TK's Emisar D4 review

[ToyKeeper]

@Agro

Played a bit with thermal regulation.
It’s less smooth than I expected, the output stays visually stable, makes a jump, stays stable for another while, jumps again etc.
Also, I noticed that if I heat it up a bit, turn it off and start at full output, the peak temperature will be much higher than originally. Also, I think it may be exceeding the level that I configured….or maybe my burnt fingers got more sensitive

It can drop down in large steps. Stepping back up only goes one small step at a time though.

The step-downs could be made smoother with a bit more code, but it also has a high risk of introducing bugs. I actually tried it at one point, but the direct approach was too buggy and the indirect approach was fragile. So I went with a more robust but also more visible method.

Also, the regulation is not based on the current temperature, but on the predicted future temperature. This means it’s sensitive to the rate of change. If it’s below the target temperature but rising quickly, it will step down pretty quickly. But if it’s above the target temperature and falling quickly, it won’t step down (and may even step up a little).

When it first turns on in turbo, it will generally exceed the target temperature before it stabilizes. This is because, even with the predictive algorithm, thermal lag still delays the measurements too much to avoid the initial peak. Basically, thermal regulation is rather unpleasantly like trying to steer a fast-moving cow around a hairpin turn on an icy race track. While wearing someone else’s glasses.

@hIKARInoob

Edit:

The higher the temperature increase rate (dT/dt), the bigger the difference between desired maximum temperature (what you felt), and actual max temperature value (Tmax set by D4).

Yes, it’s based on the predicted temperature, not the current temperature. This helps it compensate for thermal lag. Without that, the racing cow was slamming into each turn’s crash wall a lot harder.

@Agro

2. I noticed a quirk in the UI. If I ramp the light up to a fairly high level, wait for thermal management to restrain it an start ramping again, I expect the light to smoothly change output. It jumps to the previously set level and starts ramping from that.

The light makes a distinction between the actual ramp level and the target level. The button sets the target level, while thermal regulation can change only the actual level. This allows it to step back up when the actual level is lower than the target level and the temperature is no longer too high.

To change the behavior, the ramp button code could perhaps use the actual level instead of target level, when they’re different? I doubt it would be a difficult change… and patches are welcome.

@Agro

3. A weird software bug:
   Sometimes the light doesn’t notice that I cut off the power.

If I understand correctly, this isn’t a software bug. It means the power wasn’t disconnected long enough for the driver to lose power. To make this happen, do the following:

• Ramp to a low level.
• Click to turn the light off.
• Wait at least 6 seconds so the light will enter sleep mode.
• Loosen then tighten the tailcap quickly.
• Notice that the light doesn’t blink. Click to turn it on, and see it’s still at the last-used level.

This happens because the standby current is so low that it is able to tolerate short interruptions while it’s asleep.

[ToyKeeper]

@djozz

USB charging.

I wonder how hard it would be to add an inductive charging circuit in a tailcap. I really don’t like poking holes in a waterproof light for USB ports.

Not that I have an inductive charger… but it might be a nice way to do these things without incurring the usual problems associated with built-in charging.

[Fuzzylogic]

@charles lin

…

Can I ask what clips you are using here? They seem to be a much better length than the convoy ones.

[ToyKeeper]

@KawiBoy1428

This reminded me I have a bottle of GitD paint. I still haven’t decided what to do with it.

[ToyKeeper]

@RollerBoySE

Is a zero shown as something similar to half a blink?

Yes. That was one of the things I fixed. Before, ‘0’ digits weren’t shown at all.

[hank]

the regulation is not based on the current temperature, but on the predicted future temperature. This means it’s sensitive to the rate of change.

This is just amazingly smart. Thank you.

[Slayer1]

Thanks again KawiBoy1428!!. Cheers.

[electricjelly]

I received my Emisar D4 in the nichia flavor today. I’m loving it. Thank you Tom E and Toykeeper for nailing the ramping UI. Thank you Hank Wang for bringing this light to market. The quality is top notch and the price is an amazing value. I have another one on its way with the xp-g2 5D tint. I love all the choices available. I had been wanting a Zebralight SC62-3 but this light has scratched that itch for sure. Good job to everyone involved.

[Agro]

@ToyKeeper

@Agro

2. I noticed a quirk in the UI. If I ramp the light up to a fairly high level, wait for thermal management to restrain it an start ramping again, I expect the light to smoothly change output. It jumps to the previously set level and starts ramping from that.

The light makes a distinction between the actual ramp level and the target level. The button sets the target level, while thermal regulation can change only the actual level. This allows it to step back up when the actual level is lower than the target level and the temperature is no longer too high.

To change the behavior, the ramp button code could perhaps use the actual level instead of target level, when they’re different? I doubt it would be a difficult change… and patches are welcome.

I would be happy to produce such patch, but so far I don’t have a way of flashing the patched firmware and so I can’t test it. Developers seem to move away from SOIC8 chips, so investing in a SOIC8 setup at this moment doesn’t seem like a good idea….

@ToyKeeper

@Agro

3. A weird software bug:
   Sometimes the light doesn’t notice that I cut off the power.

If I understand correctly, this isn’t a software bug. It means the power wasn’t disconnected long enough for the driver to lose power. To make this happen, do the following:

• Ramp to a low level.
• Click to turn the light off.
• Wait at least 6 seconds so the light will enter sleep mode.
• Loosen then tighten the tailcap quickly.
• Notice that the light doesn’t blink. Click to turn it on, and see it’s still at the last-used level.

This happens because the standby current is so low that it is able to tolerate short interruptions while it’s asleep.

Confirmed, I can reproduce it with your steps. Why doesn’t it happen every time after a power-reset?

[Ronin42]

TK you have just won the best line in August award.

I cant decide which:

1. “When it first turns on in turbo, it will generally exceed the target temperature before it stabilizes. This is because, even with the predictive algorithm, thermal lag still delays the measurements too much to avoid the initial peak. Basically, thermal regulation is rather unpleasantly like trying to steer a fast-moving cow around a hairpin turn on an icy race track. While wearing someone else’s glasses.”

LOL

2. “Basically, thermal regulation is rather unpleasantly like trying to steer a fast-moving cow around a hairpin turn on an icy race track. While wearing someone else’s glasses.”

LMFAO

And for the rest of us I think you just made a 4th gen light (we don’t actually control it) we give it our input and then it decides the exact output. sounds about right, maybe like modern high performance aircraft we can’t react fast enough.

All Hail TK and her Fly by Wire flashlights! Were not worthy

[electricjelly]

I’d love to see a Emisar 90 degree angle headlight.

[Agro]

I have another usability comment…
I’d prefer the light not to remember the ramping direction for too long. Myself, I forget it seconds after I stop ramping. After that I never know what will the light do if I start ramping again.

[ven]

@electricjelly

I’d love to see a Emisar 90 degree angle headlight.

Hot head!!!

[BeardedRaleigh]

@electricjelly

I’d love to see a Emisar 90 degree angle headlight.

I have always wondered why we have to have the light on our forehead? Why can’t we just use a regular flashlight on the side?

I don’t care either way, just sayin.

[ven]

@BeardedRaleigh

@electricjelly

I’d love to see a Emisar 90 degree angle headlight.

I have always wondered why we have to have the light on our forehead? Why can’t we just use a regular flashlight on the side?

I don’t care either way, just sayin.

They do have their place, mainly to free up both hands when working etc. Disadvantages come when others are around(dazzle people nearby).

[Man_Without_Shadow]

DELETED due to lack of anyone reading, understanding or actually giving a rip

[Agro]

Can’t get D4 out of my head.
This light really shows how UIs of my other flashlights suck.
Very rich, but not complex. Allows to do nearly anything that I want quickly. Doesn’t really have a lot to memorize. Click, click+hold, 2-click, 3-click, unscrew slightly. Not little, but far from the 20-clicks I’ve seen in a similarly rich driver.

Nevertheless I have some thoughts, which may or may not be an improvement.
I just got stuck in the momentary mode. How many times do I need to click to escape it? I needed to look up a diagram. If screwing / unscrewing the tail would reset that, it would be nicer (though I recognise some may want to keep this mode forever, regardless of battery changes).
It would be nice to have a shortcut to roughly-max-thermally-stable mode.
If I enter thermal calibration mode, press and hold until it starts showing the current mode, then start clicking, the light behaves funny. Click turns it on, but at the next digit it turns off by itself and continues showing the number. I guess it would be better if de-pressing the button made the light quit the thermal-calibration mode immediately.

[djozz]

Have been playing with the ramping for a few days, I felt the need for an extra way to get to a maximum sustainable mode directly (so not after letting the, well working, thermal stepdown algorithm figure it out). I’m sure in most situations that is about 2.5 A in this light.

Been thinking of ways to implement this.

There is the short hickup at 350mA, easiest is create a same hickup at 2.5A but once you see it you overshoot it, it is not precise.

A way to implement is to let ramping up stop at the two locations. Letting the switch go and pressing it again resumes the ramping to the next stop.
This requires the ramping-down to have a new unique command, that could be the shortclick-hold that DrJones uses in his ramping UI.

TK and Tom and Led4power and the Dr have been tinkering much longer about these things so I realise that my thoughts may be primitive and flawed, but I wanted to post it anyway

[djozz]

Coincidentally, Agro just posted the same need for a direct access to the maximum thermally stable mode.

[joechina]

With a Lamp on your side you can get distracting light distribution if you use it at armlength.

In fog or rain left and right eye see different brightness. Thiss can be annoying

Your neck musculature isn’t trained for sideload. Waering a light sideways a few hours you will fell it the next day.
Front mounted is easier.
A LED in front, battery on the back of your head is even better.

For working in narrow spaces you dang your head often sideways seldom the front. You are usually better avoiding a slam in your face.

For a dedicated headlamp you want a broad beam over 100 degree if need it for armlength to 3 meter distance. Side mounted you have a shadow.