KD (kaidomain): Deals and new products thread

There's variations with the 7135's but we typically program them at over 8 kHz. Actually with 13A's, we use 9.4 kHz phase-correct PWM or 18.75 kHz fast-PWM.

Biscotti, STAR firmware, etc., pretty much all use these PWM settings.

You mean 9.375KHz phase-correct/18.75KHz fast PWM, Tom E?

AMC7135 PWM LED control – works? @ Pratik Panda

The aforementioned above link has errors on the period scale. Actual turn-on/turn-off times: 80μs/40μs.

Of course, 9.375/18.75KHz PWM frequency is outrageously fast (and wrong!) to drive AMC7135s. However, if you are to question my reasoning, have another cup of evidence:

HKJ's Test/Review of Qlite Rev.A 7135*8 Multiple Modes Circuit Board 3.04A

A little briefing of HKJ's quotes:

“The driver does not disable pwm at high.”

“Due to the fast pwm, the driver does not turn fully on. This makes the driver drop out of regulation at a fairly high voltage.”

“Due to the high pwm frequency, the curve it not square, but rounded.”

“In low mode the light is never in regulation.”


Fellows, the irregular shape and level of the regulator's output during the turn-on phase tells me its efficiency barely reaches 40% for that 8μs (turn-on) period.

With all of this in mind, it is clear to me that:

AMC7135's switching frequency is slow, so if PWMing them this has to be taken into account: slower PWM frequency (much slower!). I wouldn't go above 2-3KHz to minimize the impact on energetic efficiency.

0K. Time for a brief retreat.

Cheers ^:)

Originally posted on Thu, 12/08/2016 - 05:35; edited for fixups.

Isn't that what I said?? Don't tell me you are also anal bout capitalization, spaces... That's a QLite with the 380 mA 7135's HKJ reviewed - those are the bad ones, stay away from them. I've been using 350's from FastTech now for a long time - much better.

The guy in the link you gave is referring to 350's, not 380's, and says 6 kHz is ideal.

If the driver uses 8 7135, why is it not possible to use for high mode 5, for medium 2
Finally for tuning the mode spacing and low and moonlight the last with pwm?

No, i mean the tail switch, the ‘clicky’.
Especially when you push it a little.

Problem solved, thanks KD.
(and sorry about the off topic)

Tom E, my previous post wasn't explicitly directed towards you.

You said “9.4 kHz phase-correct PWM or 18.75 kHz fast-PWM”, while my question was “You mean 9.375KHz phase-correct/18.75KHz fast PWM, Tom E?”. It was aimed to explicitly comfirm the lower frequency was exactly half of the first one (and their interrelation, as I didn't knew yet what the difference between both signals was about).

I know nothing with regards to “bad” 380mA 7135s.

If binning is all the difference between 380s and 350s, then their switching behaviour and timing is likely to be exactly the same.

The guy at Pratik Panda can opine whatever he wishes, me too.

I can be anal about anything, if I wish to be so.

I've provided plenty of evidence to demonstrate how wrong is to apply such fast PWM frequencies to a device which needs a whopping 8μs just to turn-on properly. HKJ pointed this out back then, yet no one seemed to care.

Jerommel, aaaah! Interesting for you to say that. Such a behaviour indeed has to do with the fact that the frequencies at which those AMC7135s are switched lay on the audible range. Some sort of reverberation is likely to happen somewhere inside those tail clickies.

Cheers ^:)

Originally posted on Thu, 12/08/2016 - 13:03; edited for fixups.

The DTP triple parallel board is available now:

http://www.kaidomain.com/p/S026381.KDLIGHT-Triple-3TP-20-DTP-Copper-MCPCB-for-Cree-XP-Series-Nichia-219-Series-3535-LEDs-2pcs

2 for 2.98 $US

The short answer is because there are no PCB’s for it.
Silly isn’t it?

The TripleDown and TexasAvenger use a FET for Turbo and 1-6 7135’s for high mode, then PWM’d for lower modes. It is a bit of a compromise mainly limited by space on the board and the amount of pins (8) your common Attiny has to offer. Of those 8, normally one is used for power, one for ground, one for mode switching, one for programming via clip, one for voltage detection, and the other 1-3 for controlling the output regulators.

I think MikeC has done essentially what you are thinking of Lexel, but it didn’t make much buzz because it is a little complicated.

See here:

I may have misunderstood this, but i thought FET drivers are not constant current or current limited?
It just PWMs full blast, doesn’t it?
And aren’t there FETs in a 7135?
My question is actually: what’s the voltage drop over a FET and over a 7135?
if it’s almost the same i would prefer 7135s.

Switched on in banks of 1, 2, 4 and 8 for total of 15 x 7135 or 1, 3 and 9 for total 12 x 7135 or 1, 2 and 4 for 7x…etcetera…

Check the link

Thanks ! :smiley:

Edited: What the hell? User DEL published a convenient explanation with regards to what I was saying about AMC7135 switch-on times, and now it's gone?

Fellow DEL, deleting your contributions is of no service to us. :-/

Related thread link I dug up from my browser's history: https://budgetlightforum.com/t/-/35447/250

Thanks (DEL) for the explanation and all that information. I started to presume the Pratik Panda guy made some sort of mistake with the scale while determining the switch-on/switch-off times, because I just couldn't figure what the hell was going with those oscilloscope screenshots timebase, lenght of pulses and their stated PWM frequencies (?).

Yes, there I can see the 6-8μs turn-on time, with a 50μs duty cycle pulse, and he is speaking about a 2KHz PWM frequency, which is a 500μs long period… bah, nevermind, there's too much stuff I was supposing.

He said turn-on + turn-off times to be ≈12μs (83.3+KHz), and it didn't made much sense to me that the regulators could be having so much trouble with 9.375/18.75KHz frequencies.

Jerommel, I guess you'd like to drive a sort of tiny 7135 DAC. Doubling the driven regulators per pin each time, with just 3 pins we could drive from 0 to 7 regulators in parallel (and PWM combinations). You said 15, that's 4 bits (maximum unsigned integer value for 2n, n being the number of bits/outputs).

Cheers ^:)

Originally posted on Thu, 12/08/2016 - 20:56; edited for fixups and clarification.

Well,my own plan is to just use a 6x 7135 Ak47 or 105 driver and add a momentary switch to engage an extra bunch of 7135s for turbo mode.
This way you can not leave it on in turbo and when too hot you will let go of the button (which would be in the head of the flashlight).
We’ll see…
But i’m not into electronics, PCB design and / or chip programmng at all, but i like the idea of a momentary turbo button.

You said it all, Jerommel. 7 regulators driven by the microcontroller (3 bits/outputs), plus your “magic trigger” to enable the wild bunch…

Cheers ^:)

The 380 mA 7135’s pre-populated on QLites are bad? or generally all 380’s (such as these from Fasttech ) fall out of regulation?

The properties are different. Sorry, don't understand it all, but for example with a 350 7135, you can get down to a PWM value of 2 or 3, while with a 380 you can only go as low as 4 or 5. So the 350's have a better range. I don't think it's as simple as binning them differently. The problem is the 7135 has gone obsolete years ago, so what's on the market now is clones, or clones of clones, so I wouldn't trust the specs you can find, or even the source. TK says there's even differences between 350's, so it's a difficult part to depend on for specs. The ones from FastTech seem to have been consistent. I've had many buys from them over the years, sometimes by the 100, sometimes by 10-50. Also the 350's over perform a little, while the 380's under perform a little, so really there's not much difference. Forgot where but more recently, a few of us were posting about 7135's and pretty much all agreed the 350 were the better ones, and to stay with them. It's what we use now in populating the FET+1 drivers, and the triple drivers.

Maybe this thread should be renamed to "KD Deals and endless esoteric discussion about things not Deal Related that makes you regret your last numerous visits to the thread to hear about the new KD deal and actually is not a deal but a continuation of the endless discussion about some unrelated esoteric concept that should have been taken to a separate thread and then wanting to stab yourself in the head for being so dumb to return to the thread Thread".

That’s a bit long for a topic name though… :stuck_out_tongue: