I'm just wondering what "standard practice" is for re-initializing a charac ter-based LCD display (i.e., a 2x16 backlit LCD)?
I have a project in the works now that erratically displays a blank LCD. I haven't quite tracked down what might be wrong, but I could "fix" the who le issue by just re-initializing the LCD on every pass. (This is very simp ly loop software..., check some inputs, throw some relays, etc..)
I don't think it's a flaky LCD, and like I said, I'm not 100% sure why it's doing this? The power rails are good and solid, and really nothing that stands out in the code (which is hand-assembler). And probably 95%+ of tha t code is from other projects over the years which I know to be bulletproof .
So, not asking for help on the "fix".
I'm just wondering how often (if ever?) folks re-initialize the LCD display ? This particular code never "sleeps".
If it did, I could more easily justify re-setting the LCD (via initializati on) when coming out of low-power, or power-down modes. (You know, just to be sure!)
This project uses an AT89LP51ED2 8-bit, 8051 derivative -- and the standard Hitachi language LCD character display. Nothing special. (The LCD is int erfaced in 4-bit mode, and my code has complete control over it.)
Nothing unexpected if I'm in front of it, even if for hours. But leave it a few days or weeks, and sometimes the LCD will be blank even though the code is working fine otherwise. A reset fixes it, as does a sof tware switch (toggle a port pin) to call the LCD init routine.
The Hitachi-compatible 16x2 display chips vary in the delays they need after each kind of operation. If you have a flaky system, add some big delays. If it's not flaky any more, go looking for which delays fixed it and reduce them to the required duration.
Quite often the cheap displays containing these controllers have no internal caps on the power supply. I had similar problems that several times when the customer switched to cheaper displays. But your argument about the timing is also correct, the required timing seems to vary with he moon phase.
Those eBay-special displays can be pretty sensitive to brownouts/droops/noise on the supply rail; even very brief dips in the supply or EMI injected from a motor spinning up or relay switching can cause the controller IC to latch up and the display won't come back up until the power is cycled.
I would ensure the supplies and ground were actually as "solid" as I thought they were
The other thing to check if you can get at any of the LCD pins is to make sure that there is an oscillating signal present. LCD displays fade to nothing if they are DC biassed for any length of time.
That it will run for many days or weeks without a problem and then blurp all of a sudden doesn't sound like a timing issue or noise on the clock and data lines, that usually results in garbled characters but not the whole display going down.
If boards are already made and stuff re-initializing the display every so often is one of those pragmatic hacks that while not particularly elegant has probably been done before without anyone noticing.
One thing that might be a good idea to do in future for debugging this kind of problem is if one is gonna go the China-special route on "Hitachi-compatible" 16x2 LCDs whose brain is a little chip-board-blob made by God-knows-who is I picked up some display modules from the late
1980s (also on eBay) that have the actual HD44780 OEM IC from goddamned Hitachi themselves on the board.
That way one can do a "compare and contrast" on the performance and if the OEM works and the China-special doesn't you at least know the problem isn't you.
Do not know if your processor does anything else apart from driving the LCD, but check for brownout and power cycles, check that your reset is correct in those cases. You could test by switching power on/off a few times.
I re-initialize our 122 X 32 pixel (20 chars X 4 lines) every 5 minutes although I never notice it.
This was because the LCD is above a switching power supply and the EMIs want to get back to ground somehow. Does the problem go away when you move the LCD away from the processor ?
Another thing to try if necessary is to put the lines driving the LCD through a common mode ferrite to help keep the common mode signals away. More careful bypassing of EMIs in the power section have quieted things down a lot so no need for the ferrite but it's simple and free to re-initialize once in a while, lest the display justs get stuck in a bad way and have to make the user re-power the whole thing.
Now that's interesting... The left end of the display is fairly close to the 52 kHz switching regulator (more importantly, the coil). By close, I mean half an inch, maybe. Of course, it passed emissions.
MAYBE the regulator is doing something squirrely under light loads (i.e., switch from continuous mode?) Hummmm..... new ideas and places to look. Thanks.
The LCD is on a ribbon (2-inch), but I could monkey with it. Might be easier to just shield it. Need to look at the cabinet again for that. I mean, if that's the root cause and not just a ching-chang-chong issue. I wonder if we still have any legit Hitachi LCD controller styles laying around...?
If you don't find the original display board, perhaps you can extract something useful from the original data sheets:
If you want an LCD display with a real HD44780A00 chip, I have a display that I extracted from my Korg DSS-1 synthesizer: Here's the board: I just peeled off the paper label covering the chip and found that's a genuine Hitachi HD44780.
The display has an internal electroluminescent backlight that over the years had dimmed and faded away. It's difficult, but not impossible, to see the display. Also, it's not a 16x02 but is a 20x02 display.
You're welcome to the board (for free) if you want it. Send me an email (see signature) with a USPS mailing address and I'll mail it to you.
--
Jeff Liebermann jeffl@cruzio.com
150 Felker St #D http://www.LearnByDestroying.com
Santa Cruz CA 95060 http://802.11junk.com
Skype: JeffLiebermann AE6KS 831-336-2558
Forgot to mention that the most valuable change to the PCB was swapping the inner layers to make sure the ground plane layer (negative actually) was next to the bottom layer on the stackup... Also, moving some traces that were on that ground plane to another layer and using more of that ground plane for actual ground (negative). Better shielding against the EMIs trying to get back to ground through the sensitive signal lines.
Don't overlook the fact that all colander layers carrying voltage rails that are linked to the zero volt 'ground' reference rail via decoupling capacitors are effectively 'ground rails' as far as screening and 'ground return' pathways are concerned.
The PCB designer should be more than well aware of the pitfalls of unexpected parallel tuned resonance effects when using decoupling capacitors in parallel to 'make two (or multiple) separate voltage rails behave as one' with regard to the whole spectrum of frequencies being handled by the circuit components, including unwanted interference signals that might leak past any assumed external screening provided in the final build.
Even assuming the designer correctly specifies the decoupling capacitors (and their placement), it's still possible for manufacturing to derail the designer's best intentions either due to component supply logistic issues or simply 'Beancounteritis' using cheaper sub-standard substitutes for those critical parts.
In the case of 'Beancounteritis', it's often the tactic of eliminating 'surplus to requirements' parts[1] that don't appear to provide any useful function during the functional testing phase but the removal of which can save thousands of dollars on a half million production run which represents at most maybe a whole half a percent on the BoM costs.
Astonishingly, the beancounters don't seem to have any understanding of mathematics beyond 1st order "Sums" nor any appreciation of such penny pinching actions on the net worth of the company they have been placed into the unwarranted position of destroying by such short sighted 'money saving' actions.
[1] I'm forcibly reminded of one very memorable act of Beancounteritis vandalism that left one reviewer of Akai's then latest and greatest R2R tapedeck, namely the GX630DB (of which I owned one example of - hence the interest in the reviewer's article) rather puzzled by its playback performance using pre-recorded tapes made on other tape decks which had successfully recorded to levels in excess of +8dB VU but which clipped in the GX630DB's replay amp.
The 0dB VU level is normally set to 12dB below standard tape saturation (maybe corresponding to 15 to 18 dB below the saturation level of the more modern high output low noise tape formulations currently available from Japanese tape manufacturers (Maxell and TDK) at that time).
Since the only perceivable limits on performance should be that of the magnetic recording tape alone (the weakest link in the whole enterprise) rather than have anything to do with the amplifiers which could so easily outperform the limitations of the media and therefore contribute no further degradation to the process of recording and playback of analogue magnetic tape recordings, this finding was rather a disappointing surprise to the reviewer who tried to mitigate just what a disaster it truly was by claiming it could be worked around by using lower recording levels than were considered desirable with higher noise tapes to take advantage of the modern low noise tape formulations, overlooking the glaring deficiency in handling existing high level recordings.
He was obviously motivated to excuse this singular defect on a machine that so clearly outperformed everything else in the market including "professional class" machines costing ten times as much, in every other way (almost unmeasurable wow and flutter, virtually no head scrape noise due to the absence of pressure pads and glass smooth glass crystal heads (which, as a bonus, didn't so readily clog up with shed tape coating material) and with a service life at least two orders of magnitude better than the standard Mumetal based heads).
Since I had taken the precaution of obtaining the service manual for this tape deck as soon as I'd bought it, I was able to peruse the circuit diagrams to try and work out why the replay amplifiers had such a low clipping point. Luckily, the circuit diagram showed test voltage values at critical points which swiftly indicated where the problem lay.
The 'DB' suffix in the model number indicates the use of the Dolby B noise reduction system. In this case, the deck used a total of four Dolby circuit boards, two in the playback amplifier chain and two in the recording amplifier chain.
In the case of the recording chain, they sat between the bass and HF EQ sections (the bass lift being applied first meant the Dolby processor, which didn't react to this subsonic to 200Hz frequency range, could be safely used at this point in the recording amplifier chain[2]).
In the case of the playback chain, they formed the last link immediately connected to the line out sockets via the 2.7K? emitter load resistor of the emitter follower output stage. The Dolby boards were powered from a
24v rail which supplied 21.8v to the output transistor's collector so one might reasonably expect to see an emitter voltage on a class A emitter follower amplifier stage around about the half voltage mark, circa 11 volts in this case. What the circuit diagram revealed was a test voltage value of 5.5 volts, just half of what it should have been!
I didn't want to alter the amplifier gain so I added a capacitor and a resistor to the bias feedback network so I could alter the DC feedback without upsetting the audio frequency gain in order to re-bias the emitter voltage to 12 volts to eliminate the worst of the asymmetric clipping characteristics of this resistor loaded emitter follower stage.
Seeing as how driving even a 10K? line input would aggravate the asymmetric clipping behaviour (there was already a 33K? load after the output coupling capacitor to the line out socket), I decided to replace the emitter resistor with a constant current generator so as to preserve symmetry of clipping down to 3K? loadings, effectively raising the originally observed clipping level by some 10dB (If a job's worth doing... and all that). In so doing, I re-optimised the emitter voltage to a value of 10v (figuring on a 2v drop from the driver collector load resistor I'm now guessing).
It was quite gratifying to quash the handiwork of Akai's beancounters but even more gratifying was my discovering about a year later from a Wireless World magazine article which showed the original Dolby B circuit in its intended form which revealed that I had simply recreated the original biassing/feedback network that Akai's beancounters had so devilishly vandalised.
After you've read [NOTE 2] below, you'll understand my utter contempt for the beancounters of this world.
Strangely, in the case of the GX747's electronic tape counter, a version of beancounteritis could have immeasurably improved the accuracy of this hours, minutes, seconds tape counter by saving on the cost of a drive belt and a seperate optical counter disk and its bearing assembly by mounting the optical sensor directly to the rubber faced tape driven counter shaft. Ghod only knows what possessed the designer to cobble up such an abortion as this in the first place. :-(
[2] It could have more conventionally been placed ahead of the whole recording amp's EQ chain but this was a neater insertion point. However, the beancounter's act of sabotage gave rise to another observation by the reviewer with regard to bass frequency clipping and gross intermodulation when recording at moderately highish levels (above 0dBVU), a problem I'd assumed to be due to a limitation of lower flux saturation levels in the choice of ferrite over Mumetal in the recording head, an assumption I may have have adopted from the the reviewer - I can't be certain now after all this time.
Anyway, I knew something that the reviewer didn't, the placement of the Dolby recording boards between the bass and HF EQ sections of the recording amps. The penny hadn't dropped straight away since it was only months later that it occurred to me what the true nature of this poor bass frequency recording limitation might be.
I performed an almost identical modification to the recording Dolby boards which completely cured the issue of bass distortion/ intermodulation on moderate to high levels of recorded bass, thus completely exonerating the glass crystal recording head of all blame. Now, at last, I could aim for healthy recording levels into the red of the VU meter's scale just like I'd been able to with my Akai 4000D a decade earlier.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.