thefreeman’s HDR Anduril 2 high efficiency drivers - update : FWxA boost driver

397 posts / 0 new
Last post
icpart
Offline
Last seen: 11 hours 6 min ago
Joined: 04/15/2019 - 01:13
Posts: 529
Location: Bulgaria

thefreeman wrote:

LM34936/LM5176 controller, there is another one from Analog, but while both can convert single cell input voltage they require a >8V bias, I haven’t found a BB controller or powerfull converter that can be powered with single li-ion. 8V is not cool because that’s out of reach of small doubler charge pumps, need to use a boost converter to power the BB controller lol, though there are small micropower ones.

The reason I was thinking about it is because somebody I know was working on a high CRI high current 3V LED, but the project was abandoned due to the very high cost of high CRI phosphors, so I lost a bit of interest on this BB driver.

.


I thinking as you. Time ago I had same idea as you for powerful Buck-boost but with LM25118 or LM5175. That’s are previous BB controllers before new one LM5176. LM25118 is asynchronous converter. You right they will need additional boost converter for bias voltage. These days I owned Q8 from friend. I really like it but the driver is sucks without any sort of stabilization. This type of big multi emitter flashlight can benefit from powerful BB. For small one it’s pointless anyway.
I think to design PCB for it maybe with linear driver based on code of Emisar D4 with 1634. I don’t need HR like in your design. Maybe simpler one channel is also good choice. Also I like to use Attiny 1616 and DAC. That’s is big step ahead compared to pwm DAC. Also with integrated DAC we can choose different internal reference voltages. I need to see if gchart is used that in him code. With that it is also possible to achieve good resolution of current control.
thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

Yes he did, he used the dynamic PWM code TK recently implemented to switch the VREF between 2.5V and 0.55V giving a dimming range of 1:1159 with the 8bits DAC.

icpart
Offline
Last seen: 11 hours 6 min ago
Joined: 04/15/2019 - 01:13
Posts: 529
Location: Bulgaria

thefreeman wrote:
Yes he did, he used the dynamic PWM code TK recently implemented to switch the VREF between 2.5V and 0.55V giving a dimming range of 1:1159 with the 8bits DAC.

That’s are great news. So I think it is time to place my order on mouser for 1616 when they received new one.
thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

Farnell has an earlier availability estimate, Feb 2022.

icpart
Offline
Last seen: 11 hours 6 min ago
Joined: 04/15/2019 - 01:13
Posts: 529
Location: Bulgaria

thefreeman wrote:
Farnell has a sooner availability estimate, Feb 2022.

Thanks for the info. I can order from them via their distributor in my country. About replacement for opamp do you know that NCS333A. It seem to be direct replacement of OPA333 and are cheaper. Specs are almost as OPA333. But they are also out of stock for now.
thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

Yeah the specs look good, there is also the MCP6V66.
Note that I don’t use the OPA333, but its low cost identical equivalent the TLV333, in stock at Mouser and Farnell right now.

Sk9
Offline
Last seen: 1 hour 24 min ago
Joined: 02/08/2022 - 13:29
Posts: 96
Location: Canada

Is there any chance of getting a FW1A boost driver made or designed?

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

Sorry I haven’t updated this thread in a while, I’ll try to provide the Ospark PCB design and BOM soon.

Sk9 wrote:
Is there any chance of getting a FW1A boost driver made or designed?

Designing will be easy, I’ll just add the switch ring to the BST18 (18m clearance) driver.

Aside from that I’ve been working on a special driver, here is a picture of it if anyone can guess what it does Question

id30209
id30209's picture
Offline
Last seen: 4 days 7 hours ago
Joined: 05/17/2018 - 12:20
Posts: 2327
Location: Croatia

I have no idea what it does but i swear to god i’ll buy all of them from you Smile

WTB Titanium 4sevens 2xAA tube

Haukkeli
Offline
Last seen: 11 hours 6 min ago
Joined: 06/24/2021 - 05:27
Posts: 522
Location: Suomi

Good job again!

Quadrupel
Quadrupel's picture
Offline
Last seen: 8 hours 55 min ago
Joined: 12/03/2017 - 10:40
Posts: 1203
Location: Lithuania

2ch? OR x2 power? Big Smile

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

One channel Smile
Here’s a picture with the markings visible :

ArtieT59
ArtieT59's picture
Offline
Last seen: 5 days 43 min ago
Joined: 02/25/2020 - 17:55
Posts: 3598
Location: CT, USA

Thefreeman- I swear I don't get as excited about anything in this hobby as I do when you build a new driver. 

thank you!

[FLF] Five Light Friday https://budgetlightforum.com/node/78749

Check out some of my new lights (picture heavy) and quick first impressions of them here: https://budgetlightforum.com/node/77180

My Sft40 beamshots / comparison thread: https://budgetlightforum.com/node/78100

The BLF GT with SFT40 (2300lm, 2700m+ @ 30 seconds!)
https://budgetlightforum.com/node/79561

Quadrupel
Quadrupel's picture
Offline
Last seen: 8 hours 55 min ago
Joined: 12/03/2017 - 10:40
Posts: 1203
Location: Lithuania

Cant find by markings, but its probably MPS converters. Separated buck-boost? 2 step boost?

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France
Quadrupel wrote:
Cant find by markings, but its probably MPS converters. Separated buck-boot? 2 step boost?

Close guess Big Smile

The one on the right is a MPQ8633B (buck) and the left one is a MP3431 (boost), the driver first boosts the cell’s voltage (set to 4.4V here), then bucks it down to the LED Vf, Analog Devices calls this type of topology cascaded boost-buck.
So like a classic buck-boost converter it can deliver full current to the LED regardless of the input voltage, of course because it does two conversions the efficiency is not as high as with a buck or boost driver, but it has similar peak efficiency to available buck-boost converters, with much more output current :

I developed a similar driver before which was a cascaded boost-linear driver, I have the boards but never ended up testing it.

Haukkeli mentionned using a buck converter and I went with it, the boost-buck topology should be more efficient at most output currents, except maybe at max current.

Haukkeli wrote:
Good job again!
ArtieT59 wrote:

Thefreeman- I swear I don’t get as excited about anything in this hobby as I do when you build a new driver. 

thank you!

Thanks Smile

Quadrupel
Quadrupel's picture
Offline
Last seen: 8 hours 55 min ago
Joined: 12/03/2017 - 10:40
Posts: 1203
Location: Lithuania

Oookeyyy, poor batterys . Big Smile

iamlucky13
Offline
Last seen: 1 day 5 hours ago
Joined: 06/22/2018 - 09:18
Posts: 1175
Location: USA

Fascinating. Theoretically perfect regulation at max output right up until your battery dies, as long as your host can handle the heat.

Am I correctly understanding that that the distinction here is the cascaded boost-buck topology always first boosts the voltage to an adequate level, then bucks, while the more typical boost-buck topology will switch between boost or buck mode depending on the input voltage?

Were those tests powered by a bench supply or a battery?

Is pad A for single-channel aux LED’s (doesn’t look like there is room for more solder pads)?

Also on a separate topic that still involves multiple inductors – did you ever pursue the dual-channel buck driver you were working on further? I’m still intrigued by this as the ultimate D4 driver:

https://budgetlightforum.com/comment/1795118#comment-1795118

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

iamlucky13 wrote:
Fascinating. Theoretically perfect regulation at max output right up until your battery dies, as long as your host can handle the heat.

Yep, this is mostly usefull for single high Vf LEDs, for example the Yinding round LED can’t be driven to it’s full potential with a direct drive or linear driver since its Vf is so high, but even an SFT-40 with a Vf of 3.5V at 9A will start to drop relatively quickly as the cell discharges. It’s also useful for UV and Laser diodes, although I’m not particularly interested in those.

Quote:
Am I correctly understanding that that the distinction here is the cascaded boost-buck topology always first boosts the voltage to an adequate level, then bucks, while the more typical boost-buck topology will switch between boost or buck mode depending on the input voltage?

That’s right, here are schematics of the relevant topologies :

The buck-boost converter is buck and boost switches joined together with the inductor in the middle. When bucking the TOP-BST FET is always ON (BOT-BST OFF) and when boosting the TOP-BCK FET is always ON (BOT-BCK OFF)

With the cascaded boost-buck converter it’s two separate converters that are always switching.

Quote:
Were those tests powered by a bench supply or a battery?

Bench PSU, much more convenient for testing different Vin, plus if I want a constant Vin for efficiency measurements I need to adjust the PSU voltage depending on the output current due to the voltage drop, which is quite significant because I use precision current shunts for current measurements.

Quote:
Is pad A for single-channel aux LED’s (doesn’t look like there is room for more solder pads)?

Yes, I designed this one relatively quickly and made it single sided for ease of fabrication, at the end there was some space for a single AUX pad so I added it.
It has a 21mm diameter clearance, I ordered it in 26mm for Convoy M21D or M3C, it could also work in a KR1 with an added switch ring on the back, or K1/DM11 with a larger diameter.

Quote:
Also on a separate topic that still involves multiple inductors – did you ever pursue the dual-channel buck driver you were working on further? I’m still intrigued by this as the ultimate D4 driver:

https://budgetlightforum.com/comment/1795118#comment-1795118

It’s another driver I never ended up testing, for two reasons : I have very few TPS62867 buck ICs left and I prefer to keep them for other projects. Anduril tint ramping code would require some big modifications to work with HDR (could have tested without the HDR function though).

Quadrupel
Quadrupel's picture
Offline
Last seen: 8 hours 55 min ago
Joined: 12/03/2017 - 10:40
Posts: 1203
Location: Lithuania
Thumbs Up Thumbs Up Thumbs Up ! Add a linear schematics for reference. People have to know the true:D
kennybobby
kennybobby's picture
Offline
Last seen: 13 hours 55 min ago
Joined: 05/10/2017 - 09:13
Posts: 1396
Location: huntspatch, alabama

Excellent work @freeman

Now i used to think that i was cool,
drivin' around on fossil fuel,
until i saw what i was doin',
was drivin' down the road to ruin. --JT

icpart
Offline
Last seen: 11 hours 6 min ago
Joined: 04/15/2019 - 01:13
Posts: 529
Location: Bulgaria

Analog devices (Linear) have similar controller from years which work in cascade mode. They just didn’t work with low voltages. They are good for minimum 2S DC/DC driver. For example: https://www.analog.com/en/ltc7812.html

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France
Quadrupel wrote:
Thumbs Up Thumbs Up Thumbs Up ! Add a linear schematics for reference. People have to know the true:D

Well, all there is to know about linear regulator is that they are like a variable resistor.

kennybobby wrote:
Excellent work @freeman

Thanks!

icpart wrote:
Analog devices (Linear) have similar controller from years which work in cascade mode. They just didn’t work with low voltages. They are good for minimum 2S DC/DC driver. For example: https://www.analog.com/en/ltc7812.html

Yes I remember seeing that one, using converters with integrated FETs is much more convenient though, if I’m going to use a controller and external FETs might as well use a buck-boost controler.
On that subject I gave a second look at the LTC3785 , I previously dismissed it because of the 2.7V min output voltage since many LEDs can go down to 2.4~2.5V at very low current and because the efficiency curves didn’t looks particularly good.
But using LTSpice the output voltage can go lower, I guess that depends if the simulation is accurate on that point. Secondly, using their calculation tool the efficiency can be very good, ~94% at 3Vin 3.5Vout 10A with the inductor and FETs I chose, it remains to be seen if it’s also accurate.

There are a few other details to think about :

  • because the input and output voltage are low (4.2-2.8V) I would need to use sub-logic level FETs and that restrict the choices a lot, most good switching FETs and synchronous FETs are logic level (4.5V), so are dual channel FETs geared towards DC-DC conversion. In the end I wonder if it would be better to use a small boost converter or charge pump to get the required drive voltage.
  • the compensation network looks a bit complicated
  • burst mode (light load efficiency) doesn’t switch automatically to PWM above a certain current, so the MCU has to change the mode at a certain ramp level, need to additional code for that.

Back to the cascaded boost-buck driver, I’m testing which passives around the buck converter can be removed to reduce the component count

  • the soft start capacitor is in addition to the internal one but despite the datasheet saying it should be at least 22nF I don’t see any ill effects after removing it.
  • the mode resistor set the switching frequency, either 800 or 1000 kHz depending on the value, or 600kHz when the mode pin is directly connected to VCC or GND, I tested at first at 800kHz, at 600kHz the efficiency is pretty similar, just slightly lower at lower output but nothing significant, on top of not needing the mode resistor it allows for a higher duty cycle (ratio of Vout/Vin), which is limited by the minimum off time (180ns)and the switching frequency ( 1-offtime*frequency) : 0.892% At 600kHz and 0.856 % at 800kHz, in reality it’s a bit lower because the switching frequency is 50~100kHz above that, that’s at 0Aout, at high current the voltage drop due to component resistance as to be taken into account.
    Higher duty cycle means that Vout can be closer to Vin, Vin is this case is the boost voltage, which can be lowered a bit, that should help improve efficiency (a tiny bit Big Smile )
icpart
Offline
Last seen: 11 hours 6 min ago
Joined: 04/15/2019 - 01:13
Posts: 529
Location: Bulgaria

thefreeman wrote:
One channel Smile
Here’s a picture with the markings visible :


What is the size of driver 18mm? One side or dual side populated. Also what is the smaller SMD component? Can you share BOM or components sizes?

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

It’s 21mm clearance (white circle), 26mm OD in this case, I’m also working on a 18mm clearance one but I’ll probably have to use 40xx inductors (5030 on this one), there is a 3333 RPP PFET on the back (not populated) and two passives, which will be put on front since I managed to remove two passives, 0603, I’ll share the BOM once finalised.

Important to note : no HDR on this one, so pretty high minimum current, it could be added by placing the HDR FET on the back.

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

I finalised the design of 3 different sizes of single cell boost driver :

  • BST15 : for 17+mm drivers, 60*0 inductor, 3333 RPP PFET, double sided, 1*AUX LED, also supports Mike C’s OTSM firmware.
  • BST18 : for 20+mm drivers, 60*0 inductor, 5060 RPP PFET, double sided, 3*AUX LED. With a BST18-FWxA version.
  • BST21 : for 23+mm drivers, 70*0/ 60*0 inductor, single sided with 3333 RPP PFET and an additional 5060 emplacement on the back. 4*AUX LED. With BST21-KRx version.

And I assembled two BST21s, one 6V and one 12V :


Then took some efficiency measurements, 12V3A, XAL7030 1.5uH, 3333 SISS61DN on the front for RPP :

With an additional 5060 DMP2003 on the back :

Lowering the inductor value to 1uH (5mΩ vs 8.4mΩ) :

6V6A, which as expected is more efficient :

Light Veteran
Offline
Last seen: 6 days 8 hours ago
Joined: 10/17/2019 - 03:36
Posts: 150
Location: Italy

I am in

Flashlight addicted

Hoop
Hoop's picture
Offline
Last seen: 1 hour 11 min ago
Joined: 12/20/2012 - 05:33
Posts: 1332
Location: In my body.

Awesome work! Any plans for a 9V out version? (triple LED)

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France

Thanks Blushing , for 9V it just needs different sense resistors values for 4A and a different FB top resistor value.

FAB505
Offline
Last seen: 17 hours 8 min ago
Joined: 01/15/2018 - 00:58
Posts: 24

These look amazing. They are electronic switch support only, correct? Any plans for versions that support a mechanical switch? Convoy s2, c8, etc.

thefreeman
thefreeman's picture
Offline
Last seen: 26 min 19 sec ago
Joined: 01/06/2020 - 09:56
Posts: 1805
Location: France
FAB505 wrote:
These look amazing. They are electronic switch support only, correct? Any plans for versions that support a mechanical switch? Convoy s2, c8, etc.

Thanks.
BST18 and BST21 are E-switch only with Anduril 2. BST15 has additional support for MikeC’s firmware which supports clicky, E-switch and clicky+E-switch. His firmware isn’t finished though, he’ll probably resume development if I send him a bunch of BST15s.
It also has an off time capacitor emplacement, so it could work with an OTC firmware, that is if DAC and HDR support is added to the OTC firmware.

Pages