells and a boost regulator to produce 3.3v because it is impossible to find an inexpensive handheld enclosure that has a battery compartment for 3-AA cells.

Learning and Information, General Electronics

Tags:

Last Edited: Mon. Jul 11, 2016 - 02:23 AM

Kartman

Level: 10k+ Postman

Joined: Thu. Dec 30, 2004

Posts: 22408 View posts

Location: Melbourne,Australia

Posted by Kartman: Sun. Jul 10, 2016 - 11:06 PM

Jim, I'd suggest you try to find the root cause of the problem, otherwise you'll be none the wiser. It could be something as simple as the spec for a capacitor.
Switched mode supplies are complex things and there's a number of compromises you make in the design. Maybe you chose the wrong compromise? You could try TIctech support and see if they have some recommendations for your application?

ka7ehk

Level: 10k+ Postman

Joined: Fri. Nov 22, 2002

Posts: 20259 View posts

Location: Tangent, OR, USA

Posted by ka7ehk: Mon. Jul 11, 2016 - 12:05 AM

Agree with Kartman. One possible source of the problem is insufficient output capacitor size or excessive ESR. Another possible issue is resistance (real and core loss) in the inductor. Another is insufficient peak current capability in the switches (FETs). Another possible problem is core saturation. And there may be others that I am not thinking of. Most of these have a simple solution (better component) though it is not always easy to tell what "better" is until you determine what the source of the problem really is.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Last Edited: Mon. Jul 11, 2016 - 12:21 AM

Chuck99

Level: Posting Freak

Joined: Sat. Oct 31, 2009

Posts: 1371 View posts

Location: Vancouver, WA, USA

Posted by Chuck99: Mon. Jul 11, 2016 - 12:07 AM

jimlake wrote:

I decided to try 2-AA alkaline batteries and a 1.5A TPS61252 boost regulator producing a 3.3v output.

It appeared to work as advertised until I powered up the device with the incorporated Xbee radio, which draws as much as 500mA in short pulses.

First, you might get more responses if you post this to General Electronics rather than Compilers and General Programming.

You can ask a moderator to move it.

Second, you wrote that the circuit works fine if powered by a lab power supply on the 3.3V rail.

Try using the lab supply at 3.0V as a substitute for the two AA batteries and see how that works.

It could be that the batteries can't handle the current surge.

Also, what size bypass capacitor do you have across the output of the two batteries (input to the switch-mode regulator?

Spoken Link Power Supply.pdf

Level: Hangaround

Joined: Fri. Aug 5, 2005

Posts: 373 View posts

Location: Tacoma, WA

Posted by jimlake: Mon. Jul 11, 2016 - 01:28 AM

Sorry about the wrong forum. Fumble fingers, I suppose. Moderator: might you change it to General Electronics for me? * Done. Ross *

I attached the schematic of the power supply.

As I said above, I used TI's design guide to get the components. When I built the board, I started with the power supply only and ran it for hours with a .75A dummy load using the lab supply at 3.0v. It seemed fine. The problem crops up with the batteries. Note also that there is an auto-off feature in line before the power supply that has a PFET (Rds ON =0.21ohms). It may be part of the problem. However, I have the same thing with the buck regulator that I switched to, and it works with no issues.

Chuck99: Have a look at the schematic. The caps are all ceramic caps, but the 22uF output cap may have had a bit different ESR than the one proposed by TI (I had some). Other than that, all of the power supply components are the part numbers recommended by TI (see the WEBENCH design). I did use the lab supply to substitute for the batteries. I lowered the voltage to simulate battery drain and discovered that the power supply failed to supply enough current as soon as the voltage got below 2.95v. That was the main reason I gave up on it.

That's why I was querying you guys. I wondered what other folks experience was with these.

Also, the power supply is 0.5A, not 1.5A.

Attachment(s):

webench_design_836153_37_214137501.pdf

Last Edited: Mon. Jul 11, 2016 - 02:24 AM

ka7ehk

Level: 10k+ Postman

Joined: Fri. Nov 22, 2002

Posts: 20259 View posts

Location: Tangent, OR, USA

Posted by ka7ehk: Mon. Jul 11, 2016 - 01:43 AM

Batteries have a much higher internal resistance than most bench supplies. As a result, their output voltage tends to "sag" for a very short time under high transient loads. The input voltage may be dropping below the cutout voltage of the supply. The input current waveform is very different for step-up vs step-down so batteries behave quite differently in those two environments.

Jim

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

valusoft

Level: Moderator

Joined: Sat. Jul 2, 2005

Posts: 14737 View posts

Location: Melbourne, Australia

Posted by valusoft: Mon. Jul 11, 2016 - 02:28 AM

If it works on your power supply (presumably with its own large cap and low effective source resistance), you could try adding some series R to simulate the batteries' ESR and see if it barfs it. You may be asking too much of the AA batteries.

Ross McKenzie ValuSoft Melbourne Australia

http://www.conrad-electronic.co.uk/ce/en/content/ti_rechargable_batteries/Nickel-Zink-Akkus-die-neue-Alternative-zu-den-herkoemmlichen-Batterien

Level: 10k+ Postman

Joined: Tue. Jan 9, 2007

Posts: 10843 View posts

Location: Arlington, Texas, u.S.A.

Posted by gchapman: Mon. Jul 11, 2016 - 04:20 AM

jimlake wrote:

On the latest device, I decided to try 2-AA alkaline batteries and a 1.5A TPS61252 boost regulator producing a 3.3v output. Is is small, inexpensive (though a lot more than a LM317) and very efficient.

If willing to use secondary cells, there's nickel zinc; 2-AA nickel zinc cells for a battery that's 3.7V 100% SoC (3.8V EoC) and 2.6V at nearly 0% SoC.

Created by PowerGenix then the smaller cells mass production was moved to China.

If it's primary, lithium alkaline cells are 1.5V nominal with a fairly flat discharge curve.

jimlake wrote:

I spent many hours trying to track down the source of the problems, without success. Finally, I gave up and redesigned the pcb with a buck regulator and a 9v battery.

You might be closer than you think.

Try hacking the PCB that has the boost regulator to add an inductor or ferrite bead between the VR output and the radio's power; this will reduce di/dt in the VR's output loop and reduce dv/dt to the VR's error amplifier.

Will need to add capacitors on both sides of the inductor; "large" to the radio to reduce its dv/dt and a small one on the VR side to reduce di/dt (EMI).

Trial and error will work as a method since there's an excellent and repeatable stimulus (aren't radios grand)

Can use an instrument (network analyzer) to evaluate a regulator's gain and phase margins.

jimlake wrote:

... because it is impossible to find an inexpensive handheld enclosure that has a battery compartment for 3-AA cells.

An old TI calculator (rechargeable battery) had a battery compartment with a lid that held the shrink wrapped NiCd cells with short leads to a connector.

Easily replaced by an operator though an extra part for the logistics chain.

http://powergenix.com/

http://www.conrad.com/ce/en/product/252000/Conrad-energy-Rechargeable-AA-Battery-x4-pcs-NiZn-16V (for a datasheet)

http://store.batteriesamerica.com/ni-znbatteries.aspx (US; multiple distributors for other nations; try Hobby King, BetterPower Shenzhen, PK Cell, Aliexpress)

http://www.energizer.com/batteries/energizer-advanced-lithium-batteries

http://data.energizer.com/

http://data.energizer.com/PDFs/l91.pdf (Energizer L91 Ultimate Lithium datasheet; 500mA pulse is on page 2)

TI E2E Community

Blogs

Power House

Testing power supply: Measuring stability

by Bob Hanrahan

Apr 24, 2013

http://e2e.ti.com/blogs_/b/powerhouse/archive/2013/04/23/testing-power-supply-measuring-stability

WaveForms 2015 [Reference.Digilentinc]

Network Analyzer

https://reference.digilentinc.com/waveforms3/refmanual#network_analyzer

"Dare to be naïve." - Buckminster Fuller

Level: Hangaround

Joined: Fri. Aug 5, 2005

Posts: 373 View posts

Location: Tacoma, WA

Posted by jimlake: Tue. Jul 12, 2016 - 01:39 AM

I did notice another factor that may be the cause of the failure. The data sheet does say, "The TPS61252 device provides a power supply for products powered by either a three-cell alkaline, NiCd or NiMH battery, or an one-cell Li-Ion or Li-polymer battery." However, the Recommended Operating Conditions say Vin is 2.3v min to 6v max. When I chose to use two alkaline cells, I believed that as long as I was within the voltage spec, the device couldn't tell whether there were three or two cells. Now, I am not so sure.

Any opinions on this? Is there a difference between three alkaline cells at 3v and two alkaline cells at 3v other than the amount of energy available?

ka7ehk

Level: 10k+ Postman

Joined: Fri. Nov 22, 2002

Posts: 20259 View posts

Location: Tangent, OR, USA

#10

Posted by ka7ehk: Tue. Jul 12, 2016 - 02:18 AM

Yes, there is a difference. 3 alkalines discharged to 1V per cell has a much higher resistance than two at 1.5V per cell. I would expect the two good ones to work better than three discharged ones.

JIm

Jim Wagner Oregon Research Electronics, Consulting Div. Tangent, OR, USA http://www.orelectronics.net

Level: 10k+ Postman

Joined: Tue. Jan 9, 2007

Posts: 10843 View posts

Location: Arlington, Texas, u.S.A.

#11

Posted by gchapman: Tue. Jul 12, 2016 - 02:48 AM (Reply to #9)

The 3 cell spec is valid because a 100% discharged NiCd or NiMH cell is 1.0V

If willing try two AA NiZn cells into the boost VR though will be below 3V discharged but greater than the 2.3V min when discharged (am not certain this will work; might need a "large" capacitor on the battery)

Did you evaluate buck-boost VR?

http://www.linear.com/product/LTC3106 (buck-boost VR, 650mA peak, relatively expensive demo board)

https://www.monolithicpower.com/Products/Product-Detail?ProductID=683 (MP2155, buck-boost VR, 2.2A peak, new, demo board exists but couldn't locate it)

"Dare to be naïve." - Buckminster Fuller

http://www.mouser.com/new/atmel/atmel-ata8520-sigfox-transceivers/

Level: 10k+ Postman

Joined: Tue. Jan 9, 2007

Posts: 10843 View posts

Location: Arlington, Texas, u.S.A.

#12

Posted by gchapman: Tue. Jul 12, 2016 - 03:01 AM

jimlake wrote:

... the incorporated Xbee radio, which draws as much as 500mA in short pulses.

New at Mouser 4/25/2016 :

Mouser

Digi XBee SX Modules

http://www.mouser.com/new/Digi-International/digi-xbee-pro-sx-modules/

...

Adjustable output power up to 1-Watt achieves 65-mile range with high gain antenna

...

Frequency Range: ISM 902 to 928MHz

...

RF Data Rate: low: 10kb/s; middle: 110kb/s; high: 250kb/s

...

1W is similar to a 2W mobile phone.

Impressive.

P.S.

If willing to reduce the data rate by an order of magnitude there's SIGFOX (Atmel) and LoRa (Microchip).

http://www.mouser.com/new/microchip/microchip-dv164140-kit/

Edit : data rate, P.S.

"Dare to be naïve." - Buckminster Fuller

Last Edited: Tue. Jul 12, 2016 - 03:12 AM

chartman

Level: Resident

Joined: Thu. May 1, 2003

Posts: 688 View posts

Location: Fleet Hants UK

#13

Posted by chartman: Tue. Jul 12, 2016 - 11:24 AM

From the data sheet, it appears that at 3V input, the device cannot supply your 500mA demand....Take a closer look and that might be your problem. Try it with 3 cells....

Level: Hangaround

Joined: Fri. Aug 5, 2005

Posts: 373 View posts

Location: Tacoma, WA

#14

Posted by jimlake: Tue. Jul 12, 2016 - 05:50 PM

gchapman: Thanks for the links to the Monolithic Power products. I was not familiar with them. I need to look closer at the buck/boost regulators.

I am using the Xbee 900HP radio module with a wire antenna. I am very impressed with this radio. I need a range of less than one mile, and this radio is more than capable of that.The only other thing that I would like is a lower price. They are about 35$ each on Mouser. I have also found that Digi has very good technical support. Since this radio goes in an RC plane, I like that fact that the 900MHz band does not load up the 2.4GHz band where most of the RC controllers are working. The Xbee is tough too. I have had the radio survive devastating crashes with no ill effects. I mount the radio on a velcro pad. I use the 10kbs data rate. The only thing that the radio is handling is airspeed data and 10K is plenty fast for that.

I did evaluate the Microchip radios, which I initially liked, but I found Microchip's tech support was terrible for anyone not buying millions of them. They basically don't care if you buy it or not. I did not evaluate the Atmel radio.