problems with Sparkfun Serial Enabled 20x4 LCD

[Admin]

I'm using the Sparkfun Serial Enabled 20x4 LCD:
http://www.sparkfun.com/commerce/product_info.php?products_id=9568

It works perfectly fine, displays text correctly, etc . . .

until I turn on the backlight . . . and then it turns into a stubborn inconsistent dysfunctional child.

I'm using WebbotLib to send it commands, so assuming it isn't buggy, my code is fine. I'm using an Axon II with 1.5A possible from the regulator.

So my question is . . . has anyone used this display, with the backlight turned on?

(doesn't matter if you're using WebbotLib or not, I just need to verify that Sparkfun isn't selling crapware again)


Doing more research on it, there seems to a problem that it'll work fine with an Arduino, but not fine with an Arduino + shield. I'm not using an Arduino, but this definitely sounds like a flaw in their design!
http://www.arduino.cc/cgi-bin/yabb2/YaBB.pl?num=1273837091
http://forum.sparkfun.com/viewtopic.php?f=14&t=21776&start=15

[madsci1016]

Don't own one, but since they are PWM'ing the backlight, my money is on switching noise on the power lines. Maybe straight to Arduino keeps it close enough to the noise caps on the Arduino, but through a shield becomes to far.

Try adding noise caps, 10uF and .1uF, as close as you can, if not on the backpack itself.

Looking at the SFE pictures, i don't see noise caps at all on the board.

[Admin]

A possible theory . . . I'll try caps next week after I find a few of appropriate values. I don't have any on me right now other than 3300uF electrolytics

The splash screen works fine even with the backlight on, so perhaps the backlight distorts the baud rate somehow. Perhaps this is why it doesn't work with the Arduino shield, as it increases the UART wire length allowing for greater noise distortions . . .

I also just noticed, which is very typical of Sparkfun, that the actual device doesn't match their posted schematic. There are ~13 more unidentifiable components on the actual device than there is in the schematic

[madsci1016]

The backlight draws 150mA at 4.1V.  Since I don't see any resistors on the schematic, it either isn't on the schematic, or even when the backlight is on they are still PWM'ing some duty cycle to keep the LED from burning out. Or the voltage drop across the NPN could cover it; i forgot about that.

Either way, PWM of 150mA and roughly 5V is going to cause some noise, and there's only a .1uF cap in the design, which isn't enough.

Does the randomness seem like it's having internal troubles, or just like it's getting junk on the data line?

[TrickyNekro]

Provided that you have a oscilloscope, can't you check the noise levels???

I guess you don't....

[Admin]

I use the oscope at work, which I'm not at on a lovely Saturday night

I studied the board more carefully, and its obvious they put zero effort into minimizing the effect of noise. I traced the power line as far as I could, and it goes over 8 inches before even touching another component! After that its hidden under the display and a black blob, but it appears its over 15 inches long before even coming into contact with a capacitor - HORRIBLE! Same with the ground wire. Considering it has a high powered PWM going through it, think of the RF from that! The wire is quite thin for 150mA+, especially with the vias that are equally too tiny - probably has a huge voltage drop across the length.

At least the Rx line is an inch long, which is ok.

There is only one capacitor on the board (besides two for a timing crystal), and if the schematic is correct, its only 0.1uF . . . your right, thats not enough for wires this long. I think I'll unsolder it and put on something much better. But even then, its *far* away from the high current components on the board, so still not a perfect solution . . .

[Soeren]

Hi,

There is only one capacitor on the board (besides two for a timing crystal), and if the schematic is correct, its only 0.1uF . . . your right, thats not enough for wires this long. I think I'll unsolder it and put on something much better. But even then, its *far* away from the high current components on the board, so still not a perfect solution . . .

Might be worth getting rid of the trace with an exacto knife and run a shorter Kynar (?) wire instead - not a nice solution, but sometimes very effective.

[TrickyNekro]

Hi,

There is only one capacitor on the board (besides two for a timing crystal), and if the schematic is correct, its only 0.1uF . . . your right, thats not enough for wires this long. I think I'll unsolder it and put on something much better. But even then, its *far* away from the high current components on the board, so still not a perfect solution . . .

Might be worth getting rid of the trace with an exacto knife and run a shorter Kynar (?) wire instead - not a nice solution, but sometimes very effective.

Or you could just scratch the plastic/covering/protecting/ what ever gives that cool color layer, and place some extra capacitors ;-)

PS... sorry long day on the road...

[Admin]

Another problem I found with it - they didn't set a brown-out fuse. I have to completely disconnect the wiring to power down and reset the display. The capacitors on my Axon II 5V line keep a voltage for quite some time after power down.

[Admin]

The displays were working fine yesterday. I turned everything on today, making no changes, yet the baud rate changed to some other unknown rate. Trying to fix it, suddenly the backlight turns on again, although I issued no command to do so. An unknown baud rate + a backlight that breaks commands = I give up.

[madsci1016]

Sounds like it was receiving junk on the serial line, and the junk happened to be accepted commands for some things.

Like the review you left for it. Sometimes I think Sparkfun's 'engineering' team is full of anything but engineers.

[Admin]

Sounds like it was receiving junk on the serial line, and the junk happened to be accepted commands for some things.

Like the review you left for it. Sometimes I think Sparkfun's 'engineering' team is full of anything but engineers.

I'm actually thinking they hire cheap summer interns to pump out sub-quality products.

That said, I will no longer buy any GPS or LCD display designed by SF, as they sell these items without properly designing/testing them.

(if it was 'made in China', what would we say? )

[madsci1016]

That said, I will no longer buy any GPS or LCD display designed by SF,

Don't forget about Xbee regulated boards!

[Admin]

That said, I will no longer buy any GPS or LCD display designed by SF,

Don't forget about Xbee regulated boards!

Thanks for reminding me - I just left a 'proper' review of it on SF. Hopefully it saves someone money, time, and frustration . . .

And now that I think of it, their BlueSmirf bluetooth devices seem to spontaneously fail more often than they should . . . three inexplicably died on me over the years . . .

[madsci1016]

They started off as garage hobbyists, and turned into a professional company. But I think they are still mostly garage hobbyists, as seen by the quality of some of their designs.

Did you see my comment about the Xbee board? I wonder if anywhere there even took notice. 

[Admin]

Did you see my comment about the Xbee board? I wonder if anywhere there even took notice. 

I just read it, and your comment was definitely better written than mine. I also noticed a -2.0 next to your user rating . . . wtf?

I clicked the smiley next to your name, and it changed to -1.9.

Does a mod not like you at SF?

[madsci1016]

Does a mod not like you at SF?

Possibly, and the feeling would be mutual. I have mostly been critical in my reviews, but always try to stay constructive.

I noticed big jumps in their rating system (for my rating) as i navigate to different pages. They also don't list all your comments on any one page so i couldn't find where all the negatives were coming from.  I tried googling it once and found posts of users in the -100's. I chalk it up to another poorly designed Sparkfun system!

[madsci1016]

and suddenly it's +1.9 on the homepage comment. good grief.

[knossos]

Did you see my comment about the Xbee board? I wonder if anywhere there even took notice. 

I just read it, and your comment was definitely better written than mine. I also noticed a -2.0 next to your user rating . . . wtf?

I clicked the smiley next to your name, and it changed to -1.9.

Does a mod not like you at SF?

I clicked the smiley too and it went from -2.4 to -2.6... ? lol sounds like they just have problems.

[Webbot]

If its of any help then I really recommend the displays from Matrix Orbital http://www.matrixorbital.com

They come with excellent online PDF manuals and most of the (serial) boards allow you (via jumpers) to use either TTL serial or RS232 serial - so you can connect them direct to your PC or to your robot. They also have parallel and I2C.

They have dealers all over the place: USA, Canada, Australia, Asia, Europe/UK.

May not be the cheapest option - but they work. They also have all sorts of more complex stuff: graphic displays and touch screen etc

I've used them for a few years now without problems and NO - I'm not on commission ! Of course that will be the kiss of death and I'm now expecting a storm of replies saying 'I had problems!' or 'Ive looked at the schematic and its got the same noise issues' ....

[madsci1016]

Did you see my comment about the Xbee board? I wonder if anywhere there even took notice. 

I just read it, and your comment was definitely better written than mine. I also noticed a -2.0 next to your user rating . . . wtf?

I clicked the smiley next to your name, and it changed to -1.9.

Does a mod not like you at SF?

Sparkfun added a search for comments now. Seems my low score was because I was one of the first critical comments on the "Free Day Recap" post. Apparently that made a few fanbois upset.

Also there's been talk on the Sparkfun Xbee Regulated Board comments including a SFE employee, but no reply to our findings. Shame on you Sparkfun!

[Admin]

To add to my SF complaints, I ordered two AVR development boards from them. The chips were tiny QFNs.

You know where this is going . . . THEY DIDN'T WORK!

After looking carefully, pins were shorted, and half of them were cold joints.

It annoyed me even more that they didn't leave any space between the ISP connector pins - it's impossible to plug in a battery or uart headers when the ISP is plugged in . . .

And the PCB traces . . . it looks like they used Eagle auto-router, and didn't bother to clean up all the ridiculousness . . .

I was really a SF holdout, praising their greatness, but now . . . ehhhh, the Chinese can produce better quality products at the same price.

On the bright side, after sending them magnified images of the problem, they promised to replace the units. I haven't gotten them yet, but they want me to send back the non-working units when I do.