Attempt to design an alternative I5R

1.Purpose

As we saw in the article on the observation of an I5R, Olight made certain choices to evolve the i5T in I5R. The purpose of this article is to think yourself as an Odesigner and try to make different choices.

2.Specifications

 Regarding the mechanics, we will start with the same choice and keep exactly the same. The only thing to modify is (if it is possible):

Regarding the batteries, it must accept:

• Alkaline battery AA
• NimH AA
• Lithium AA
• 14500 3.6V Lithium battery

For the light power, we will try to keet it equivalent for AA and go to higher for 14500.

3.Analysis of the specifications

 As we saw in the observation of the i5R, this will force us to hace a so-called Buck-Boost voltage conversion topology, The bulk of the work be at this level because the available space is low.

Then Olight always delivers a battery with lamps, thus we will have to find a good 14500. As we are ambitious and we are going to try to get a lot of lumens, we will choose a model with very low internal resistance and a relatively high capacity.

As we want to output high power the temperature may be our main problem, indeed the user should not burn himself.

We will start with the battery, the the electronics ans its integration and  the thermal will be measured which will allow things to be corrected if it is too hot.

4.Battery

 For the battery, we will choose a Vapcell INR14500 H10, It seems to be a very good candidate and in addition, I have this model.

Its internal resistance is about 0.07Ohm and the capacity is 1000mAh, copared with the Baton 3 16340, it is more efficient. This is rather a good omen because the Baton 3 can draw up to 1200 lumens.

It can delivers 10A continuously and much more over shorter times.

But be careful with this battery fully charged in a normal I5T, the electronics of the lamp are destroyed on the fisrt switch.

5.Electronics

 If you haven't already read it, I recommended you the following 2 articles

• I5R Observation (French)
  
• How a flashlight works, case of the Olight Baton II

So as we have seen:

• Space available in an I5T is extremely low.
• Schematic of a Buck-Boost requires more space than a Boost (I5T) and I have not found integrated circuit which fulfills this function at very low voltage.
• In the S1R Baton II, Olight used a Boost followed by a linear regulator which takes up little space but which can lead to significant losses.

5.1.Electronics choice

As I can hardly see myself starting from zero for a discrete solution (no control integrated circuit + switching MOSFET) in such a small volume, we will go for a similar choice to that of Olight S1R Baton II. The advantage is to have only one "big" transistor to add and some slight electronic modifications. But this transistor will be hot ant it will have to be thermally transferred.

Regarding the light power we will also base ourselves on the baton II, whose maximum LED current is 3A, and start at 2A, given the difference in the volume of material between I5T et S1R to arrive at an equivalent heating.

5.2.Modifications compared to I5T

I5T Mode are:

We will quickly see that we will have to change that. Indeed the maximum power of the I5T is the same regardless of the battery voltage. We want to try to reach 2A which correspond to a LED voltage of 3,2V which gives 6,4W. However at 0,9V input voltage a Boost has a maximum efficiency of 80% which leads to a current of 8,9A (which is totally incompatible with standard AA an integrated Boost circuit of I5T).

It will therefore be necessary to modify the actual schematic so that the maximum current depends on the voltage of the battery. And we will therefore keep the decrease over time and the gradual changes (i.e. 100% for 3min, then 50% for 25min, then 10%) with each of the 3 progressive transitions. For LOW mode, 15 lumens correspond to 4%.

5.2.1.Current variation in the LED as a variation of its voltage

In principle, the micro-controller outputs a PWM signal which is averaged by a low-pass filter. This average is used as a reference for the error amplifier which compares it to the voltage of the shunt (image of the current in the LED). And so the output of the error amplifier attacks the driving of the boost. The electrical schematics of the I5T is therefore as follows:

The micro-controller being powered by the Boost output via a 3V LDO, its high voltage state voltage varies slightly between 2,65V and 3V but it is far from being enough for our modification.

Here is the schematic on which we will start for our alternative design of the I5R:

Adding R4, R7, R11 and CR1 gives us the following theoretical curves:

On the x-axis we have the battery voltage. Compare the yellow curve (current I5T 100%) and the blue curve (I5R alternative 100%). We see that the curves intersect towards 1,25V in fact I chose this voltage because I wanted equivalent result with NiMH batteries, which means that we will get almost 800mA with a 1,5V battery (which is equivalent to 350 lumen).
As for the temporal, we should have the blue curve for "min, then the red curve for 25min then the green curve if the battery allows it.

The curve for the LOW mode is as follows:

5.2.2. Shunt resistor

As the schematic shows us, we are going to stay on the same value which will allow the heat to de distributed and the MOS Q1 to be a little heated less. However 2A in a resistor of 0,05Ohm gives 0,2W, it is a bit to much for a 0805, thus we are going to switch to a 1206 (bigger size).

5.2.3.Inductor

The L1 inductor is already a very good inductor and should drawn 2A current with no problem (I measured its internal resistance at 0,011Ohm). However the height of this component is 3,3mm, and we need an inductor of 2,3mm max to manage the thermal of Q1. We will therefore start with a model form the manufacturer BOURNS, an  SRP4018FA-R68M which has a heigh of 1,8mm +/- 0,2mm.

5.2.4.MOSFET Q1

I won't go into the explanations too much, but we needed a MOS with a VGS threshold max of 1V, the lowest possible RDSon, a small case with good thermal resistance (see bellow). My choice fell on the SISS61DN from VISHAY.

This component is the most critical, because it is it that will take most of the voltage between the LED and the battery with a 14500. A numerical application will be more meaningful, when the battery is fully charged, its voltage is 4,2V and the LED is 3V for 2A (most critical case). The electronics will therefore manage to present a resistance of 1,2V / 2A (i.e. 0,6 Ohm). These 0,6Ohm are distributed between inductor (0,0082Ohm), IC1 (0,058Ohm), Q1 (the remainder) and the shunt (0,05Ohm). So that is 0,4838Ohm for the Q1 so almost 2W of losses.

In a classic I5T, the LED in which there is about 1,82W is mounted on IMS (Insulated metallic substrate) and not PCB. So if we do not want this MOS to break with high temperature, we will have to transfer it thermally.

5.2.5.PCB with modification

 New component with red Marking

5.3.Measure after modification

 

In yellow the battery voltage and in green the current in the LED. 

We see:

• a plateau at the level of the current towards the beginning, which is due to an approximation in my theoretical calculations. To remedy this it would be necessary to decrease the voltage across R4, however I think 1.85A is sufficient for this test.
• the green line which suddenly thickens towards 3V is when we go from Buck to Boost.
• that it is rather in accordance with the theory

On the I5R OLIGHT announces 350lumens with a current of 800mA. 

If we look at the datasheet of the LED going from 800A to 1.85A we should have a flux of about 700lumens to 1.85A.

6.Assembly 

We saw on the I5R that to manage the thermal the LED was mounted on an IMS then transferred using thermal grease on an aluminum part which is inserted by force into the body of the I5R.

The principle for our test is to be able to put an SMI on each side of the aluminum part, one for the LED and the other for the transistor Q1. The other advantage of this solution is to increase the thickness of the aluminum part by 30% and therefore this will make it possible to reduce the temperature gradient between the body of the I5T and the elements which heat up. The drawing below shows us this assembly.

It looks simple on paper but seen the actual size it looks like a very tight part of TETRIS😀.

A photo of the IMS with the MOS Q1, some small components and the 3 connections to the modified PCB of the I5T (sorry for the quality).

The change of the inductor will allow us to place the PCB with the electronics in its initially planned place. To prevent the inductor, the Boost integrated circuit and the shunt resistor from getting too hot in HIGH mode, I add a thermal Gappad to their level. In the photo below we do not see the difference with a classic I5T.

All that remains is to insert it by force into the body of an I5T and off to testing😎.

7.Thermal

As a reminder:

This is the weak point of this test, because I cannot modify the times of the operating modes. 

The images below are taken with the thermal imager and the time at which they are taken is indicated below.

If we look at the 3min and 3min30 pictures we see the impact of the change from 100% to 50%. Indeed the temperature of the head decreased while the temperature of the body continued to rise.

It is the same thing between 20min and 25min with the change from 50% to 10%. 

The temperature rises to 55 ° C (burning sensation) however this temperature is maximized because if the lflashlight had been held in one hand there would have been a heat exchange and we would have climbed lower. 

After 45min of HIGH mode, the battery voltage is 3.33V which means that there is approximately 19% battery charge remaining.

8.Rendering

In this photo we have an I5R in HIGH mode and our modified I5T in HIGH mode.

The 2 flashlights are placed 40cm from the screen.

9.Conclusion

This attempt is rather conclusive, changing the time to 100% in HIGH mode would be fine. However as expected, the integration is complicated with a min clearance of 0.2mm. It is really not easy to industrialize for me I had a hard time making this prototype. So I understand why Olight did not go for this choice.