0

I'm trying to implement the circuit shown in the schematic below (not complete, just for reference,) where the idea is to turn on only one LED at a time -hence the ADG1406- with a frequency of about 1kHz.

The characteristic of the circuit, which I try to make evident with the color gradient, is the high resistance of the traces, where green is a low resistance (15Ω) and red high (250Ω.) Because of that, I decided to implement a current sink (TLC5947) to keep the brightness constant along the whole matrix.

Using the TLV61048 (boost converter) a voltage of 14V is defined with a maximum current of 300mA, for my application, the LEDs require a current of 10mA.

Schematic

Now, my question is.

With the implementation of the TLC5947 is the supplied voltage (14V) going to be adapted according to the conditions of each LED plus keeping the current constant following more or less this equation

$$V=R\times I_{LED}+V_{LED}$$

or in the case of implementing a current sink, must the voltage supply be different?

There must be a working principle/theory that I'm not getting.

EDIT: I'm aware that the current set resistor (R_{IREF}) is missing, the main point of the question is to clarify the behavior of the voltage across the matrix given a current sink.

Also, the high trace resistance is given by the really small cross-section of the copper.

Second approach

The difference with respect to the first approach is, as proposed in the comments, change the TLC5947 for another ADG1406 (yeah, expensive IC) and the voltage regulator was also changed by a single-channel LED driver, the TPS92612.

enter image description here

Note: there is a mistake in the schematic, the VDD of both ADG shouldn't be connected to the output of the LED driver.

Smart_Celery
  • 89
  • 8
  • 1
    You're not using the IREF pin of the TLC5947, the IREF pin needs a resistor to GND which sets the maximum LED current. The TLC5947 needs about 1 V between an OUTXX pin and GND so your supply doesn't need to be 14 V. I think around 6 V would be enough. Also the TLC5947 is quite complex, do you need the **dimming** function of the TLC5947? If not then **I** would replace the TLC5947 with a 2nd AG1406 and from its D pin connect a resistor to GND to set the current or even a simple 2-transistor current source so that the trace resistance matters less. – Bimpelrekkie Feb 05 '21 at 11:21
  • 1
    Also note that a very smart PCB layouter can make the LED matrix such that each LED will have almost the same trace resistance! – Bimpelrekkie Feb 05 '21 at 11:24
  • Thanks for your answer. I'm aware of that resistor, that's why I said the schematic was just for reference. The main point of the question is the behavior of the supplied voltage. Another ADG1406 plus a resistance in the D port IMO wouldn't define a constant current along all LEDs since, as I also said, there is a big difference in trace resistance between one LED and another. – Smart_Celery Feb 05 '21 at 11:27
  • with respect to the dimming function, it's true, I don't need it. However, can that function be deactivated by defining the duty-cycle between 0 / 50% / 100%? An alternative to the TLC5947 might be the TLC5923. – Smart_Celery Feb 05 '21 at 11:34
  • 2
    @Smart_Celery: The point of using a constant current source is that you don't have to concern yourself (much) with the voltage. 14V is more than enough for the LEDs. The current regulation sees to it that the LED only gets the amount of current you set. The voltage across the LED is a result of the current, not its cause. – JRE Feb 05 '21 at 12:09
  • 1
    The one thing you do need to consider is the waste heat produced by the constant current supply. To reduce the heat, you can start with a lower voltage. – JRE Feb 05 '21 at 12:10
  • 1
    The adg1406 would not be my first choice for this application. Why? It is probably not cheap esp when compared with jellybean mosfets. It may not appreciate the 80mA through it as well. – Kartman Feb 05 '21 at 12:10
  • 1
    *wouldn't define a constant current along all LEDs* You're over-focusing on that constant current. If you would make the design such that there is a 10% current difference between the LED with the lowest and highest current (like 10 mA and 11 mA) I bet that **no one will notice**. Do some measurements on two LEDs, how much current difference can be allowed before your eyes start to notice? WOW that ADG1406 is € 10 I agree with @Kartman that with some "jellybean" components considerable cost savings can be made. – Bimpelrekkie Feb 05 '21 at 12:23
  • @Kartman I agree that the IC is not cheap, but jellybean components can beat the 5x5 size, bear in mind that the largest array is 16x16. Something I didn't completely get, where did you get the 80mA? – Smart_Celery Feb 05 '21 at 12:45
  • @JRE that is exactly my point, with a current sink, the voltage wouldn't be "that" important. However, my question is still the relationship between the 14V that the TLV61048 supplies and how it changes by the condition "imposed" by the TLC5947 and the trace resistance. I'm sure there is something I still don't understand with current-sinking and sourcing. – Smart_Celery Feb 05 '21 at 12:48
  • @Bimpelrekkie But what is the problem of overfocusing on maintaining a constant current if there are the tools to do it. Also, as I mentioned, a difference of 200Ω I don't consider to be that low. Also, considering your first proposal for a second ADG1406. What if the D terminal of the 2nd is connected to GND and the D terminal of the first one is connected to a current source? I'm going to make a schematic and then upload it here as a proposal. – Smart_Celery Feb 05 '21 at 12:53
  • 1
    @Smart_Celery - if each led can draw 10mA and there’s 8 per row, then 80mA. Have you looked at hub75 modules? Cheaper and easier than building your own. Or strings of ws2812 leds. – Kartman Feb 05 '21 at 13:18
  • @Kartman yup I see what you mean, but what I want is to turn on and off just one LED at a time, and switch between LEDs with a frequency of 100kHz. With respect to the other LEDs options, as I said, the array is very particular, and it's already done, fixed, I cant work with it or changed it, I just need to control it. and it has these particular limitations. Maybe the idea of two ADG is not bad at all. Actually, right now I'm looking for a current source. – Smart_Celery Feb 05 '21 at 13:26
  • *But what is the problem of overfocusing on maintaining a constant current if there are the tools to do it.* Having the tools is not a good argument, over focusing will result in a design that is over **complex** and complex things have complex problems. Take it from any experienced engineer that the simplest solutions are often the best solutions. So "keeping the current constant" is a goal but the question is **how constant**. 10.00 mA +/- 0.01% is overkill (I hope you agree), is 10 mA +/- 10% still OK ? Because, if it is, then we can use a **very** simple circuit. – Bimpelrekkie Feb 05 '21 at 13:41
  • Current source, here's an example I would try: https://electronics.stackexchange.com/questions/360175/leftover-voltage-when-using-constant-current – Bimpelrekkie Feb 05 '21 at 13:42
  • @Smart_Celery: With the constant current source driving your LEDs, the resistance of the traces doesn't matter (unless you have traces of several hundred ohms.) I'm having trouble imagining that the trace resistance will vary enough to make a visible difference. Also, keep in mind that the output channels of the TLC5947 can vary from one to another by up to 2%. – JRE Feb 05 '21 at 14:01
  • @Bimpelrekkie I was studying the circuit explained by jonk and it is indeed a very good option (I even made a simulation with a 4x4 matrix). However, the components needed for a 16x16 matrix are too many and the area starts to be too big (my limit is about the size of an arduino nano). The positive thing now is that I understand the theory and yes, I agree with what you say (0.01% is overkill), but I have yet to define a circuit that can do the job efficiently. – Smart_Celery Feb 05 '21 at 17:24

0 Answers0