TK's Emisar D4 review

I have the D4 in 219c and xpl hi and have noticed with battery voltage check the xpl hi reads .1volt lower than the 219c using the same battery Has anybody encountered this as well?

Can we buy D4's driver separately in future?

You can buy a FET+1 @ Mountain Electronics with D4 V2 Ramping firmware.

… in various different diameters. My shipment arrived today! Next project will probably swapping one into a DQG Tiny III triple.

Again no reply from the driver gurus?

I’m not a driver guru, but my guess is potting the driver with silicone, or squirting into the driver cavity lots of thermal grease, or putting a thermally conductive sponge on top of the MCU with the other side of the sponge pressing the bottom of the shelf all would have the same effect: Faster thermal transfer to the heat sensor and possibly a faster response in ramping down when it gets hot.

The potential disadvantage is that Toykeeper calibrated the thermal sensor for this exact light in its stock configuration. She said it doesn’t ramp down temp based on current temperature, but rather it looks at how fast temp is changing and then adjusts output down based on predicted future temperature.

Increasing the thermal conduction to the sensor might cause the temperature to change too rapidly at the sensor and result in excessive stepdown.

Still… only one way to truly find out…… you should test it! :sunglasses:

Okay, its spec page says…

  • H1 1100 Lm (PID, approx 2 hr)
    * H2:
    • 670 Lm (PID, approx 2.5 hrs)
    • 356 Lm (3.9 hrs)
    • 162 Lm (11 hrs)

So, I guess its highest level without thermal regulation (PID) is 356 lm.

This isn’t really surprising, since they have about the same amount of aluminum, same amount of thermal mass and surface area, so they have roughly the same ability to deal with heat. Maximum output at a given level of heat will be similar until we make LEDs and drivers more efficient or add extra cooling.

The Q8, for example, should be able to run significantly brighter at the same temperature, simply because it’s bigger.

I’m glad to see it’s so consistent, and that it reaches a fairly stable state in under a minute. I guess I must have done something right… The measurements are still pretty noisy though, so even with two lowpass filters it still jumps around sometimes at random.

However, I’m not sure I’ll ever make sense of the rainbow. The rainbow remains a mystery.

Thanks mate, been a ruff day’ please excuse my mood all.

Thanks! For a moment, I was starting to think that maybe Cree “5” tints might not be so bad, but your pics reminded me why I don’t use them. I might not mind a high-CRI Nichia at 4000K once in a while, but the Cree 5D looks unpleasantly yellow to me.

The measurement relies on the calibration of the attiny MCU, which is known to have significant variation between units. Voltage measurement varies by about 0.1V and temperature varies by like 12 C.

… maybe? It would probably improve things a little, but I haven’t tried it. I have some thermal transfer foam, but it’s not easy to wedge it between the MCU and the pill, especially in such a tight space with wires in the way.

Exactly.

One-upped again. More lumens, more mAh, more bling. You really don’t know how not to customize anything. :smiley:

Sometimes I mess things up…

You guys are far more accurate/ articulate/ nerdy than I am so your results would be more interesting than mine. There is nothing I would like to change about the firmware except the inyoface LVP warning so for now don’t wanna pull the driver out
Just an idea dudes…

Now we’re all going to have send our D4s to DB Custom to get blinged!

That’s pretty easy to disable. Look around line 1600 or so and comment out the calls to BlinkLVP().

Not sure if this has been discussed yet or possible but here’s an idea: instead of stepping down to around 500 lumens, allow to user to choose the output level (current), depending on the ambient temperature of the environment being used this can be adjusted to suit everyone’s needs. It doesn’t need to be lot of options, maybe just 3: 1/1.5/2A

In my case I think the D4 can run comfortably at ~750lm in 25C ambient temp like my triple S2+. After seeing maukka’s graph and how slowly the output recovers after step down, I think it is better to dumb down the temperature regulation.

Thanks for the tip, but it is bearable for the moment :+1:
I haven’t entered the game of firmware development yet as I anticipate rapid improvements in certain areas within the next year or two, although what’s going on just now is kool as can be, I think my sincerest efforts will be more appreciated some time future.

Are those necessary for 20700 or are you just having fun?

I don’t think calculating temperature from Vf would help. I’ll explain via analogy.

Let’s say I’m blind and I want to go outside, but my guide dog is very photosensitive so I need to select an appropriately-dark pair of doggie sunglasses first. I can’t measure the brightness outside directly, but I do at least have a lux meter and a window, so I check that to determine which glasses to use. This is sort of what the D4 does now.

However, I also have a GPS clock which tells me the time, date, and location. So maybe I could calculate the average brightness at this location based on the time and date plus a reference chart for average seasonal weather conditions. This is like calculating temperature based on emitter Vf.

It seems like it would be less effective.