H6Flex 6.6Amp Adjustable Buck

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
Finally got some time to finish assembling the protoype h6flex driver now that the h6cc is shipping and in use.

I figure as something different I'll document some of the steps involved in taking a design into production. Will (wquiles) suggested some folk may get a better appreciation of what goes into a driver design.

Given I've just started on the h6flex, it seems like an opportune time to take pictures and document some of the steps.

So, here's a picture of a prototype assembled and ready for programming and initial testing:

h6flex.jpg


The prototypes are made as production boards, but to save cost and allow for quick turn and also to allow running several 'flavours' of design and other boards, they have no soldermask/silkscreen and they are also on uncut panels. One key advantage of the bare boards is the ease of making any cuts/jumps and the ease of probing any traces since there is no insulating solder mask in place.

Not having a soldermask makes soldering a little more tricky since not having the mask causes the solder to bridge more easily. But, with good quality flux (I use an RMA flux that is water soluble and washable in a $40 ultrasonic cleaner). With experience I can easily solder fine pitch IC's like the switcher IC on the board.

Once all soldered down it's a case of buzzing out key 'suspect' connections to make sure everything is good enough for power to be applied. I use a current limited supply and slowly raise voltage up from zero to ensure no shorts or high current draw.

The next step is to modify the source code from the other *flex boards and prepare to program it into the prototype board.

h6flexc.jpg


The programming & test jig for the h6flex. When laying out the h6flex prototype PCB I also layed out a programming & test jig. The programming part has a 6 pin ISP header to go to an Atmel Dragon (programmer/debugger) and pogo pins to mate with the needed 6 connections on the h6flex board. The test part of the jig has pogo pins that mate with the input power/LED output drive, switch inputs and STAT LED connections.

h6flexb.jpg


Here's a shot of the test area of my work bench. Two power supplies and a Tek 200MHz digital scope. Just visible in the lower right is a 50MHz Tek current probe interface that connects to the Tek scope. This allows me to view voltage & current on the scope.

h6flexa.jpg


After testing various PWM values and charting the resulting output current I have the necessary data to create the various current tables for the ui-uni2 firmware. Typical of all my flex drivers I provide the user several current tables to choose from. Given this driver is capable of up to 6.6A output I will likely provide 6 tables (tentatively):

1400mA
2000mA
2800mA
5000mA
5600mA
6600mA

Here you can see the excel spreadsheet showing PWM versus output current - very nice & linear.

h6flexd.gif


I'll continue to document the process in this thread as testing continues and then the steps to finish up the design for production boards.

Here's a screen grab to show the PCB layout as it stands, with some minor changes already implemented as refinements to the prototype design are discovered.

h6flexe.gif


cheers,
george.
 

netprince

Enlightened
Joined
Sep 7, 2008
Messages
547
Interesting details. I've been looking forward to this driver for a while. Great work...:clap:
 

Al Combs

Enlightened
Joined
Jul 2, 2007
Messages
872
While it's still in the development stage, could you make it fit in a Mag-D battery tube? That would be awesome.
 

Hill

Enlightened
Joined
Apr 11, 2008
Messages
318
Location
MA
George,

I think many of us are drooling while waiting for this to become available. I have the host ready to go. Just need this beauty to bring it to life!

Thanks for making this a reality!

Hill
 

Mettee

Enlightened
Joined
Aug 5, 2008
Messages
678
Location
Arizona
Nice work George,

I didnt see a pot for current adjustment as in the H6CC but I did notice you gave some current steps...

1400mA
2000mA
2800mA
5000mA
5600mA
6600mA

Are those going to be fixed levels or will current be programed? ETA: I read it again, where you mention current tables....didnt see that at first :)

drew
 
Last edited:

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
Drew, no trimpot needed with the h6flex - all the current selections are 'digital' as per my other flex drivers (as you figured out...).

Anyhow, I've added a 7th current table, so my proposed choice of current tables are now:

1400mA
2000mA
2800mA
3600mA <- new one added
5000mA
5600mA
6600mA

These of course are the maximum current outputs that each table will support. Folk that aren't use to this idea can read one of the flex manuals (e.g. hipflex) on my website.

The firmware of course inherits UIP, UIF and UIB since it is the same UI-UNI2 firmware as my other flex drivers.

I've decided to change the uController on the driver. Currently an attiny85 (as per all my other flex drivers), but I need an extra I/O so will use an MLF attiny861 instead.
----------

Anyhow, here's a few more pictures to keep folk in the loop as the proto goes through its paces. The h6flex architecture has digitally adjustable constant current output AND PWM capability. With the adjustable current I can reliably reach around 100mA output, though to be conservative I'll set 200mA as the lowest output current (in constant current mode). To go lower I'll use the PWM hardware to give lower 'average' current (PWM the 200mA low to go lower...).

So, here's a picture of 2 P7's (in parallel to emulate an SST50 type load) running at 20mA average output (this from a driver that can just be clicked to 6.6A output), not bad dynamic current control range :)

The green lit LED on the test jig was caught as it flashed on (the voltage monitoring STAT LED) warning that the input voltage is too low (set for a nominal 14.4V li-ion pack, but the power supply is set to just under 12V).

h6flexf.jpg


And this image capture of my Tek scope shows the 1 in 10 duty cycle of the 200mA output - average 20mA output. As can be seen the rise/fall time of the LED pulses is excellent. PWM frequency is nominally 370Hz

h6flexg.gif


I can easily go even lower than 20mA average, but I think that's more than low enough for this type of driver.

I've measured standby current with the driver 'clicked' off at about 80uA, so that's sufficiently low to not drain batteries over a few months of non-use.

I don't see any need to run efficiency tests since it uses the same switcher core as the H6CC and I've already run quite a few tests (see the H6CC thread if interested).

I have a few more tests to run with the prototype and then I'll tidy up the PCB layout with the new uController and see about doing another PCB run.

cheers,
george.
 

HarryN

Flashlight Enthusiast
Joined
Jan 22, 2004
Messages
3,976
Location
Pleasanton (Bay Area), CA, USA
Hi George, Very nice work.

I always have been impressed by your work and products, but even more so recently as I have had to contract with a guy for a couple of boards. It is just amazing that you can keep all of that power and functionality on a small board like that.
 

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
I didn't realize it's that big, so it won't fit mag C tube then?

Nope - a C mag would have insufficient heatsinking for any appreciable runtime at 6+ amps.

Given the design goal of this driver is to have components on one side only (for easy heatsinking) and be stable at 6+ amps indefinitely and has the full ui-uni2 firmware with voltage monitoring/warning and reverse polarity protection - you would be really pushing to have it fit in smaller area AND still be easy for a user to solder the wiring in place.

Remember, this driver has been tested with 6.6A to 3 pairs of P7's - i.e. 70+ watts driven, we're not talking an insignificant power level for a 1.3" diameter board...

The diameter was chosen to fit a D mag.

cheers,
george.
 

PCC

Flashlight Enthusiast
Joined
Oct 28, 2007
Messages
2,326
Location
Sitting' on the dock o' The Bay...
The green lit LED on the test jig was caught as it flashed on (the voltage monitoring STAT LED) warning that the input voltage is too low (set for a nominal 14.4V li-ion pack, but the power supply is set to just under 12V).
Does this mean that we would need to run this on a minimum of 14.8V (4 Li-Ion cells)? I was hoping to run this on 11.1V.
 

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
Does this mean that we would need to run this on a minimum of 14.8V (4 Li-Ion cells)? I was hoping to run this on 11.1V.

The voltage monitoring is configurable (read the hipflex manual).

The driver will operate from 6V to 24V. So, for one SST-50 use 2 li-ion (or more). For 2 SST-50 use 3 li-ion (or more). For 3 SST-50 use 4 li-ion.

The driver will ship configured for 4 li-ion as hipflex ships - but you can change that via the menu system, again, read the hipflex manual.

cheers,
george.
 

Mettee

Enlightened
Joined
Aug 5, 2008
Messages
678
Location
Arizona
I cant wait for this board it will finish a build I am currently working on...hopefully George, you wont mind me really pushing the board hard :)

drew
 

pepko

Newly Enlightened
Joined
Jan 11, 2008
Messages
143
Location
Slovakia
350mA (super low mode)
1400mA
2800mA
4000mA
5000mA
5600mA
6600mA
 
Last edited:

georges80

Flashlight Enthusiast
Joined
Oct 23, 2002
Messages
1,262
Location
Sunnyvale, CA
350mA (super low mode)
1400mA
2800mA
4000mA
5000mA
5600mA
6600mA

You are confusing current tables with levels. The super low will be 20mA as per the tests above.

You will have a choice of 7 current tables as outlined in my previous post. i.e. you will choose (you can change it via the menu system) one of the following current tables that will correspond to the maximum output current:

1400mA
2000mA
2800mA
3600mA <- new one added
5000mA
5600mA
6600mA

For each current table you will have a choice of 5 (UIP), 8 (UIF) or 2/3/5 (UIB) levels from 20mA UP to the max for the chosen current table. Levels will be roughly a doubling of current per step for reasonably linear 'eyeball' changes of intensity.

PLEASE read the hipflex manual to see how current tables and levels interact...

cheers,
george.
 

ednspace

Newly Enlightened
Joined
May 19, 2009
Messages
6
As an electronics tinkerer I was interested in your testing and programming jig. It looks like you have some spring loaded pins that contact the main driver board for programming and test points.

Is this whats going on, or could you explain a little bit more about your development process? How did you go about making your test jig?

I have some RGB LED boards I have been developing and have been programming my surface mount microcontroller with a 3M chip clip. It works but is very finicky with the contacts. I am hesitant to add a full on programming port, as once I get the code down these should be quick flashed with the firmware and done.

Thanks for the inspiring post!
I really love seeing the design side of the electronics for these drivers.
 
Last edited:

SmokedCPU

Newly Enlightened
Joined
Jan 9, 2010
Messages
85
Location
Quebec, Canada
making these level of devices from a single guy, not from a big corporation nor a butching factory, all Quality checked before shipment, it is mesmerizing !

Congratulation !
 
Top