Quality appears as good as any other displays I've seen at least when lighting one segment at a time. They are very bright and should hold up to a lot of multiplexed dimming. Speed was pretty good. I believe it took about a week and half to get them. I paid via PayPal, but they also accept bank transfer. The other fees and shipping were pretty high, so I would recommend you buy a large enough quantity to make it worth it. They also make custom displays, so if you need an LED display with symbols and custom colors they can do that too. The ones I got were alphanumeric .8 inch and dual .5 inch ultra green (525nm) and blue (470nm). My only complaint is that the .8 inch do not have decimals that light up even though there is a place where a decimal appears to be. I'm not having a lot of luck getting these displays driven yet as I can't seem to get SPI working out of our PIC. There are a lot of modes and settings to choose from and I can't seem to find one that works. As soon as I can get more than one segment lit up I will post a picture of them in operation. Stu
Kwality LED Photos It looks like we may have figured out how to do the SPI. So far we can enable test mode. We still can't figure out how to do anything but test mode, but I think we just have to decode the register maps in the datasheet of the MAX6954. Take a look at the photos below. The black stuff on the green LEDs is no part of the displays. It comes off easily. Stu
Mike, I have used something similar in a PIC based clock some time ago (though I applied the correction term once per second). One way to make it easier for users to adjust the trim value is to relate it to an error they can measure. This is how I did it... T = (E*I)/(D*0.0864) Where T is the required trim value, and E is the clock error in seconds out over D days. I is the number of timer counts (instruction cycles in my case) per microsecond. By setting D such that I/(D*0.0864) = 1, the error term E input by the user is equal to T. For Fosc = 20MHz, D should be set to 57.87 days (or 58 days, rounded). Granted, 58 days is a long time to wait, though 58/8 is approx. one week. Multiplication of the user input trim value by 8 is really easy and the errors created by this simplification are small enough to tolerate for ease of use.
I also used something similar but also applied the correction term once per second which gave excellent accuracy. Applying the term on each rollover was not accurate enough for me. Have you guys considered using lower frequency crystals to give you better results? 20MHz is pretty high just for a clock.
Am I glad I use DCF77.... I indeed do use a lower 4.096000MHz Xtal wich gives exactly 1.000sec for 2x250ticks of Timer0 with Prescaler=8, but then long term accuracy is not much of a concern to me.... CD
The DCF77 module is certainly an advantage there but I'm cheap so can't justify the extra expense . The other option is using the Timer1 oscillator pins for a low power 32.768kHz crystal in addition to the system clock for far better accuracy (you could even use the internal 8MHz oscillator (+PLL for 32MHz) for the system clock in this instance if the PICs you guys use support INTPLL). There'll be some more pictures of the Matrix display soon as construction is almost complete. Unfortunately nothing fancy is being done with the clock as that is not the display's primary purpose. How is your matrix clock going Chris?
A DCF receiver module from Conrad.nl (ref 641138) is only 9.95 Euro (GBP 6.81) without shipping. And I know DCF77 reception works in North Yorkshire since Richard lives there. I'm still using the 16F PIC series controllers as my Proteus VSM version (6.3) doesn't support the 18F series...(although I have a few 18F's laying around). I would love to see more pictures of your matrix display. I actually paused the development on my dot matrix LED clock because of my company move. I should have more time in the next few weeks. CD
I'll get some pictures up soon. There's no proper functionality on there yet as it's all in a bunch of different asm files for each function (I'm still testing most of the functions on my breadboard before making the final files for the displays). My avatar shows a tiny fraction of the action <------- There's five lines of text on the display, but they're quite well spaced out (about 100mm gap between the bottom of one line of text and the top of the next). Total cost of LEDs was about £200 at around 3p per LED
Hi Steve, I am applying the correction once per second using Timer 2 (no roll-overs)... Perhaps I'm missing your point? Why would a lower frequency 50-ppm crystal give better results than a 20-MHz 50-ppm crystal? If you're adjusting timer frequency once per second with resolution of 1 Tcyc it would seem that the higher frequency crystal provides better/finer tuning... What am I missing? Kind regards, Mike
Re: your correction - I misunderstood when you were applying the correction. Re: crystal accuracy - The 32.768kHz crystals available in the UK are usually 10ppm or better giving better accuracy. However if only using 50ppm crystals, a lower frequency crystal will be more accurate because at worst case for a 20MHz crystal your crystal could vary by +-1kHz compared to just 200Hz for a 4MHz crystal. This leads to the result that you would notice five times as much error with a 20MHz crystal than with a 4MHz crystal in the same period of time. I apologise for the poor explanation but it's pretty late...
Hi Steve, Your explanation is fine but I'm not sure about your logic... Assuming you're using timers, your 4-MHz xtal clock code happily counts off 1,000,000 Tcyc's to produce a 1-second 'heartbeat'... My 20-MHz xtal clock code counts off 5,000,000 Tcyc's... If you do the math you'll find the Tcyc error in your 4.000200-MHz xtal and the Tcyc error in my 20.001000-MHz xtal both produce 59.997-seconds per minute... Assuming the resolution for trimming the timers is one count or 1 Tcyc you can only 'tune' or 'trim' the timer to within approx 1-usec/second with the 4-MHz circuit compared to 200-nsec/second with the 20-MHz circuit... The 20-MHz circuit with its 200-nsec Tcyc provides finer tuning/trimming and more accuracy than the 4-MHz circuit with its 1-usec Tcyc... Take care... Have fun... Kind regards, Mike
You're probably right actually, I just remember reading this somehwere and took it as read without checking the logic behind it, unless they were only talking about the more accurate and stable 32kHz crystals...
Of course it is possible, it all depends on how you implement the keyboard routines (key scanning, key debouncing, actions on key presses, ...) There are lots of example matrix keyboard PIC routines available. CD
I think he's enquiring regarding the functions on the keys rather than how to interface the keypad, but either way - it's all possible.
Hi Chris, So any thoughts on the pictures I posted? I think getting some white ones would be good as they can be filtered to display almost any color. Stu
Cool Thanks Turbokeu The alarm clock is coming on slowly but rapidly. I have got the multiplexing for the 4 &-Segment LED's and getting on well. I have already got loads of different ideas now to progress with this.
They seem perfect, nice color! How did you ask for quotes? I'm not able to send the enquiry form on Kwality's website (404 error) I have 0.8" white displays and I like them very much. Unfortunately LC-Led still doesn't have the 4.0" white displays on stock... I would be interested in the Kwality 6", 8", or 12" jumbo segments. And they do carry 4.0" 5x7 dotmatrix displays CD
Hmm. I just e-mailed them at "kwality at kwalityindia.com" and told them what I wanted. That's the address I've been communicating with them on. For an idea on prices, the dual blues in the picture are $7.35 each which I think is a very good price. Hope this helps, Stu
