DIY 1A BUCK - BOOST DRIVER

125 posts / 0 new
Last post

Pages

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon
DIY 1A BUCK - BOOST DRIVER

There have been several conversations about available buck boost drivers. So here ya go!

This is some artwork for a 17mm, 2.7-5.5V, 1A Buck Boost driver using the Linear LTC3454. The design is taken right from the data sheet. A PIC10F322 was added to run a bare bones kind of UI. Battery board copper on bottom layer/side. Man it was a tight fit.

!1A Buck Boost Driver photo LTC3454.png!

Parts are on there way, but it will be a few weeks at least before the boards would come in. I see this as an open hardware, open source, or even a BLF project. These switch mode drivers are still a bit of a mystery to me, so fingers crossed. Comments welcome!

Kent

Edited by: nickelflipper on 02/01/2013 - 21:49
scaru
scaru's picture
Offline
Last seen: 5 years 1 month ago
Joined: 03/22/2012 - 13:36
Posts: 6946
Location: Virginia

Any chance of posting links for the parts and a file so we all can order some? Also, what is the advantage of this over a 7135 driver? As I understand it they are both very similar in operating range and in efficiency (both are linear). 

nitro
nitro's picture
Offline
Last seen: 3 weeks 6 days ago
Joined: 05/22/2011 - 13:32
Posts: 2706
Location: ohio

Looks awesome

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

scaru wrote:

Any chance of posting links for the parts and a file so we all can order some? Also, what is the advantage of this over a 7135 driver? As I understand it they are both very similar in operating range and in efficiency (both are linear). 


I will work up a BOM with part numbers, and eventually the gerber files. Too bad OSH Park doesn’t have their share a project feature going.

The LTC3454 is a switch mode driver (not linear), and would be quite a bit more efficient?, than the AMC7135. The data sheet claims 85-90% efficient. Of course they cherry pick their data point voltage at 3.6V. I think the main advantage would be when fresh batteries are in play. The LTC3454 would buck down the extra voltage, while the AMC7135 would be burning it off as heat.

Click on the embedded link in the first post for the data sheet..

Kent

scaru
scaru's picture
Offline
Last seen: 5 years 1 month ago
Joined: 03/22/2012 - 13:36
Posts: 6946
Location: Virginia

Oh, my bad. I just glanced at what you labeled the link as: "Linear LTC3454" Wink Once these start to become available count me in for one, I'm assuming this is one sided?

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

nitro wrote:
Looks awesome

Thanks.

Now that I have stepped back from it, I see several opportunities to clean things up and get rid of some vias and bottom copper. That would be by moving the PIC and running the LTC3454 ground right thru the middle of the package.

Kent

moderator007
moderator007's picture
Online
Last seen: 1 min 43 sec ago
Joined: 12/23/2012 - 04:47
Posts: 3777
Location: North Carolina

Should allow the driver to run almost constant current so there’s no loss in brightness until the battery is depleted.
A amc7135 decreases in brightness once the battery voltage (taking into account battery sag) becomes lower than the vf of the led and other losses, at its set current becoming direct drive until the battery is depleted. If a quality battery isn’t used with high current draws the amc7135 can drop out of regulation rather quickly depending on current and the sag of the battery.

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593

There are quite a few people here who have the ability to program the Atmel ATTiny chips and there are is quite a bit of source code for nifty drivers using it. Using that would be sweeter than a PIC design.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

texaspyro wrote:
There are quite a few people here who have the ability to program the Atmel ATTiny chips and there are is quite a bit of source code for nifty drivers using it. Using that would be sweeter than a PIC design.

I anticipated this, and by some fluke the ATtiny six pin devices have the supply, gnd, and even the PWM pins in the same locations as the PIC. They both are SOT23-6 packages and should be a straight swap, although I haven’t tried personally. The spare pin is brought out to a thru hole pad. Lucky for the ATtiny folks who can use the MCLR/RESET pin as an analog input.

Kent

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593
nickelflipper wrote:
texaspyro wrote:
There are quite a few people here who have the ability to program the Atmel ATTiny chips and there are is quite a bit of source code for nifty drivers using it. Using that would be sweeter than a PIC design.
I anticipated this, and by some fluke the ATtiny six pin devices have the supply, gnd, and even the PWM pins in the same locations as the PIC. They both are SOT23-6 packages and should be a straight swap, although I haven’t tried personally. The spare pin is brought out to a thru hole pad. Lucky for the ATtiny folks who can use the MCLR/RESET pin as an analog input.

It would be nice to use the SO8 package… people already have the SOIC clips for glomming on to the chip.

Slim Pickens
Slim Pickens's picture
Offline
Last seen: 2 days 9 hours ago
Joined: 04/28/2012 - 19:50
Posts: 1231
Location: Lehigh Valley

I hope you ordered enough to sell a few.

nickelflipper wrote:
The LTC3454 is a switch mode driver (not linear), and would be quite a bit more efficient?, than the AMC7135. I think the main advantage would be when fresh batteries are in play.

+1. Linear regulators seem to be a necessary evil these days. Many buck drivers need 2×16340 in order to regulate current to a white LED, and it’d be great to just use 1×18650 with a buck/boost driver.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

scaru wrote:

Oh, my bad. I just glanced at what you labeled the link as: “Linear LTC3454Wink Once these start to become available count me in for one, I’m assuming this is one sided?


All IC and passive components on the topside copper, but it is a two sided board. The bottom copper will accommodate a battery spring in the center, like a Nanjg 105C, and there’s the outside annular ground ring too.

Kent

moderator007
moderator007's picture
Online
Last seen: 1 min 43 sec ago
Joined: 12/23/2012 - 04:47
Posts: 3777
Location: North Carolina

Nickelflipper do you have a round about number on what the efficiency would be once the driver switched to boost.
Not going to hold you to it, just curious. Wink
Edit: I guess I should have read the datasheet first. Says 90% from 2.7V to 4.2v

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593
moderator007 wrote:
Edit: I guess I should have read the datasheet first. Says 90% from 2.7V to 4.2v

Read/look at graph a little closer and… at 3.1V in and 1A out it is around 78% efficient. 90% is only reached at 150 mA. At 1A the efficiency never gets above 85%.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

texaspyro wrote:
It would be nice to use the SO8 package… people already have the SOIC clips for glomming on to the chip.

Absolutely no room for an SOIC8 package. I can appreciate the programming situation isn’t so hot, no room for pads either, which is what I would prefer. The only thing that comes to mind is making up a SOT23 to DIP adapter gadget. I have SOT23 programming adapter for the PICKit programmer, which is kind of expensive to justify for the casual user.

My goal with a single board construction like this was to keep it simple, and top side only components. Maybe I’m not thinking hard enough?, I’m still listening.

Kent

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593

A SOIC should be do-able. Look at what’s on those Chinese drivers. What size passives are you using?

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

@texaspyro

Passives are 0603, with a couple 10uf caps 0805, inductor is 5.2mm which is a small form factor for 1.85A saturation current. The problem with some of these fancy LED drivers are is their high component count. My idea is to dap some solder paste on the pads, put it on the hotplate, done. So, a two sided component construction is not appealing to me.

Kent

NightCrawl
Offline
Last seen: 5 years 10 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Great idea, I guess you have some experience with electronics? Big Smile

Sadly 1A is not enough for me.. Wink

moderator007
moderator007's picture
Online
Last seen: 1 min 43 sec ago
Joined: 12/23/2012 - 04:47
Posts: 3777
Location: North Carolina
texaspyro wrote:
moderator007 wrote:
Edit: I guess I should have read the datasheet first. Says 90% from 2.7V to 4.2v

Read/look at graph a little closer and… at 3.1V in and 1A out it is around 78% efficient. 90% is only reached at 150 mA. At 1A the efficiency never gets above 85%.


Thanks Texaspyro, I guess I should have read on down the datasheet. At the top it says greater than 90% can be achieved. Well thats only at 150ma like you said. That was kind of the reason I was asking, boost circuits are usually not all that efficient.
nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

NightCrawl wrote:

Great idea, I guess you have some experience with electronics? Big Smile

Sadly 1A is not enough for me.. Wink

No formal training in electronics, but have been doing it for my own projects for some time now. I have some parts coming for a 3A buck boost board, but getting the 1A board going is first up.

@texaspyro

I will look into putting the 2 × 3 AVR programming pads on the bottom copper, I think this might be doable?

Kent

NightCrawl
Offline
Last seen: 5 years 10 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Sounds great. I sent you a PM, might be interesting.. keep us updated. Smile

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

O.K. then,……. I will be putting aside my PIC bias in favor of the AVR for the collective good? :).

It looks like the SOIC8 chip footprint is going to fit, with some re-juggling of the passives.

@NightCrawl – back at you. If you want to further discuss a higher amp version you can continue it in your 1 cell buck bosst thread (I,m subscribed).

Kent

scaru
scaru's picture
Offline
Last seen: 5 years 1 month ago
Joined: 03/22/2012 - 13:36
Posts: 6946
Location: Virginia

Could 3 of those buck/boost chips be used to get up to 3 amps? If so, would they fit if the board was double sided (components on both sides)?

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

scaru wrote:

Could 3 of those buck/boost chips be used to get up to 3 amps? If so, would they fit if the board was double sided (components on both sides)?

I don’t think so, as how would you synchronize the outputs? If you diode OR’d them together then there would be the reduced efficiency/output? I think you would go to a different single chip solution for the sake of board space (at least for the 17mm form factor).

A 3A solution should be a separate thread, and like I have said, I’m not going to put much of my own time until the 1A is sorted.

Questions for AVR junkies: Any reason why the ADC1 pin couldn’t read the VCC pin directly? You don’t have to have an inductor/cap combo for reliable ADC readings do you?

Kent

NightCrawl
Offline
Last seen: 5 years 10 months ago
Joined: 01/22/2012 - 08:20
Posts: 3071
Location: Karlsruhe, Germany

Dont worry, we wont derail this thread. Smile

But a quick look at the linear.com page shows several boost-buck chips, some handle more than 1A. But sadly I dont understand quite enough of that stuff.

JackCY
JackCY's picture
Offline
Last seen: 6 years 8 months ago
Joined: 12/28/2012 - 16:30
Posts: 588
Location: Czech

PIC? Tab closed, why would you use PIC over Atmel? Not to start a flame, just want to see what made you pick PIC over Atmel. Size, price, features, speed, memory, consumption, programmability, …

I can’t say I have a need for a 1A buck-boost, 7135s do the buck well. I would be great thought to have a driver that can supply constant power to the LED by drawing more amps out of the battery when it has lower voltage to supply.

I guess it’s still useful for people with NiMhs to have buck-boost 1A. At this power I think I’ve seen some though.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

JackCY wrote:
PIC? Tab closed, why would you use PIC over Atmel? Not to start a flame, just want to see what made you pick PIC over Atmel. Size, price, features, speed, memory, consumption, programmability, …

I can’t say I have a need for a 1A buck-boost, 7135s do the buck well. I would be great thought to have a driver that can supply constant power to the LED by drawing more amps out of the battery when it has lower voltage to supply.

I guess it’s still useful for people with NiMhs to have buck-boost 1A. At this power I think I’ve seen some though.

Hah, hah, I’ve heard countless discussions about the PIC vs. AVR. Sometimes it just boils down to what you started with (PIC for me), the time invested, and familiarity. I have nothing against AVR, just not as familiar with them.

Kent

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593
nickelflipper wrote:
Questions for AVR junkies: Any reason why the ADC1 pin couldn’t read the VCC pin directly? You don’t have to have an inductor/cap combo for reliable ADC readings do you?

To read the battery voltage you will need to configure the chip to use the internal 1.1-1.2V bandgap reference. Usually the ADC can only measure voltages up to the reference voltage so you would need a voltage divider on the battery. Note that high impedance voltage dividers can affect the reading due to the time it takes to charge the sample and hold cap. You may need to adjust for the reading that you get.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

texaspyro wrote:
To read the battery voltage you will need to configure the chip to use the internal 1.1-1.2V bandgap reference. Usually the ADC can only measure voltages up to the reference voltage so you would need a voltage divider on the battery. Note that high impedance voltage dividers can affect the reading due to the time it takes to charge the sample and hold cap. You may need to adjust for the reading that you get.
Close tab…minus 1 for AVR. The PIC also has a 1.024V internal reference, but with the advantage of a 2x and 4x multiplier, no voltage divider needed. Boo Hoo.

I may go back to the SOT23-6 with the programming pads broke out for the PIC or AVR?

Kent

texaspyro
Offline
Last seen: 2 years 6 months ago
Joined: 04/29/2011 - 12:43
Posts: 4593
nickelflipper wrote:
The PIC also has a 1.024V internal reference, but with the advantage of a 2x and 4x multiplier, no voltage divider needed.

Even with a 4X multiplier you can’t measure a fully charged lithium cell… but may not matter if all you are doing is low voltage cutoff.

Also, forgot to mention that the TINY85 has an on-chip temperature sensor. Not very accurate unless you apply a calibration factor.

nickelflipper
nickelflipper's picture
Offline
Last seen: 3 weeks 4 days ago
Joined: 10/20/2012 - 09:31
Posts: 498
Location: Oregon

texaspyro wrote:
Even with a 4X multiplier you can’t measure a fully charged lithium cell… but may not matter if all you are doing is low voltage cutoff.

Also, forgot to mention that the TINY85 has an on-chip temperature sensor. Not very accurate unless you apply a calibration factor.

Yea, not interested in voltages over 4V, lets hope the charger is doing its job. I’ve tried a newer PIC with the internal temp sensor, and danged if I could get any sensible numbers.

Kent

Pages