Desperatly looking for driver (updated kennan driver?)

[jtr1962]

Re: Desperatly looking for driver

$3 or $4 a driver might be very doable if those driver ICs would drop a bit in price. For example, if I could make a driver with about 4 or 5 parts costing less than $1 total, including the PC board, then I can sell it for perhaps $3.50 or $4 each, and still make a reasonable profit, provided I can get interest in 250 to 500 pieces. I'll add "ultra-low cost driver" to my list of "to be designed", but I probably won't have time to do it anytime soon. Also note that these things get way cheaper when size isn't an issue. A wall wart is huge compared to something to fit in a AA flashlight. That accounts for a good deal of the higher cost. I do agree $30 for a driver is ridiculous in most cases.

I use wall-warts or plain AC transformers, combined with one of my drivers, for all my 120VAC stuff. I agree those 120VAC LED drivers are way overpriced (and most are huge-the size of fluorescent ballasts).

BestHongKong has some 12V drivers for $5.95 here. Still a little higher than you're looking for, but not anywhere near $23.

 

[frenzee]

Re: Desperatly looking for driver

I was wondering if anyone here has ever used the Linear LT3467A in a boost circuit. This thing is unbelievably tiny, smaller than an ant, but it's supposed to deliver up to 1.1A of switching current. Although the LTSpice simulator that Linear has shows that it can come up to and sustain my needed current in under 1 millisecond using only a 3.2µH choke, I am having a hard time believing those figures. I am about to start building the circuit, but I imagine it will be quite a challenge considering the ultra-small form factor. I was wondering if anyone has had any experience with this particular driver or with soldering the DFN form factor in general (3mm x 2mm, 8 legs).

 

[snarfer]

Re: Desperatly looking for driver

I've been doing a lot of research on drivers lately.

Not so sure on the LT3467/3467A. On the plus side it's small and has an integrated switch. Also switches very fast. But nearly two dollars a chip in quantity seems pretty steep. Also it appears to be voltage-mode control, not current-control, so how do you deal with Vf variation and heat?

My latest conclusion is that the best solution, at least for my purposes, where I need to be able to control via a microprocessor at fairly high speed for color mixing, is actually the ZXSC310 (mouser sells them for something like 70 cents, and they're only 38 cents in quantity!) run in buck mode with an external transistor switch. The other cool thing about the ZXSC is that the sense resistance threshold is only something like 20 mV, so a nice gain in efficiency there over other constant current solutions. And in boost mode it can source current from some ridiculously low voltages.

Well I haven't got them yet. I'm in Thailand and it takes forever to get stuff from the US. But anyway just wanted to share that, since it seemed like a much better
(and cheaper!) driver solution than anything else I've seen out there.

 

[jtr1962]

Re: Desperatly looking for driver

The ZXSC310 won't work here-the maximum input voltage is only 8V, requiring an external voltage regulator. I remember when I was researching drivers a lot of the Zetex stuff at first seemed good. However, once you looked at the data sheets, you found the need for external pass elements (that's another part to be soldered, irrespective of cost), and in some cases (as with the ZXSC310), the need for extremely low-value (i.e. expensive) sense resistors.

Unfortunately most (all?) of the integrated buck drivers, which is what is needed in the original poster's application, do indeed run $2 or more even in quantity. A lot of those which might be marginally cheaper (i.e. external pass element) had other restrictions.

One Zetex driver I'm liking more and more is the ZXLD1360. This is a buck driver with an input voltage range of 7V to 30V. Only $1.40 in 500s, internal pass element, 100 mV sense voltage (thus can use normal value cheap sense resistors), low parts count ( only need the inductor, Schottky diode, sense resistor, and input capacitor ), and an external input for either current modulation or PWM. Still not cheap enough to make a $4 retail driver board with (for me anyway), but at least a step in the right direction. I could probably sell a ZXLD1360 based driver board at perhaps $6 each were I to get interest enough to build 500 pieces.

 

[snarfer]

Re: Desperatly looking for driver

I was just looking at prices for sense resistors. 0.05 ohm "current sense" resistors are like 40 cents apiece! But then when you check the "regular" 0.10 ohm resistors, they are like 2 cents. I guess with only 19mV you might be able to use a bit of PCB board instead.

I hadn't thought about how annoying it is to have the extra pass components to deal with. In my case it was not such a big deal as I was researching how to put a very large number on one board, so a power supply was a given anyway.

Actually the OP said 12-15 volts input, so the boost regulation was kind of a sidetrack to that.

I see what you mean about the ZXLD1360. That part appears to be the best fit for the needs of the OP.

 

[space]

Re: Desperatly looking for driver

The LT3467 isn't really a good choice for hi power (500-1000ma) LED driving. It just won't be able to produce enough "juice", and efficiency will hurt when one start to squize much current from it. Current feedback also is more complicated unless one goes for low efficiency design.
For constant voltage design and output up to 350ma I think it could do ok.

(I've used the LM2623A myself which is close to the same IC but with 2,2A switching current.)


space

 

[nanotech17]

Re: Desperatly looking for driver

OK, so I bought these drivers in the past:
http://www.kaidomain.com/WEBUI/ProductDetail.aspx?TranID=2982

They are absolutely good stuff, very reliable and efficient. And they have the input range that suits me (automotive purposes).

The problem is that I ordered them once again from Kaidomain in mid November. I paid for 20 pieces. I don't remember if they were available at the time, but I think I would have seen the 'out of stock' text.

The thing is that the drivers have not arrived yet, after three months. Fortunately I have managed to enter into a conversation with people from KD (which is like a miracle, as they usually do not reply) and they told me that the supplier give the item in really small batches, so I'm waiting in queue. I can get a refund if I want.

But I want the drivers because I need them. So does anybody know where I can buy (at a similar price) drivers for 12-15V (this range interests me)?

I am stuck with my projects for three months, waiting like an idiot.

would like to try this - http://www.besthongkong.com/product_info.php?cPath=19_51&products_id=447

 

[jtr1962]

Re: Desperatly looking for driver

I was just looking at prices for sense resistors. 0.05 ohm "current sense" resistors are like 40 cents apiece! But then when you check the "regular" 0.10 ohm resistors, they are like 2 cents. I guess with only 19mV you might be able to use a bit of PCB board instead.

That's probably exactly what the designers of the ZXSC310 have in mind. It can even work if precision isn't too important. At best with PC board traces you'll be able to get perhaps 20% repeatability from lot to lot due to variations in copper thickness, undercutting, etc. Add to that variations in copper resistance over temperature and that's probably 25% tolerance. Still good enough for a lot of applications I suppose, although I usually aim for 5% or less in the stuff I do. Last run most of my boards fell within a 2% error band of nominal.

I hadn't thought about how annoying it is to have the extra pass components to deal with. In my case it was not such a big deal as I was researching how to put a very large number on one board, so a power supply was a given anyway.

A power supply to a bunch of driver ICs on a single board is quite feasible. Regarding the extra pass components-it's another cost if you're doing mass production. I figure $0.15 to $0.25 to mount each extra component when I calculate costs. One driver design I did startd out with 14 components. Next interation will be only 7. That's going to cut in excess of $1 off the final price to the customer, as well as making my life easier assembling these things.

I suspect the price of these driver ICs will crop dramatically in the next few years. For IC manufacture, hundreds of thousands for such niche products is what results in these high costs. Once LEDs become ubiquitous in cars and homes, they will be selling millions of these driver chips. Probably the low price on ZXSC310 is already the result of this driver being used by the millions. We can hopefully expect a lot more of the $3 or $4 driver chips to fall under $1 in a couple of years, making $3 driver boards a reality.

 

[MikePL]

Re: Desperatly looking for driver

Soo.. when I was about to give up, the drivers from Kaidomain arrived. After over three months. But either I got the wrong driver or maybe the driver was updated.

The result is that I have 'some' drivers and I don't know if it's the updated kennan or how to connect the cables to it. Below are photos of the typical kennan driver and the photo of my new kennan? driver.

The typical one:

My new driver:

 

[jtr1962]

It's the same IC (PT4105). This looks like it's probably the datasheet. Just figure out where the leads went on the old board, and make sure they connect to the same places on the new one. Ground is probably the cathode on the Schottky diode (i.e. the side without the band) and also pin 7 of the IC. +12V will be pin 3 on the IC. The LED anode should be that round pad on the left of the inductor. The LED cathode will be pin 8 on the IC (the FB pin).

 

[MikePL]

So please correct me if I'm wrong. I only marked the LED connections as the input positive and negative cables can be soldered on the other side of the board.

Please, can someone confirm that these connections are right? I do not want to send my drivers to heaven and have to wait another 3 months for them...

 

[jtr1962]

So please correct me if I'm wrong. I only marked the LED connections as the input positive and negative cables can be soldered on the other side of the board.

Please, can someone confirm that these connections are right? I do not want to send my drivers to heaven and have to wait another 3 months for them...

That's definitely correct. You can't destroy a buck board by connecting the LED wrong anyway, only by connecting the power leads wrong.

 

[MikePL]

OK, thanks for help. If anything out of the ordinary happens then I will keep you posted. I've also heard that this driver was to be updated and probably I got the updated version.

 

[Gryloc]

Was that the reason that it took so long to ship? I am still interested in these, but I am afraid to be stuck waiting. I looked the board over and it seems silly to make the layouts so much more complex. The old driver was so simply laid out, and it is hard to mess things up since the components were pretty spread out. I see the very same components in the new one as the driver before. I studied the new one to see how it compares to the old one, as well as the datasheet schematics, but I could not see where all the traces went (hidden under the IC and hard to see). I think that everything is fine. I hope it is. :shrug:

-Tony

 

[MikePL]

They are still marked as 'out of stock' so probably they are shipping the drivers to all those who ordered them long ago.

 

[PeterC]

Just collected my long oustanding order for 5-off Kennan (KD sku 2982) from the post office. These are of the new build (with diode wires supplied, but not fitted).

Ordered on 7th Jan '08
Delivered on 5th March '08

Patience is obviously required if you order these!

The circuit of the original design was very close to the typical application circuit shown on the Micro-bridge data sheet for the PT4105 (and applied the recommendations about pcb layout given on the sheet). The new design appears to be based on the same application circuit, but has a couple of additional components (with spaces for several more).

It is not easy to see exactly how the components are connected within the circuit! A bright light, magnifier, and continuity meter/DMM will be needed to reverse engineer the design and see exactly what we have.

My initial thoughts, however (if we're lucky), is that the extra pads and pcb tracking may give some scope for easier mods to the driver board than the original one offered (e.g. changing the feedback circuit to alter the current supplied to the LED.

 

[jtr1962]

My initial thoughts, however (if we're lucky), is that the extra pads and pcb tracking may give some scope for easier mods to the driver board than the original one offered (e.g. changing the feedback circuit to alter the current supplied to the LED.

That's exactly what it looks like to me. See page 5 of the datasheet for some typical dimming methods. The probably have the extra pads so one or even all of these methods may be employed. Good idea since it adds nothing to the cost of manufacture but greatly increases the flexibility.

 

[MikePL]

Hmmm... I've got a problem. It seems that the connection above is wrong. Today I wired two drivers that way and connected an LED. The LED chip was barely giving some light. I thought that maybe the driver is dead, so I used the other one. I got the same results, the LED was so dim that it was barely visible. Input voltage was 12V from a computer PSU.

So do you have any other suggestions on how to connect the LED wires?

 

[jtr1962]

Try connecting to the other side of R2 (there's a pad between R1 and R2 which looks like a connection pad). It seems R2 is there to protect the IC. Problem is when you connect directly to pin 8 as per the schematics in the datasheet current will be limited to well under 1 mA due to the presence of R2.

 

[MikePL]

Thanks for help. It now works absolutely fine. Hooray, I finally have them damn drivers

Edit: just to clarify what connection we are talking about, I've decided to post another image.

This is the right connection of the LED wires: