gfxgfx
 
Please login or register.

Login with username, password and session length
logo
 
gfx gfx
gfx
19852 Posts in 1275 Topics by 5182 Members - Latest Member: charbuild March 28, 2024, 04:41:21 PM
*
gfx* Home | Help | Search | Login | Register | gfx
gfx
Breaktru Forum  |  eCigarette Forum  |  Modding  |  Topic: Working on this
gfx
gfxgfx
 

Author Topic: Working on this  (Read 19707 times)

0 Members and 2 Guests are viewing this topic.

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Working on this
« on: June 11, 2014, 07:53:02 AM »
I don't post pics of mods I build here much since they all look pretty much the same.  What changes is the controller boards I put in them, always doing upgrades and trying new designs.  So to avoid looking like all I do is talk a lot on this forum, here's what I'm working on.

This is a cool little job I'm about about to submit to the PCB fab.  Instead of building a whole board, I'm going to do it in two parts this time.  Don't want to find it doesn't work after spending a whole bunch of time designing a complete board.  So I'm doing the power converter first.  Once I make sure it works to my satisfaction, I'll do the whole controller board.

This converter is pretty trick in that it uses a regular boost controller to create a buck-boost converter by using a virtual ground at atomizer negative.  The beauty is in its simplicity.  The only change in topology it requires is a relocation of the atomizer ground tap.

If all goes well, I'll get a 40W/15A output off this converter with a range of 2.5 to 8V running from a single cell.  Simulator says it works so I'll see.

Here's the top of the board.  The layout looks inefficient, but this is for testing as part of a complete board.  The complete controller board will have other stuff around those blank spaces (the forum automatically resizes the images so just click on them to make them bigger).



Here's the bottom of the board;



This is a 4 layer board so here it is with all the layers showing.  I need those two ground planes in the middle to handle the big currents.  The one downside with this converter design is it increases switching currents quite a bit, need big paths and parts with absolute minimal losses.

« Last Edit: June 11, 2014, 08:22:57 AM by CraigHB »

Online Breaktru

  • Administrator
  • PV Master
  • *****
  • Joined: Nov 2008
  • Location: NY
  • Posts: 3344
  • Karma: +792/-4
  • Gender: Male
  • PLEASE participate in this forum
Re: Working on this
« Reply #1 on: June 11, 2014, 08:49:53 AM »
Outstanding work Craig. You never cease to amaze me.

Offline Visus

  • PV Master
  • *******
  • Joined: Oct 2013
  • Location: Nexxus
  • Posts: 929
  • Karma: +62/-0
  • Gender: Male
Re: Working on this
« Reply #2 on: June 11, 2014, 02:29:47 PM »
Beautiful

The rub is from your chats I actually are looking and understanding.  Seen your powerblok in action, imo it hits better than any ecig mech or regulated,  there is no doubt the soupy richness of the vapor is unmistakable in attention to detail and no compromise parts selection,  how it performs.   Ya build Bugatti's of the ecig world,  I know this one is as awesome..  If only ya could get a look at some of those secretive chinese chips to have choice at using..  Heckuva useful hobby,  combined w/ chats help contain and annihilate the analog urge for many of us..

 :thumbsup:

Offline Alexander Mundy

  • Jr. Member
  • **
  • Joined: Jul 2013
  • Posts: 21
  • Karma: +3/-0
Re: Working on this
« Reply #3 on: June 11, 2014, 05:21:28 PM »
Had a look at the boost data sheet and your board.
I can follow most including the fact you are feeding Vin from the charge pump and firing with a pmos on the low side of the inductor.
Maybe I cant see the forest for the trees, but where are you taking atty negative from?
Common connection of high side driver, low side driver, and inductor?
I need my mind expanded a little to understand how that wold work if that is the case.

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #4 on: June 11, 2014, 06:07:44 PM »
Ya build Bugatti's of the ecig world,  I know this one is as awesome.

Thanks for the compliment Visus, I try to make my boards kind of stand apart from what else is out there.

Quote
If only ya could get a look at some of those secretive Chinese chips to have choice at using.

It would be nice to have access to some of those domestic Chinese chips.  I'd have to translate the Chinese data sheets.  Used to be the US and Japanese companies had all the best IC designs, even if produced elsewhere, but the Chinese are coming up with some cool stuff.

Though, I trust the quality a lot more on the stuff designed by the big US and Japanese names like the parts that come from TI.  This converter is actually an all TI design.  All of the semiconductor parts on that board are TI parts.  The Chinese like to skimp and cheat wherever possible and using their domestic parts becomes the determining factor in the reliability of your design.

Just about all electronic components are produced overseas and the majority of them in China.  Though the company that handles the design and runs the production facility makes all the difference in quality.  Doesn't really matter where a part is made, just who makes it.

The reliability of a design is highly dependent on the parts used in it.  Most of the time when something electronic works it works, but stuff does fail, especially when operating at the limits of its capability.  I build my controller boards with the highest quality parts I can get and the highest reliability in mind.  So far they haven't let me down.

I can follow most including the fact you are feeding Vin from the charge pump and firing with a pmos on the low side of the inductor.

I think I understand what you're saying, but just to clarify, the PMOS is only the main power switch.  For testing, I need a manual power switch in there.  The compete board actually fires that MOSFET using a complimentary small signal MOSFET pair driven by the MCU.  The converter itself uses two NMOS switches, hence the charge pump.

I made a bit of misnomer in the notes on that board design.  The charge pump is required to drive the high side MOSFET only in a round about way.

When using high side NMOS switches, you need a voltage higher than input to turn the switch on and off.  The converter controller does that with a bootstrap diode and capacitor, but it also requires a high enough voltage on Vin which supplies that circuit.  Since Vin is minimally 4.5V on that part, I use a charge pump to boost supply voltage from a single cell within the acceptable range for the controller supply.  The only reason the controller has a minimum Vin that high is because of the high side NMOS (in lieu of a PMOS).

There are performance advantages in using an NMOS over a PMOS on the high side switch.  NMOS switches have lower "on" resistance and lower gate charge than PMOS switches.  That higher Vin comes as a cost for the higher performance switch.  This is the case for any controller that uses a high side NMOS switch.

Quote
Maybe I cant see the forest for the trees, but where are you taking atty negative from?  Common connection of high side driver, low side driver, and inductor?

I'm picking up atomizer negative at battery positive, after the main switch.  There's a pad that isn't shown for the wire.  It's just behind the inductor were you see the divot in the PCB.  Atomizer positive picks up at the output capactors like a normal boost converter.

This simple change in topology changes the boost converter into a buck-boost converter.  The controller is perfectly happy supplying the load with only the voltage generated by the inductor instead of "battery plus inductor".  The down side is the inductor has to provide all of the voltage output which pretty much doubles the switching current, not quite, but just about.  It makes for some tricky part selection and PCB design in consideration of power loss.

Offline Alexander Mundy

  • Jr. Member
  • **
  • Joined: Jul 2013
  • Posts: 21
  • Karma: +3/-0
Re: Working on this
« Reply #5 on: June 11, 2014, 07:10:30 PM »
Quote
I think I understand what you're saying, but just to clarify......

Yes, I followed the circuit as you describe.

Quote
I'm picking up atomizer negative at battery positive, after the main switch.

Ah, you have enlightened me. Very clever.  :thumbsup:

At first thought it seems that the output voltage would increase as B+ voltage decreases unless the feedback loop was altered? I haven't went back and traced that as I am mobile now.

Offline Claviger

  • Sr. Member
  • ****
  • Joined: May 2014
  • Location: Hawaii
  • Posts: 94
  • Karma: +4/-1
  • Gender: Male
Re: Working on this
« Reply #6 on: June 11, 2014, 07:18:33 PM »
happy to see you are continuing development of your boost/buck controller.  who do yuo use or your pcb prototypes if you dont mind?

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #7 on: June 11, 2014, 07:34:57 PM »
At first thought it seems that the output voltage would increase as B+ voltage decreases unless the feedback loop was altered?

Yes, the difference is only where the atomizer picks up ground, strictly.  The feedback circuit remains unchanged from the boost configuration.  That presents a problem in that the controller will regulate to "battery plus inductor" like it normally does in a boost configuration, but voltage on the atomizer is only inductor voltage.  Common sense would prompt you to put feedback on the same ground as the atomizer, but that doesn't work since the controller is not using the same ground as the atomizer.

So, the solution is that output voltage must be regulated by the MCU.  I'll use a differential amplifier with fractional gain on atomizer positive and negative to provide an analog feedback voltage level to the MCU.  The MCU processes the level and forwards data to the digital pot to maintain the correct atomizer voltage.


« Last Edit: June 11, 2014, 07:53:05 PM by CraigHB »

Offline Visus

  • PV Master
  • *******
  • Joined: Oct 2013
  • Location: Nexxus
  • Posts: 929
  • Karma: +62/-0
  • Gender: Male
Re: Working on this
« Reply #8 on: June 11, 2014, 07:38:13 PM »
Yes, I followed the circuit as you describe.

Ah, you have enlightened me. Very clever.  :thumbsup:

At first thought it seems that the output voltage would increase as B+ voltage decreases unless the feedback loop was altered? I haven't went back and traced that as I am mobile now.

Same as i thought an open loop using a matching resistor to flip the neg
meh'  too much info running around at once with no hands on experience in me brain..   :facepalm:

He did say in other thread, ground at the atomizer,  he truly meant ground at the atomizer

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #9 on: June 11, 2014, 07:48:42 PM »
who do yuo use or your pcb prototypes if you dont mind?

I use Osh Park (https://oshpark.com).  I used to a pay a lot more for prototype PCBs until I found these guys.  You have to order in sets of three, but it's so cheap, it's not a big deal and it's always better to have more than one board on-hand anyway.  When testing stuff, there's a high likelyhood of frying something so you might need another board anyway.

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #10 on: June 12, 2014, 09:22:20 PM »
I submitted the boards in the OP last night.  They only do the 4 layer panels twice a month so it sucks when you submit at the beginning of a cycle.  You have the two week lead time plus the two week order cycle.  I so lucked out this time.  I submitted my boards the day before the panels went off to the fab.  Could not have timed it better.  I should have them in about a week and a half.  They say two weeks, but it's always less.

You know what's seriously amazing, the order was a mere $14.  I just can't believe how cheap it is to get proto boards through Osh Park.  They make building your own circuits a cheap hobby.

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #11 on: June 16, 2014, 04:25:38 AM »
Hi guys, i'm new here!

wow, great works and development Craig! i've been hurting my head :wallbash: trying to build my own booster circuit, how i wished i would have seen your works earlier!! :help:

just to share, this is around 10W buck circuit i produced with a friend. usually i play with low current 5V circuitry, this is my first time trying to build high current circuit so i solely rely on my friend for the high current part. so he produced the schematic of the H-Bridge without both knowing what output the circuit could achieve Doh:

there are some mistakes with the connections of the 4 layer board (also 1st time for me to produce such PCB), with some drilling and jumpers, the PCB works except for the OLED part.



i learned A LOT on this new experience, so much new stuff!!

currently we are building a new revision based on the boost topology. the design & PCB is ready, however i'm waiting for my OLED breakout that i should receive in a few days so that i can confirm the connections. it should be straight forward especially if using 3.3V microC, however the previous design used 5V and i also chose the wrong level converter IC (CD4504B).

hoping to learn more from this forum especially when i have the time to read back all the posts, TQ!

Online Breaktru

  • Administrator
  • PV Master
  • *****
  • Joined: Nov 2008
  • Location: NY
  • Posts: 3344
  • Karma: +792/-4
  • Gender: Male
  • PLEASE participate in this forum
Re: Working on this
« Reply #12 on: June 16, 2014, 08:27:24 AM »
Hi guys, i'm new here!

wow, great works and development Craig! i've been hurting my head :wallbash: trying to build my own booster circuit, how i wished i would have seen your works earlier!! :help:

just to share, this is around 10W buck circuit i produced with a friend. usually i play with low current 5V circuitry, this is my first time trying to build high current circuit so i solely rely on my friend for the high current part. so he produced the schematic of the H-Bridge without both knowing what output the circuit could achieve Doh:

there are some mistakes with the connections of the 4 layer board (also 1st time for me to produce such PCB), with some drilling and jumpers, the PCB works except for the OLED part.

i learned A LOT on this new experience, so much new stuff!!

currently we are building a new revision based on the boost topology. the design & PCB is ready, however i'm waiting for my OLED breakout that i should receive in a few days so that i can confirm the connections. it should be straight forward especially if using 3.3V microC, however the previous design used 5V and i also chose the wrong level converter IC (CD4504B).

hoping to learn more from this forum especially when i have the time to read back all the posts, TQ!


Welcome to the forum miskol. Nice going on your board and design. Would luv to see it completed

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #13 on: June 16, 2014, 01:04:30 PM »
wow, great works and development Craig! i've been hurting my head trying to build my own booster circuit, how i wished i would have seen your works earlier!!

Power converters are one of the more esoteric areas of circuit design.  They can make you go "what the hell" a lot.  One thing is there's lots of documentation about designing them available through the various makers of the controller chips.  Just about anything involving them can be researched easily enough.  The math can make your head explode, but you can get around that with a good simulator, pretty critical tool for this stuff.

I actually got into a chat with one of the power engineers at TI about this converter I'm trying to do.  He said it's something he hasn't seen before, but it's reminiscent of obscure toplologies like floating buck and inverting buck.  If you Google those terms some TI papers will come right up.  Interesting stuff.  He says my design could be called a floating buck-boost.

It's going to be interesting to see what happens with this for my own edification if nothing else.  Boards coming next week.  There's still the 4 switch buck-boost topology if this floating buck-boost doesn't work out.  Just waiting on a controller to be released for that.

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #14 on: June 16, 2014, 01:34:26 PM »
TI engineers are helpful, especially when you are asking about your design that is using their product (not sure if other manufacturers are helpful as well). a friend of mine received advices and recommendations from TI engineers about the circuit that our research group have been using for years! he followed the advices given, simulated it and sees the results exactly like what the TI engineers discussed previously with him. just submitted the board for fab, probably next week will receive it.

anyway back to vaping, our next revision is really based on the simple boost topology, we have the inductor, mosfet, diode and capacitors. our simulation on the design works well with current up to 15A... so probably in real world hoping to at least achieve around 8 - 10A.

i have (or had) two backup plans, one was to explore the use of LTC3785 (i just read ur post on that, TQ for the works u did, at least i know NOT to explore it). so now my only backup is to repair our H-Bridge circuit and at least have a 10W vaping mod of our own. other than that, probably to scour for new controller that is not made by Linear.


Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #15 on: June 16, 2014, 02:19:12 PM »
My advice is stay away from the LTC3785.  In fact, I'll never use another Linear part again.  TI always and I mean always pads their stuff so no matter how bad you screw up on the engineering, your circuit won't blow up.

When Linear said 10V as an absolute maximum, they meant 8V.  That stupid converter controller (and I can't say that with enough prejudice) could not handle transients on the SW input at the maximal output voltage I had designed of 8V even though they advertise a 10V range for the part.  Two controllers blew out on me before I gave up on the design.  There were some other failures to meet specs too.  There was way more regulation error than the specs indicated.  The transition from buck to boost had more ripple than the specs indicated.

Not that I'm never without fault, but I know what I'm doing as much as any other engineer so my design was good enough for the converter to function to spec.  I'm so steaming mad at Linear, still.  If you're going to advertise a 10V part, make sure it can handle 10V so poor schmucks like me don't waste several months on a design only to scrap it due to a chip maker's failure to advertise realistic specifications.

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #16 on: June 16, 2014, 02:32:12 PM »
if it happened to me, i think it might take me years to finally figure out that it was the manufacturers mistake and NOT mine!

this is the bottom image of our previous design. i put it away after some time and wow, the inductor got rusty, bad component!


Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #17 on: June 16, 2014, 03:19:13 PM »
Yeah, there's always the chance it's bad engineering on my part over a design flaw on the chip maker's end.  I'm about 90% sure it was a poorly designed chip over a poorly designed circuit board on that last one.  But even so, any part should be "bad engineer" proof to a certain extent. 

Did you mention before that's a 10W boost board?  What controller is that if you don't mind me asking?  Should be able to get 40W output with a board like that.  Looks like you've got the right switches assuming they're under 6 mOhms. 

That looks like a 4.7uH inductor which is too high a value for a high output design.  If you run the frequency around 600kHz and drop that inductor 1uH or under, you should be able to get 40 Watts out of it no problem.  The tricky part is getting it unconditionally stable which is the tricky part for any boost converter. 

If that's a voltage mode boost controller, you'll be spending a lot of time stabilizing it for a 40W output.  It's possible though, I've done it.  Current mode is easier, but I haven't done a small 40W boost converter in current mode yet.  I will be soon.

Boosters are generally harder to stabilize then buck converters.  Mathematically, they have an extra pole in the transfer function which makes it harder to get the phase margin you need.  That's why you don't see high output boost modules like you see with the buck modules.  Buck is easier to design for high outputs.


« Last Edit: June 16, 2014, 03:53:02 PM by CraigHB »

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #18 on: June 16, 2014, 03:50:52 PM »
No, its based on Half-Bridge Configuration (followed the example in the datasheet), buck topology (i'm sorry if i used the wrong term), the output voltage is below the input voltage.

the controller used is LM5106. it could?!! now that i think it over, maybe it really could at least higher than 10W. since i'm really new in this whole thing, i remembered i tested the circuit with only Rload around 1.2ohm, able to get max voltage around 3.6V and got around 1.8A ... i should have tested with lower Rload and stretch its limits... however i dont really understand this "1.8A Peak Output Sink Current", does it mean using this Controller, the max current output is 1.8A?

Did u mean the Rds, if yes then its Rds is 1.85 mOhms.

Initially i used 1uH inductor with lower Amp rating, not fully understanding the circuit i thought the Inductor was limiting the current so i changed it with what i have on hand, which is 4.7uH with higher Amp rating with no significant changes with the output.

I'm not sure if it's voltage or current mode controller, sorry.

How to determine the lowest and highest Rload the circuit could handle, is it solely based on real test try & error (with what Breaktru did with the Beta DNA 20), or is there a calculation?

The problem with the H-Bridge configuration is that, the output is pulse ( i want it to be DC ) and i want my mod to use a single 18650 battery, so i need to use the boost topology. Is there any simple filtering to get DC output and is there a way to use this Controller for boost topology?

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #19 on: June 16, 2014, 04:21:20 PM »
That sounds interesting, let me look at that data sheet. 

That's a gate driver, different animal.  That's a multi-chip solution so its kind of the old school way of doing it.  Used to be you needed a separate gate driver, but modern single chip controller solutions have pretty good gate drivers built into them.  They're usually under 3 Ohms which gives you good enough rise and fall times for frequencies up to a mega-Hertz.  In some cases you still need a separate gate driver, high voltage stuff would be one, high gate charge would be another, but none of the e-cig stuff requires it.

You would have to be using a controller with that which will determine the control mode.  You could actually use that gate driver for either buck or boost.  It only depends on the controller and which side of the inductor the switches and gate driver live on.

In terms of the inductor, you could probably get things to work at high outputs with a 4.7uH inductor, but the problem is the heating you get with the higher DCR.  And then there are the losses which impact efficiency.  For a high output design, you really want your switches and inductor no more than 6 mOhms. 

Your switches are good at 1.85 mOhms, but your inductor DCR is going to be way too high for big outputs.  I actually run a 680nH inductor on my boost board with a frequency of 600KHz.  My buck-boost board will be running a 470nH inductor with a somewhat higher frequency.  It's looking like 750kHz right now, but not until I bench test it will I know for sure.  750kHz is the absolute lowest I can get away with when using an inductor that low of a value, but I need the lowest DCR I can get for that one.

There's only one way to boost voltage from a battery and that's with a boost converter so yeah, that's how you have to do it.  You pick your controller and design around that.  There's not a huge selection of external switch boost controllers like there are buck controllers so the field is a pretty limited there.  You can go internal switch and you may be able to find ones that can do maybe 10 or 20 Watts, but for the big outputs you have to use external switches.


« Last Edit: June 16, 2014, 04:32:38 PM by CraigHB »

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #20 on: June 20, 2014, 09:20:55 AM »
TQ for your comment, i've done some reading to understand the greek language lol

based on my previous design which i planned to continue using on my latest design, i just found out that my multi-chip solution is based on voltage-mode control.

in terms of the switching frequency, my calculation used 50k freq and thus the components selection and values are based on that freq. further reading, i found that frequency selection is based on some aspects such as:
- ensuring DC/DC converter not operating at the same freq as other sensitive components in the circuit
- faster freq means smaller inductor size
- 500kHz is generally regarded as the sweet spot

if i were to follow this sweet spot 500kHz freq it doesnt necessarily means i have to find a new Inductor right? it just gives me an option to use a smaller Inductor value if i want to. i would love not to go hunting for new components again if dont need to hehe. also my chosen current MCU could only achieve up to 60kHz.

the LM5106 have Pin EN, which makes it easy to disable or enable the gate driver (previous design used this Pin), as a result effectively turning on and off the output, so i could use this Pin connected directly to a switch or could use a MCU to control it. my question for this particular part, is it recommended to do so or is it better to use a mosfet at the output.
« Last Edit: June 20, 2014, 11:07:28 AM by miskol »

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #21 on: June 20, 2014, 03:57:25 PM »
When you simulate your converter, you'll find a fairly linear relationship between switching frequency, inductor value, and the resulting inductor ripple current.  There's no hard and fast rule about ripple current, but most designers try to keep it between 20 and 30% of average inductor current.  The reason for this is to minimize electrical noise.  However, with an e-cig we're not particularly concerned about electrical noise unless it impacts the more sensitive parts of the circuit.

If you use good ground plane isolation and electrical decoupling on your logic circuits, you can get away with much higher ripple currents than the recommended 20 to 30%.  This is good since it allows you to relax requirements for inductor values and frequencies.

Another consideration with inductor ripple is that it's usually discussed in terms of percentage of overall inductor current, but that percentage varies with load.  It's not a good way to look at inductor ripple for an e-cig converter since load can cover a wide range.  What I usually do is look at it in terms of raw amplitude.  For example, I try to keep it 6 Amps and under for these e-cig converters I do. 

A final point about higher inductor ripple is that you transistion from CCM to DCM sooner under lighter loads.  I'm not sure if you're already aware of these operational modes, but if not, CCM is continuous conduction mode and DCM is discontinuous conduction mode.  The controller has to gracefully handle these transitions.  This may or may not be a problem depending on your controller.  Most of the time it's not an issue, but it can be so you have to check it in testing.

I'm going to try to stretch inductor ripple to the limit with this latest one I'm doing, boards here Monday.  Need the really low inductor DCR for the much higher switching currents with this one. 

Yes, 500kHz seems to be a switching frequency sweet spot for the most part.  When you go higher, you get more switching losses which reduce efficiency.  When you go lower, you need higher inductor values resulting in higher DCR and higher losses.  That's not a hard and fast rule though.  There's no reason you can't run as low as 250kHz or as high as 750kHz if that's preferable for whatever reason.

Running up switching frequency leads to more problems than just switching losses.  It means you need faster MOSFETs with lower gate charge and that increases "on" resistance for the switches.  Higher frequency is a double whammy, higher switching losses driving the gates faster and more resistive losses in the switches themselves. 

Designing these converters can be quite a balancing act when pushing them to the limit.

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #22 on: June 25, 2014, 04:55:39 AM »
Well, I built it and did some bench testing on it, photos below.  It's stable and it can put out a boatload of power, ran it up to 60 Watts and no boom.  Though I'd cap it at 40W due to heat considerations.  Puts out up to 15 Amps and runs from a range of 2.5V to 8V as designed. 

I have one little issue (there's always something).  It's with something called dead time which has to do with how long the high side switch waits to turn on.  It's a problem since if there's too much, it kills efficiency and there's too much.  I have to ask my TI guy.  If I can work that out, I can use this converter design to build a single cell mod controller board with a really wide voltage range and good top end output current.






« Last Edit: June 25, 2014, 05:16:04 AM by CraigHB »

Offline Visus

  • PV Master
  • *******
  • Joined: Oct 2013
  • Location: Nexxus
  • Posts: 929
  • Karma: +62/-0
  • Gender: Male
Re: Working on this
« Reply #23 on: June 25, 2014, 05:41:30 AM »
Sweet comings along there,  way out my league.

I do know that if a tree falls in the forrest and no one is around to hear question is irrelevant..
so haha

Although
 I coulda done a mean job of handing you a beer, whilst..
Nice welding there craig
awesome

Online Breaktru

  • Administrator
  • PV Master
  • *****
  • Joined: Nov 2008
  • Location: NY
  • Posts: 3344
  • Karma: +792/-4
  • Gender: Male
  • PLEASE participate in this forum
Re: Working on this
« Reply #24 on: June 25, 2014, 07:00:11 AM »
Amazing results craig. Your designs are getting smaller and using less components yet they are getting more powerful.
I hope you resolve the "dead time".

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #25 on: June 25, 2014, 08:38:41 AM »
awesome single cell booster Craig! nice circuit design and layout, seems complicated  :thumbsup:

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #26 on: June 25, 2014, 01:39:20 PM »
That's probably about small as I can possibly go.  Keep in mind that's just a converter not a whole mod controller board.  That's the layout the converter would be in if it was part of a whole board.  There would be other stuff in those blank spaces and the board would be a lot bigger.

Have to make sure any particular converter works the way I want before I build a whole board with it.  This is not something that can be done on a breadboard, has to be done on a PCB.  I made that mistake before, did a whole board around an untested converter and got screwed when the converter didn't work to my satisfaction, won't do that again.

Need to capture some waveforms off the scope and send them to the TI guy, see what he says.  Though the topology works without issue and that's good to know.  My dead time issue is is not due to the unusual topology, it does the same thing when configured as a normal boost converter.  No idea what's causing it at this point.

In any case, it's pretty cool.  Just relocate the load return to batt + instead of batt - and you turn a booster into a buck-boost.  The one down side is that it has to be a high voltage booster since it has to handle battery voltage plus load voltage instead of just load voltage.  Boost controllers are not exactly a dime a dozen, let alone ones capable of 12V output.

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #27 on: June 25, 2014, 03:10:59 PM »
i've been wondering, what makes any topology have its max output current?  is it because of the Inductor used?

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #28 on: June 25, 2014, 05:27:47 PM »
It's mainly about losses.  You can usually run things up until parts get too hot.  The lower the resistance of the switches and inductor, the higher the currents it can handle without getting too hot.  The frequency and inductance are not a major factor, though there are loading considerations that come into play due to those characteristics. 

For the converter I posted in this thread, I'm using the lowest resistance switches and inductor I can get a hold of within my size constraints.  The inductor is only 2.6 mOhms and the switches are only 3 mOhms at 4.5V VGS.  This allows me to run those big currents without major heating issues. 

One nice thing about e-cigs is you don't have to design for continuous outputs, makes a big difference in heating considerations.  Most controllers have a 10 second atomizer cutout which mitigates converter heating.  I use a 30 second cutout to provide a much more liberal atomizer on time so I have to design for better thermal performance.

There's also the consideration of PCB heating and it's why the current paths on my board are so large.  Typically smaller ones are used, but it results in higher losses and additional heating.  I'm using the largest paths I can to minimize those losses as much as possible. 

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #29 on: June 29, 2014, 04:57:26 AM »
So I came up with yet another topology I can try.  I haven't heard back from my TI guy so I don't know if I can get the last design to work to my satisfaction.  Decided to push ahead with something different.

It's not an uncommon thing to use a buck controller to build an inverting buck-boost converter.  There's lots of papers on it if you look it up.  It's done by swapping the switch positions and changing the ground reference.  The problem in using one of these to build a mod is you end up with negative output voltages which are impossible to deal with in terms of measuring outputs with an MCU.

I played around with this topology in the simulator for some time and discovered a way to make it non-inverting.  This is really cool.  Similar to the first one (non-inverting buck-boost using a boost controller), I only have to change one connection.  All I have to do is change the battery negative connection from ground to the low side of the inductor and voila, I have a non-inverting buck-boost using a buck controller. 

I've been simulating this topology all evening and it works beautifully.  The sim shows I can run up to 50W/20A no problem.  However, if I build it I'm not going to run that high of a limit, probably do 40W/15A like the last one, just don't need any more than that.

One of the big problems with the first one (non-inverting buck-boost using a boost controller) is I end up with a voltage shift at the atomizer with respect to ground.  Not a problem functionally, but it makes atomizer ground not the same as battery ground.  This can be be a problem for current loops through the charger if the atomizer makes contact with a grounded object.

This new one is not without issue either.  It does a similar thing but on the battery side instead of the atomizer side.  It shifts battery positive by atomizer voltage.  This is a problem since it requires everything connected to the battery to be 12V tolerant.  I need a regulator for the MCU and a charge-pump that can handle those voltages.  It's one of those things that impacts a lot of other things.  Also, being based on a a buck topology the input is noisy and output is quiet.  Boost is opposite, output is noisy, input is quiet.  I would have to more vigilant about decoupling logic circuits from noise with this one which is kind of a ding.

I think I'm going to try out this new topology on the bench so I'll post some drawings when I get to it.  It simulates nicely so I expect it to work well.  It may be the one to base a design on.  The first one works really well in terms of output capability, but that switch dead time is a showstopper, kills efficiency and I won't skimp on that.  I'm not going to build a mod with a design that's anything short of outstanding.

Offline Visus

  • PV Master
  • *******
  • Joined: Oct 2013
  • Location: Nexxus
  • Posts: 929
  • Karma: +62/-0
  • Gender: Male
Re: Working on this
« Reply #30 on: June 29, 2014, 05:39:05 AM »
Woot more things; and now small fry mods you can make in basic just pot, batt, enclosure,  out of this one you just proofed built, swap out batts lol...  No reason to shelve it  :laughing2:  Oh yeah

For inspirational Sunday the China builders boast 96% on the sx 350     btw  freaked_out:


Gotta get em -- home mod curve 94.5%  :cheer:

Offline miskol

  • Super Member
  • *****
  • Joined: Mar 2014
  • Location: Malaysia
  • Posts: 106
  • Karma: +9/-1
  • Gender: Male
Re: Working on this
« Reply #31 on: June 29, 2014, 05:42:09 AM »
one of these days Craig... i'm gonna have full understanding of what you're saying  :laughing2: but until then, please keep up ur good work and keep sharing ur knowledge.

Good luck and thank you!!

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #32 on: June 29, 2014, 02:52:12 PM »
the China builders boast 96% on the sx 350

It's easy to claim whatever specs you want and China makers tend to lie and cheat about specs wherever they can.  Is this the case for the YiHi products?  Maybe, maybe not.  I'd have to bench test one of their boards to say for sure.  From the drive-by posts I've read here, they're at the top of game when it comes to e-cig controller boards, but I also find it hard to swallow all the the hype coming from a Chinese maker. 

This forum is about building your own stuff for your own use anyway.  Does it matter if it's not as good as something you can get already produced.   If it does, it kind of makes mod building at any level rather pointless, doesn't it?  I know exactly what goes into the boards I make and it's an interesting hobby to build them anyway.

In the end, the high power stuff is probably more grandstanding than anything else.  I admit I'm guilty of that to some extent.  The vast majority of vapers have no need for that kind of power. 

Online Breaktru

  • Administrator
  • PV Master
  • *****
  • Joined: Nov 2008
  • Location: NY
  • Posts: 3344
  • Karma: +792/-4
  • Gender: Male
  • PLEASE participate in this forum
Re: Working on this
« Reply #33 on: June 29, 2014, 04:17:23 PM »
In the end, the high power stuff is probably more grandstanding than anything else.  I admit I'm guilty of that to some extent.  The vast majority of vapers have no need for that kind of power. 

Amen to that brother

Offline CraigHB

  • PV Master
  • *******
  • Joined: Nov 2011
  • Location: Reno, Nevada
  • Posts: 2023
  • Karma: +246/-1
  • Gender: Male
Re: Working on this
« Reply #34 on: June 30, 2014, 04:28:54 PM »
Yes, sometimes you have to step back a little and put things in perspective.  It's easy to get caught up in things that are not all that important.

Back to my power converter ramblings, I swore off Linear parts because of my dissatisfaction with their 4 switch buck-boost controller I tried a few months ago.  It had a tendency to burn out when driven up to the maximal voltages I was trying to achieve, though it worked well other than that. 

I'm back looking at the Linear parts again.  They just have such a wide selection of power parts with designs that meet unusual requirements, it's hard to avoid them.  Just have to treat their parts on a one to one basis, try each one on the bench and see how it does.  I've used other parts of their's before with good results.

I found this Linear charge pump that is one of the coolest electronic components I've seen (http://www.linear.com/product/LTC3245).  It has a really wide input voltage range and can provide outputs both above and below input voltage.  It has micro-current draw making it good for minimizing power consumption when the electronics are idle. 

It's pretty much impossible to find a charge pump with an input range like that and it makes it possible to do this buck controller based buck-boost converter I want to try.  To do this design, I need a 3.3V and 5V source independent of battery voltage and this part fits that bill.  Another nice thing about that part is with such a wide range I cold run series cells if I want. 

Breaktru Forum  |  eCigarette Forum  |  Modding  |  Topic: Working on this
 

gfxgfx
gfx gfx
Powered by MySQL Powered by PHP Valid XHTML 1.0! Valid CSS!