I am working on a circuit design that must drive 4 10 segment LED bar graph displays. (40 LED's in all). There seems to be a lot of choices when it comes to LED constant current driver chips. I would like PWM dimming along with constant current 10-20mA.

One option would be to drive each 10 segment display with something like the attached datasheet item.

An FPGA/CPLD needs to control all 4 of the 10 segment displays. This appears to be very easy to do with this chip being it is a simple SPI interface.

My question: Does anyone have experience using these types of chips and which one should I use? I have also found a single chip that could drive up to 48 LED's at one time, but it appears to have R/G/B outputs, so it has more capability than I need.

Thanks for any advice!

 

I have played with the TM1638 (datasheet). 8 digits x 10 segments, every simple SPI-type interface, and cheap (about 25 cents at LCSC).

It's a simple chip, where you send the bits you want on and off. No decoding to work around to display what you want.

 

I have played with the TM1638 (datasheet). 8 digits x 10 segments, every simple SPI-type interface, and cheap (about 25 cents at LCSC).

It's a simple chip, where you send the bits you want on and off. No decoding to work around to display what you want.

Interesting, but it appears the datasheet for the part itself does not seem to exist anywhere on the internet. A search for it on Digikey or Mouser shows the part number does not seem to exist. The test board exist and the IC chip on the board has that part number:

'

I cannot find a supplier for the chip or even a datasheet for that part. The 'datasheet' link is just for the demo board.

 

ok, LCSC has a datasheet but it is in Chinese. I think I would prefer to stick to a chip that is available from most major electronics suppliers and is in English. Thanks for the idea. I do appreciate the suggestion.

 

The link I posted – TM1638 (datasheet) – strangely enough takes you to.... guess what .... a datasheet. In English. Reasonably clear and well-written.

LCSC has been a reliable supplier for me, often with "traditional" parts that aren't available anywhere else.

For me, the choice is pretty obvious – a 25 CENT part that can handle twice as many LEDs than you need, or FOUR chips at $6 EACH for the same functionality. The one-chip 25 cent solution is also in an easier-to-use package.

 

I have used the Texas parts
https://www.ti.com/power-management/led-drivers/led-display-drivers/products.html
There are some parts with analogue dimming, and some which just have a constant current output which is determined by one common resistor.
They are easy to interface, and reliable.

 

The link I posted – TM1638 (datasheet) – strangely enough takes you to.... guess what .... a datasheet. In English. Reasonably clear and well-written.

LCSC has been a reliable supplier for me, often with "traditional" parts that aren't available anywhere else.

For me, the choice is pretty obvious – a 25 CENT part that can handle twice as many LEDs than you need, or FOUR chips at $6 EACH for the same functionality. The one-chip 25 cent solution is also in an easier-to-use package.

That is bizarre, the link takes me to a paper on a PCB board that looks like this:

 

Crap. Wrong link apparently. Sorry.

Here's the data sheet.

 

I have used the Texas parts
https://www.ti.com/power-management/led-drivers/led-display-drivers/products.html
There are some parts with analogue dimming, and some which just have a constant current output which is determined by one common resistor.
They are easy to interface, and reliable.

Nice and they have a lot of choices and meets my requirement for PWM dimming. Thanks!

 

Crap. Wrong link apparently. Sorry.

Here's the data sheet.

Nice, thanks! I have it now.

 

I have used the Texas parts
https://www.ti.com/power-management/led-drivers/led-display-drivers/products.html
There are some parts with analogue dimming, and some which just have a constant current output which is determined by one common resistor.
They are easy to interface, and reliable.

After much thought, I decided to use the following chip from the series you recommend from TI:

TLC6C5712-Q1

Plan to use 4 chips instead of one of the larger more output channel chips in order to make the routing easier. I can keep each chip close to each bar graph on the PCB. I like the fact that it has a PWM input that can be applied for all output channels at once. That gives me the ability to change brightness to just about any fine tuning amount I may desire.

Though the TM1638 is definitely cheaper, I prefer to have the PWM for the dimmable capabilities. You can dim the TM1638 but only in discrete levels (16) and only by using the serial interface.

 

I have used the Texas parts
https://www.ti.com/power-management/led-drivers/led-display-drivers/products.html
There are some parts with analogue dimming, and some which just have a constant current output which is determined by one common resistor.
They are easy to interface, and reliable.

Question: TI parts have a thermal pad on the bottom of the part. I use solder reflow and a hot plate along with a stencil so that is ok. The datasheet is not exactly clear what pin (if any) the thermal pad connects to. Common sense tells me the ground pin or the PGND pin for the TLC6C5712Q. The datasheet sort of 'indicates' that from the following (assuming the gray area is a copper fill):



but it is subtle and does not simply come out and state that. But it is never good to assume anything. Do you know what pin the thermal pad connects to from past uses of this chip? Thanks.

 

Question: TI parts have a thermal pad on the bottom of the part. I use solder reflow and a hot plate along with a stencil so that is ok. The datasheet is not exactly clear what pin (if any) the thermal pad connects to. Common sense tells me the ground pin or the PGND pin for the TLC6C5712Q. The datasheet sort of 'indicates' that from the following (assuming the gray area is a copper fill):

View attachment 293015

but it is subtle and does not simply come out and state that. But it is never good to assume anything. Do you know what pin the thermal pad connects to from past uses of this chip? Thanks.

ok, never mind. I found TI's evaluation module for this chip and it's schematic makes it very clear it needs to be tied to ground:

 

I've used PowerPAD devices from TI before. I've always tied the pad to a PCB ground.
https://www.ti.com/lit/an/slma004b/slma004b.pdf

Connect all holes, including those within the thermal pad area and outside the pad area, to the internal ground plane or other internal copper plane.

 

I'm not trying to convince you of anything, but I did just have a thought about your comment of keeping the connections simple.

With a multiplexed driver like the TM1638 and many other LED drivers, the connections are actually pretty simple, especially with bargraphs.

The ten segments of all four bargraphs are connected in parallel (i.e., segment 1 of each bargraph are connected together, segment 2 all connected together, etc.) These are connected to the segment i/o of the chip. All of the pins on the other side of a bargrapgh are connected together and connected to a digit i/o. 40 segments with 14 traces. Going up to 8 bargraphs only adds 4 more.

Just something to consider.

 

I'm not trying to convince you of anything, but I did just have a thought about your comment of keeping the connections simple.

With a multiplexed driver like the TM1638 and many other LED drivers, the connections are actually pretty simple, especially with bargraphs.

The ten segments of all four bargraphs are connected in parallel (i.e., segment 1 of each bargraph are connected together, segment 2 all connected together, etc.) These are connected to the segment i/o of the chip. All of the pins on the other side of a bargrapgh are connected together and connected to a digit i/o. 40 segments with 14 traces. Going up to 8 bargraphs only adds 4 more.

Just something to consider.

In other words using the scanning capability of the chip, as shown in the application schematic and just replace each 8 segment display with a bar graph and add SEG9 and SEG10.

Yes, that would definitely reduce parts count considerably. For this chip you do loose some of the fine grain dimming capability being it only has 8 levels where the other chip I would looking at has a PWM input for controlling brightness allowing for a much more fine grain control.

Some other concerns per the TM1638 datasheet:

The segment drive current is 30mA typical. Being that we are scanning with a 25% duty cycle, that is only 7.5mA average current per LED. Being that this application sits on the dash of a car, we need a bright display for daytime use, and a much less bright display during night time use, thus high brightness is required . But actually for this chip the situation is much worse than that being that the typical grid drive current is 140mA typical, that is only 14mA per LED, again for just 25% duty cycle, now the average is down to only 3.5mA average current per LED. That might be fine for a household application that is not exposed to sunlight conditions. No doubt the scanning reduces complexity but it comes at a cost.

You did give me an idea though of possibly using scanning with a CPLD where the CPLD does all the timing, then drive each segment with a transistor in saturation with current limiting resistor for the LED along with a high side transistor for controlling each bar graph during the scan. For dimming to work you would then need an 'adjustable' power supply on the high side that again could be controlled by the CPLD. The only drawback with this idea is you need an adjustable power supply that can be programmed via a digital interface of some sort. You also end up with more board components such as transistors and resistors but you could offset that some by getting larger packages with arrays of those components.

Just my thoughts on that. I like the suggestion in that it does reduce complexity. Maybe a different chip could have the scanning ability without compromising the power output and still have the fine tune dimming ability. I may look around at some more parts.

 

Any driver chip designed for multiplexing will work the same way, but most are designed for 8 segments instead of 10, so 2 segments of each bargraph will need to be split up on another "digit" – easy enough to sort out in firmware.

A hint for LED displays – a smoke/gray lens in front of the LEDs improves contrast significantly while allowing the colors to be easily visible.

For maximum visibility in bright light/sunlight, 3M used to have a rigid (~1/16") Vikuiti "film" that's somewhat like a polarizer on a large scale. Slats in the film permit viewing on-axis, but prevent light from off-axis from making the LEDs impossible to see. I first used this to make a dim data collector 7-segment display visible in direct Dominican Republic sunlight visible. Without it, you couldn't even tell if the display was illuminated. With it, the display was clearly readable.

I went on to use it in airport control towers to make 7-segment displays readable in all light conditions. This film DOES NOT interfere with polarized sunglasses.

Sadly, this material doesn't seem to be available any more, at least not under that name. However, computer screen "privacy protectors" seem to be make of an extremely similar if not identical material.

The pictures show an example....but taking pictures of LED displays never seem to do them justice.