Train Describes using 4 digit 7 segment displays?

[SHMD]

In the input diagram you missed 2 junctions, on the left, Sin1. Also, I'd add the 5V power to the connector, if only for convenience.

 

Are you referring to the last full (black) schematic that I posted in post 55?

If so then it is just an effect of pixilation, screen resolution (yours and/or mine), and the zoom level I used to screen grab (I have no pdf printing ability at home at the mo - win 7).

 

If I hover the mouse pointer, over the Net in question, then the software will highlight the Net (and all connections) as a quick test.

(Windows helpfully hides the pointer when screen grabbing!)

 

If I click on the Net "sin1", in the Design Manager section, then it also highlights the Net in question whilst giving a list of pins connected to that Net.

 

There used to be a BS Standard, for drawing circuit diagrams, that stipulated that the joining of two wires could only be by means of a "T" junction. You could not have two wires crossing "+" and then a Junction being plonked at the intersection. The reason (then) was that poor photo copies would add/lose these junctions randomly.

This is why I have 'staggered' the pull-up/pull-down resisters R28/R26.

 

 

With regards to having 5v near the Sin1 and Sin2 inputs - I agree entirely, which is why there is one on the finished PCB. (The inputs and outputs 'share' the 0v and 5v between them but within the same block of 7 screw terminals.)

 

 

Kev.

 

Edit to add that I have just spotted the two missing Junctions!

 

They are there but the CAD has put them on top of each other. If I move the net then the become visible.

 

 

I will amend the schematic - as I like "junction dots" - but in theory, they are not needed.

 

Edited May 1, 2017 by SHMD

[SHMD]

Novarm's "DipTrace" also automagically adds junction dots when wire traces are joined together on screen.

It's when you move the components later, to tidy up the layout, that junction dots can get left behind/doubled-up.

 

 

Now for the plug. (Sorry!)

 

No connection except 20+ years of PCB CAD use of which the last 7 (amateur) years have been with DipTrace.

I like this CAD package a lot and you can use it free under two main proviso's :-

 

1/ You are limited to 300 pins per design. (You can increase this to 500 pins if you can justify it - I have   )

2/ All your work is produced on a "not for profit" basis.

 

Otherwise you need to purchase a license.

(For example, AND1 has 143 pins.)

 

 

Kev.

[raymw]

I used "Eagle", when it was developed/owned by a company in Germany, iirc. It's now owned by Autodesk. They had various licensing agreements, including free, not for profit and student versions. I found the original icons unclear, mentioned it to them, they said 'can you do better', so I sent my designs to them, and they changed theirs to mine. Not sure what it looks like now.  The autorouting was pretty good, but I would move their tracks around to suit my more crudely made pcbs. I never liked tracks running between dil Pads, and it paid me to keep most tracks on one side, with just a few links on the other side of a pcb.

[SHMD]

After the schematic, it's PCB time!

 

It is good to start with some specifications for the board. For example :-

 

The displays must be in-line!!!

All connectors are to be at the boards edges, (Auto-place routines do not implicitly 'know' to do this, but they are getting better.)

As many component designators to be visible as possible,

Board connections functions to be displayed ON the PCB,

Each 'block' of screw terminals must include 0 and 5 volts.

No surface mount components!

 

Good house keeping :-

Minimum board size,

Minimum length of total traces,

Maximum amount of 'copper pour',

Minimum number of Vias,

Pad sizes to be optimised for hand (Antex) soldering, (Don't just leave it to the defaults!)

Hole sizes optimised for the correct components lead sizes.

 

Here's the finished design :-

(That was easy wasn't it!)

 

 

Here's a little more detail...

 

This view is of the Top Copper layer.

It really demonstrates the advantage of 'multiplexing' the display digits together.

Just connect one display to the uC and then 'bus' the rest along to the other 5. Simples? Well, not quite but still a massive saving in time, effort, connections and uC I/O.

 

Whilst the Top Copper predominately has its' traces travelling horizontal, the Bottom Copper layer has its' traces travelling vertically.

It doesn't really matter which layer does which but the effect is to dramatically reduce trace length, complexity and board space.

(Of course, simplifying the hardware is paid for by an increase in software complexity.)

 

I spend 'ages' optimising ALL the traces on a PCB.

I do use the auto-route function but only as guide to test if the components positions, within the PCB's chosen size, is actually possible.

I then erase the auto-routed traces, move the components, shrink the PCB size again, and repeat until I am happy.

 

Again, put as much info as you can on the silk screen - it will save you rooting around for schematic diagrams in the future so many times!

And, at this stage, it's free. Put some thought into how/which orientation they will need to be in the final product.

Note that with R7 to R23 I have put the segment name that they connect to to aid any diagnostics in the future.

I have also done this for the Darlington Driver outputs too.

Again, function descriptions are a MUST for all user connections.

 

 

Looking (and spinning around) the PCB all throughout the PCB optimisation process is also a good idea.

 

The Boards vital statistics :-

 

5.5" width by 2.4" high

60 Patterns

67 Nets

150.21" total length of traces

290 pins

29 vias

319 total number of holes in 5 different sizes.

 

 

Kev.

 

[SHMD]

Kicad is Open Source and has no limitations on size.

http://kicad-pcb.org

 

This might be of interest to you, perhaps:

http://www.rocketscream.com/blog/2016/02/19/from-diptrace-to-kicad/

 

HTH!

 

I had a look at this and the examples look very good.

I note that this is 'open source' and the examples given are also open source.

Other 'commercial' CAD packages are aimed at industry and their users are much more protective of their designs and thus there is not the same number of examples to inspect.

 

But, if I was starting out now, I would probably choose kicad.

Unfortunately, all CAD packages require a large amount of user effort investing before you know whether it is good or not.

 

One area not often "checked out", when choosing a CAD package, is the management/creation of components and libraries.

This is my least liked aspect of PCB CAD and there is definitely a wide range of usefulness in the library tools supplied with CAD packages.

Each part requires 2, (or 3!), CAD patterns making. One for the schematic and one for the PCB. Now there is a third model required too - for the 3D representation of the part.

 

Novarm's DipTrace had very large comprehensive libraries that ARE being maintained and added to.

But the 3D models are less so well represented and they do not have an editor for 3D models either.

The worst aspect of DipTrace is the library search function - but it is improving.

 

 

Kev.

[Brian]

<Snip>

Using thumbwheel switches for the digit inputs. (What does an NX Panel look like?)

 

Like this one perhaps?  https://www.youtube.com/watch?v=76fAHAxv0Dw

[SHMD]

Like this one perhaps?  https://www.youtube.com/watch?v=76fAHAxv0Dw

 

Not really no, but the requirement was to 'easily' display 7segment 4 digit "nx" style user selectable displays with little electronics experience.

(I just wish my end-to-end was the same physical dimensions as the panel in the you-tube video!)

 

 

Regards,

Kev.

[MarshLane]

Now for the main event - the Micro Controller (uC).

 

The PIC18F4550 is a 40 pin device with 32k Bytes of program space, 2k Bytes of RAM and 256 Bytes of EEPROM (Memory that's remembered even when the power is lost).

Plus it has a whole raft of in-built peripherals - 35 I/O pins, Serial Comms (USB, (a)synchronous, I2C, SPI), 4 timers, 13 * 10bit ADC inputs, lots of RESET choices, lots of CLOCK modes, In-Circuit-Serial-Programming, Comparators, powerful (flexible/self contained) PWM, loads of Interrupt sources, etc etc...

 

Kev.

 

 

Hi Kev,

Sorry to hijack the thread, but I've always had an interest in electronics, and am hoping to expand my knowledge a little bit to have a go at a few different things over the coming months.  But I'm really starting at the bottom with all this, although I understand the principals of resisters, relays, capacitors etc.. But I've been looking at experimenting with PICAXE processors - I note on this project you've used a Microchip PIC processor.  Is there any major difference between the two types or is it personal preference?  Is one easier to code than the other - I've done Basic, ASP, PHP and a little bit of C programming in the past.  Do you find that one type offers more options (i.e. I/o pins, timers, interrupts etc) than the other?

 

Also, I'm looking at creating some circuits (I presume you can get a test/experimental breadboard or something for PIC processors?) and then having some PCBs made (once I've proven the circuit) the software you've mentioned that you use on here, would you recommend that?

 

Fascinated with what you've created with the six character display - really interesting, and thanks for posting all the schematics and background reference.  Don't totally understand it all, but its all part of learning, as I want to be able to understand what I'm doing!

 

Cheers

Rich

[SHMD]

Hi Kev,

Sorry to hijack the thread, but I've always had an interest in electronics, and am hoping to expand my knowledge a little bit to have a go at a few different things over the coming months.  But I'm really starting at the bottom with all this, although I understand the principals of resisters, relays, capacitors etc.. But I've been looking at experimenting with PICAXE processors - I note on this project you've used a Microchip PIC processor.  Is there any major difference between the two types or is it personal preference?  Is one easier to code than the other - I've done Basic, ASP, PHP and a little bit of C programming in the past.  Do you find that one type offers more options (i.e. I/o pins, timers, interrupts etc) than the other?

 

Also, I'm looking at creating some circuits (I presume you can get a test/experimental breadboard or something for PIC processors?) and then having some PCBs made (once I've proven the circuit) the software you've mentioned that you use on here, would you recommend that?

 

Fascinated with what you've created with the six character display - really interesting, and thanks for posting all the schematics and background reference.  Don't totally understand it all, but its all part of learning, as I want to be able to understand what I'm doing!

 

Cheers

Rich

 

Hi Rich, thanks for the compliments regarding AND1 - much appreciated.

AND1 is proving to be quite a useful gadget! (But more on that later..)

 

The PICAXE is based on (a small) subset of PIC processors and their software (interpreter and IDE) are tailored to these chips in particular.

There are advantages and disadvantages to running the PIC via ASSEMBLY, COMPILED or INTERPRETED code but to begin with I feel that the most important thing is to be familiar with the hardware - no matter which language/platform you choose.

 

You NEED to be comfortable with (and be able) to make changes to your design AS YOU LEARN.

This is an iterative process so editing code, downloading it AND seeing the instant results is the most important thing at this stage.

Get an LED to flash. Then make it flash to a certain pattern. Then make it flash randomly!

To do this for the first time will take all evening BUT by the end of the night you will be able to make changes and gain confidence to achieve more next time.

 

NEVER bite off more than you can chew.

All development is done in little chunks and verified BEFORE moving onto the next line of code.

Remember - when something doesn't work as planned, that is when you will start to learn!

 

I will add more tomorrow but yes, the PICAXE is a good platform to learn and make things AS it is well supported.

(I have never actually used a PICAXE though, but will outline the advantages/disadvantages of the various choices available soon. There is rarely a simple yes/no answer in electronics!)

 

  

..regarding the thread hi-jacking, don't worry - it's not my thread!

 

 

Kev.

[MarshLane]

Thanks Kev,

Look forward to your next post, but that is certainly food for thought. I'll msg Andy Y and ask him if he can move my question and your answer to a new thread on this board, as it might be useful for others as well.

 

Rich

[MarshLane]

Hi Kev,

Just wondered if you'd had chance to put the next part that you mentioned together?

Rich

[bescotbeast]

Hi Kev,

Just wondered if you'd had chance to put the next part that you mentioned together?

Rich

I'm wondering if we have any pictures of the completed project to view, I'd love a Train describer for my layout but haven't the foggiest on what to do.

[BG John]

I was reading up on what to do with the MAX7219 in my Arduino starter kit yesterday, and it all looked so simple that I wished I could find a use for it. Now I'm not so sure .

[SHMD]

Sorry gents, i'll put something on here tonight.

 

I think I have the pictures with me and I have nothing else to do tonight as I am in Oslo this week and the beer is eye wateringly expensive - £10 a (short measure) pint!!!!

 

 

Kev.

[SHMD]

The next stage in the "AND1" PCB build process is to get it manufactured.

 

As I have previously said, the finished CAD files will have to be sent to the manufacturer in a form that the manufacturer can open / manipulate / download to their CAM machines.

Most PCB manufacturers will take the more popular (usually expensive) CAD files directly now but, traditionally, the CAD programs used to printout the "artwork" and this was sent to mask the photo-resist PCBs. Later, "Gerber" files were used as the standard (instead of physical artwork) to etch/mask/screen each layer of the PCB.

 

Gerber files are still accepted everywhere.

I would suggest that you get a "Gerber Viewer" program (off the NET but can be part of the PCB CAD program) so that you can inspect each layer. You will be amazed at how much tidying up/mistakes you will spot doing this! It will also give you confidence in your finished design.

Always include "crop marks" and the layer name - outside the PCB dimensions. The manufacturer will remove these whilst obeying them.

 

As well as the Gerber files you will need to have the "NC Drill" file.

Again, make sure the holes are the right size for the components!!!

 

 

Next, choose who you want to make your PCBs.

Your choice should take into account the following:- Price, Quality, Delivery, Lead time, and if you can - Recommendation.

I use quick-teck as I am familiar with them and, for a batch of PCBs (20 to 30), they are not bad on price.

I would recommend that you get at least 5 PCBs even if you are convinced (now) that you only need 1.

Also, be prepared to learn from your mistakes. That means this first attempt may not work(!!!) and you need to re-work your artwork.

 

Here is a typical on-line quote page :-

Just enter the dimensions and the quantity you want and leave the other settings as the defaults.

Then look at the price - £42.12 / PCB!!!!

Now adjust the quantity...

 

Here is the quotes for the following quantities :-

2 off for £84.23 or £42.12p/PCB

5 off for £86.81 or £17.36p/PCB

10 off for £91.11 or £9.11p/PCB

20 off for £102.04 or £5.10p/PCB

30 off for £114.37 or £3.81p/PCB

 

Notice how the price/PCB comes down dramatically whilst the overall cost has only increased by £30.

Batch manufacturing - even if you only want one.

(I think you would be mad not to pay the extra 6 quid to change the quantity from 2 to 10!)

 

But now, on-line, there are quite a few manufacturers offering small "prototype" quantities for under $50.

So shop around and let us know your findings, value, quality and the service you received.

 

 

Here's what Quick-teck delivered...

 

 

 

 

Kev.

 

[SHMD]

Soldering the AND1 boards...

 

There is an excellent thread on soldering. It is called the "Dread of soldering".

This thread has all the info you need to solder well but I do think it should really be renamed to something more apt like "How to get good results soldering"!

 

I use a cheap Antex 25W soldering iron.

I wipe the tip regularly on a damp sponge and then 'wet' the tip with a little bit of the multi-cored 60/40 lead/tin solder. I then hold the tip to the components leg, (and NOT the PCB pad), where the tiny amount of solder aids heat transfer to the component only. I then apply the required amount of solder to the other side of the leg (to the iron's tip) and let it melt, run, and flow into the hole until it pops all round the pad. I then remove the tip and wipe it again ready to repeat the procedure.

Never try to solder by applying the required solder to the tip and then trying to transfer it to the component. I always have the 'flat' of my iron tip to the component's leg and NOT the PCB's pad.

 

Here if a close-up of an new PCB...

Note that:-

the holes are not exactly in the centre of the pads/vias,

or that the solder resist is exactly centred on the pads,

the solder resist is bigger than the pads,

the pads connected directly to the solid copper ground plane are connected by spokes that allow for easier soldering,

the vias are covered (in this case) fully with solder resist,

and that the tracks between the pads have plenty of space and are fully covered by the very effective solder resist.

 

And here is a close-up of my soldering...

(So, so cruel!)

Note that I have not 'cleaned' the excess flux away.

Also, of note is the "Solder Mask" enlargement. The "Solder Mask" enlargement is the increase in size around pads before the (green in this case) solder resist starts. The solder mask (or solder resist) is not usually placed exactly in registration on the PCB so should be made a little bit bigger than the pads. I use 8 thou for this. 

 

Here's two completed examples.

(I think I have made 6 or 7 so far.)

 

This shows "Water." but through a red filter (but still with protective plastic covers both sides).

 

.. but, (after much research), you can use Xmas chocolate wrappers as a cheap and effective alternative!

 

 

Kev.

[BG John]

Aliexpress is a good source for various technical stuff, including reflow ovens and electronic components. (and lots more!)  My experience with them so far is good, only had to claim a refund twice in over 5(-ish) years and on double-digit numbers of transactions as the order wasn't forthcoming. Best to use a credit card.

 

HTH!

Doh! I've just paid £1.99 for five 74HC595 from a UK supplier, which I thought was OK as it included postage. The first item on a search for "electronic components" on AliExpress was 10 for 51p, with free postage. That's 5p less than posting a second class letter here, without the price of the contents! I've used AliExpress twice so far, but waiting several weeks for delivery isn't helpful when you're in the middle of making something you want bits for. I must plan ahead!

[MarshLane]

Soldering the AND1 boards...

 

There is an excellent thread on soldering. It is called the "Dread of soldering".

This thread has all the info you need to solder well but I do think it should really be renamed to something more apt like "How to get good results soldering"!

 

I use a cheap Antex 25W soldering iron.

I wipe the tip regularly on a damp sponge and then 'wet' the tip with a little bit of the multi-cored 60/40 lead/tin solder. I then hold the tip to the components leg, (and NOT the PCB pad), where the tiny amount of solder aids heat transfer to the component only. I then apply the required amount of solder to the other side of the leg (to the iron's tip) and let it melt, run, and flow into the hole until it pops all round the pad. I then remove the tip and wipe it again ready to repeat the procedure.

Never try to solder by applying the required solder to the tip and then trying to transfer it to the component. I always have the 'flat' of my iron tip to the component's leg and NOT the PCB's pad.

 

Here if a close-up of an new PCB...

PCBx400a.png

Note that:-

the holes are not exactly in the centre of the pads/vias,

or that the solder resist is exactly centred on the pads,

the solder resist is bigger than the pads,

the pads connected directly to the solid copper ground plane are connected by spokes that allow for easier soldering,

the vias are covered (in this case) fully with solder resist,

and that the tracks between the pads have plenty of space and are fully covered by the very effective solder resist.

 

And here is a close-up of my soldering...

(So, so cruel!)

PCBx400b.png

Note that I have not 'cleaned' the excess flux away.

Also, of note is the "Solder Mask" enlargement. The "Solder Mask" enlargement is the increase in size around pads before the (green in this case) solder resist starts. The solder mask (or solder resist) is not usually placed exactly in registration on the PCB so should be made a little bit bigger than the pads. I use 8 thou for this. 

 

Here's two completed examples.

20170606_212448a.jpg

(I think I have made 6 or 7 so far.)

 

This shows "Water." but through a red filter (but still with protective plastic covers both sides).

20170606_212724a.jpg

 

.. but, (after much research), you can use Xmas chocolate wrappers as a cheap and effective alternative!

20170606_212742a.jpg

 

 

Kev.

 

 

Hi Kev,

Im going to print all this out and go through it again!  I'm hoping I can work it all out to understand what is going on, and convert the display to four characters, and include the characters J, L and O.  Im also hoping I can come up with a way of create a mini network, so the PIC's talk to the next one down the line, and on a given input can pass the headcode to the next PIC to display.  Could be quite interesting to achieve!  Still, first task is to understand the code and schematics (and more importantly understand why each component is there) then get one working, then a second, and finally see if I can get the two to talk!

 

I may be back with question Kev!

 

Rich

[MarshLane]

Now to make the digits "light-up".

 

There is a resister each for every LED segment and YES you do need one / LED!

Each Display's common is then 'controllably' connected to 0v to switch that particular digit on. Because there are 17 segments going into 1 wire, the switching device must be capable of switching all that current on and off safely. I chose the Industry standard ULN2803A Darlington Drive IC which is eight channels and can switch 0.5A on/off per channel. It also has the input 'base' resisters built in thus saving PCB space, money and soldering time.

AND1_LEDpower.png

 

The Pic can sink or source 25mA per digital output pin. This is a lot.

Nominally, the output swing is from 0 to 5 volts but, in reality, this will be a little bit less.

 

I have six displays.

Therefore there are two Darlington Driver outputs spare. I have brought these out to the boards edge as general purpose output control lines capable of driving anything from an LED to a relay. (Snubbing options and pull-up resisters are also included in the design to aid flexibility.)

 

 

Kev.

 

Hi Kev,

As I suspected, I'm here with the questions!! A quiet hour at work has allowed me to sneak off to the coffee machine with my printouts of this thread and read!

 

Firstly, I'm following your diagram initially.  Can you confirm my understanding of the Darlington Driver is right.  In essence (sorry if I get the terminology wrong) this is eight independent darlington pairs?  A darlington pair being two transistors that are amplifying the current going in, not once but twice, to give a greater power at the output?

 

Im assuming that the light blue and thick dark blue lines on the above diagram are 'thick' connections for the PCB?  In which case I have two things that I don't understand.  Firstly why the resisters are there (R7-R23) and what purpose they are serving?  From what I can see (and I'm obviously missing something) you have marked each one as corresponding to a specific pin on the displays, but they all appear to connect to the same track linking the resistors and Darlington Driver to the displays?  Hence how can they relate to a specific pin?

 

Am I correct in understanding that while the displays are connected together, the individual displays are lit for a 1/1000th (or some other value) of a second, and the code moves to light the next segment, and by rotating round (multiplexing) there is no visible 'off' period to the human eye?

 

My plan is to try and replicate this but with a four character display using 5x7 LED dot-matrix segments.  I appreciate these are a little bigger than the alpha-numeric ones you were using, but I think they should allow all characters, and be closer to the type used as TD berth describers in signalling centres.  Im thinking of these http://uk.farnell.com/kingbright/tc12-11cgkwa/dot-matrix-1-2-cmn-cathode-green/dp/2080073 Can you see any problems with doing that? There appears to be 35 dots on the display, but only 14 pins, so I haven't yet worked out how specific characters are displayed, and the data sheet doesn't appear to help, so a bit more digging is required there!

 

Being only four characters, Im assuming a future expansion could be to get the Darlington driver to control two separate displays, each using four channels?

 

Rich

[MarshLane]

The key is in the name: dot-matrix. These led's are arranged in a matrix, meaning there's a number of rows and columns with a led between them. The exact layout is also in the name: 5x7, meaning there's 5 columns (vertical) and 7 rows (horizontal). Together that's just 12 pins, no idea what the other 2 do w/o consulting the data sheet for it

 

Hi DM,

All joking aside, thats what I thought, but I couldn't get my head around how 12 pins could individually set 35 LEDs! Or is this another multiplexing scenario, where each setting is one LED and the software, rapidly turns them on and off so that the change is invisible to the human eye?

 

Rich

[raymw]

The diagram referenced by MarshLane is not a 3 by 7dot matrix display. The thick line is simply a short hand way of showing more or less a parallel bus. Generally it would be annotated and linked to the same bus on another part of the circuit diagram. The resistors are to limit the current to each segment of the alpha/numeric display. hth

[MarshLane]

Yes, it's indeed a multiplexing job. The software periodically turns on each row, then for each led on that row determines if the corresponding column needs to be high (off) or low (on). That's indeed happening in the kHz range so the human eye cannot see the flicker of it. Digital matrices on buses and platforms work the same way, and if you take a photo with a digital camera you can only see part of the word(s) as the camera can 'see' the flickering due to its construction and speed.

 

Hi DM,

Ah thanks - yes that makes more sense and I can understand that concept more.

 

The diagram referenced by MarshLane is not a 3 by 7dot matrix display. The thick line is simply a short hand way of showing more or less a parallel bus. Generally it would be annotated and linked to the same bus on another part of the circuit diagram. The resistors are to limit the current to each segment of the alpha/numeric display. hth

 

 

Hi Ray,

Thanks, yes I appreciate that the original circuit from Kev was the 7 or 14 segment displays that are a bit different.  When you say a parallel bus, do I understand that correctly as meaning there is, in reality, a separate track from each resistor to the segment pin on Kev's diagram, just for ease of display, it is shown as one thicker bus?  If so, that makes more sense and I can understand how the resisters are doing their job.

 

Rich

[MarshLane]

Hi Ray,

Thanks, yes I appreciate that the original circuit from Kev was the 7 or 14 segment displays that are a bit different.  When you say a parallel bus, do I understand that correctly as meaning there is, in reality, a separate track from each resistor to the segment pin on Kev's diagram, just for ease of display, it is shown as one thicker bus?  If so, that makes more sense and I can understand how the resisters are doing their job.

 

Rich

 

 

Ray,

Sorry, I've just looked at the PCB layout and realised I've answered my own question!  Yes they are separate tracks.

 

Is there some information on the data sheet somewhere that tells you the resister value, or the maximum voltage that should be sent to the 14 pins on the dot-matrix display?

 

Rich

[raymw]

Hi Rich,

 

I'm confused as to what display you are looking at. The data sheet you link to is for a 5 by 7 dot matrix display, not the segment alpha numeric display in the circuit diagram before. Are you re-designing Kev's complete circuit? There is listed on the data sheets working values for current/voltage and maximum values, too. You select the resistance value to limit the current through the LED to be less than the maximum. The actual value of the resistor will depend on the voltage and the pulse rate duty cycle. For long life, use the highest value resistor that allows the led to be clearly illuminated, not a low value that allows the LED to be as bright as possible.

 

So, if you're applying 3V, and the limit is 30mA, then the resistor will be in the region of 100ohm, but I'd try a 1k, see if it was bright enough, then work towards 100ohm. .

[SHMD]

Hi Kev,

As I suspected, I'm here with the questions!! A quiet hour at work has allowed me to sneak off to the coffee machine with my printouts of this thread and read!

 

Firstly, I'm following your diagram initially.  Can you confirm my understanding of the Darlington Driver is right.  In essence (sorry if I get the terminology wrong) this is eight independent darlington pairs?  A darlington pair being two transistors that are amplifying the current going in, not once but twice, to give a greater power at the output?

 

Rich

 

Hi Rich, there are good answers above but, for the Darlington Driver, you are essentially correct - with a few provisos...

 

These Darlington Driver ICs come in 4 basic types - 8 or 7 Channel with/without Input Bias Resistors.

(Also in a variety of packages.)

 

Yes the ULN2803A has 8 channels. I use six to switch (one at a time) a 17-segment digit display on.

(The extra 2 I have brought out, to the edge of the PCB, for general purpose output use, with quite a lot more "umph!" than what the uC can give.)

NOTE! - These ICs do not have a positive supply pin to them!

 

 

 

Each Channel of the Darlington Driver, (with Input Bias Resister), can be simplified to the following:-

 

First, the Parasitic Diodes.

These are naturally formed when the ICs are made. Basically when one type of silicon is placed on top of another type of silicon forming a PN Junction, or Diode.

These can be ignored (pretty much) except in the following two cases:-

 

1/ When the input, or output, goes negative with respect to GND. This Parasitic Diode will then conduct and limit/pull-up/get-rid-of/dissipate the negative voltage. This is (usually) a GOOD THING.

 

2/ This “natural” PN Junction has a capacitance. Small(ish) but in the pF range. This limits the speed of the Transistors switching on (very very slightly) but will limit the upper frequency with which can be switched. This is generally a BAD THING.

 

I reckon you can completely ignore both these effects and so can forget all about Parasitic Diodes.

So why did I waffle on about them? You may not need them, now, but you do need to be aware of them wherever they pop up.

(There are many such examples of “not often needed” knowledge coming into play in lots of circuits and situations in Electrical and Electronic Engineering!)

 

 

 

Now the two Transistors.

Yes a small amount of current through the input transistor, will switch on (HARD), the output transistor. This means that this device is really a “digital” device. Your only ever want to switch it ON or OFF (and not fuss about with all that complicated analogue inbetween stuff!)

 

 

 

The three circuit Resistors.

The three resistors are used (together) to Bias the transistors. That is to ensure that they do not switch on until the input voltage has gone above a certain value. The input resistor also limits the input current (when used, in this case, with 5v ish circuits). This resistor, being built in, saves you buying/soldering/taking-space extra components on the final circuit. The two Base-Emitter resistors also “tie-down” the input, when the input is left floating, thus preventing the output spuriously switching ON randomly.

 

 

 

Finally, the Free Wheeling Diodes.

These can also be referred to as “Snubbing Diodes” or “Clamping Diodes”.

Again, these can be ignored, for the most part, but are very useful when the Darlington Driver Outputs are being used to switch inductive loads. When an inductive load is switched off, the magnetic field in it collapses causing a current. If this current has no where to go, then it will translate into a voltage that builds up very substantially very quickly. (DC) Relays and Solenoids need a diode, across them, to give the current somewhere to go thus stopping a voltage spike being created. These “Free Wheeling Diodes” fore fill this function, and again save needing extra components. If required, just connect this “common(!!!)” pin to the supply of the inductors.

LEDs do not have inductance, so you can ignore this pin and leave it open circuit.

 

 

Kev.