Construction of another digital clock

[Miguel Lopez]

Hi everybody

I have begun the construction of my next digital clock. As I always want to make something new, for fun and learning, this time I will use three twin multiplexed seven segment lamps, with commom-cathode LEDs. I have never built a multiplexed display before. I will use SN74XX TTL chips and some CMOS.

There are several block which need to be designed:
- Lamps circuit, with segment handling, and digit selection inputs
- Digit selection and data selection circuits
- The clock (I will use the same circuit that i used for my Nixie Clock)
- Power supply and UPS system (same as previous one)

So, I will focus on the first two blocks. I have already designed the PCBs and mounted the components. I'm using about 1 kHz for display sweeping, so digits appear very steady.

The seven segment decoder will be the old good 7447. As this lamps are commom-cathode type, and the 7447 is designed for common-anode operation, each segment will be powered by a PNP transistor driven by the 7447 outputs. This way the transistors act as inverters, and so the common-cathode operation is achieved.

Then, each digit (7S lamp) will be enabled using an NPN transistor which will be driven by the selection circuit. Sequence is:
1.) All lamps OFF
2.) First BCD number is selected (no lamp is ON yet)
3.) Lamp corresponding to that BCD number is enabled (ON)
4.) This lamps is turned OFF (all lamps OFF again)
5.) Next BCD number is selected.
6.) Lamp corresponding to that BCD number is enabled (ON)
7.) Repeat for the rest of the lamps, and then repeat the entire sequence continuosly.

The data selection will be using 7403 open collector gates. The BCD output of the same order of every digit will be shorted and connected with a 1k resistor to VCC. When a logic "1" is applied to the other input of the NAND gate, the output is then dependable of the other input and the it will appear at the output inverted, so the four outputs of the 7403s must be inverted before be applied to the 7447.

Here some pictures to show at what stage I am.

First picture shows all of the lamps ON, althought indeed there is only one ON at the same time, but as the refresh time is fast, they seem to be all ON.

Second picture is the data selection and lamp enabling circuit. Uses a Soviet K133LB1 for oscillation, a 4518 for frequency division, a 7442 for decimal counting, and two 7400, and two 7404 for combinational logic.

Third picture is the back side of the display, which will contain the data selection circuit and inverter ( five 7403, one 7400, one 7447 (already mounted), and some transistors)

[Miguel Lopez]

This is the circuit that I'm using for sweeping. The outputs of the 7404s select the right number, while the output of the 7400s (once inverted) provide enable pulse for the right lamp. As it is said in the picture, I replaced the 7490s by a single 4518.

I'm using about 1 kHz for display sweeping

Good show, lots (most) multiplexed displays use a much lower frequency and look flickery.

[Miguel Lopez]

Originally Posted by merlinmaxwell

multiplexed displays use a much lower frequency and look flickery

I have never understood why. It is not very difficult to increase frequency to avoid flickery.

[julie_m]

I always thought, it was because you can have any frequency which is a factor of 100 Hz. (or 120 Hz. in some countries) gratis, without going to the trouble of building your own oscillator

without going to the trouble of building your own oscillator

Even with new microcontroller run kit it can be (and usually is) painfully flickery, even 1kHz would only use a small fraction of a percent of the micros time.

[julie_m]

The 1970s / 80s LED clocks I'm thinking of used to use a single IC driving the display directly; and the unsmoothed, rectified supply was the only source of timing. The drive arrangements were such that the total current through all lit segments of any digit was limited, so they were brighter at eleven minutes past one in the morning (1:11 = fewest segments lit) and dimmer at eight minutes past eight in the evening (20:08 = most segments lit)!

Even some stereo cassette decks used to use the supply frequency to multiplex both LED level meters from a single driver IC. Though the duty cycle was higher at 50%, as opposed to 25% for a four-digit clock display, so the flicker would be much less noticeable.

[Mike. Watterson]

The cheap LED mains clock radios sold in Argos and Tesco today mostly differ from 30 years ago in having poorer radio and a single LED module. The clock bit and multiplexing is pretty similar. I've bought the Argos ones sometimes just for parts!

[Miguel Lopez]

The other thread was closed, but this project is still ongoing, so in order to share with you about it, I open this new thread. The clock was "mothballed" for several months but now I have resume it.

I finally mounted the 7403 chips (in fact, TESLA 8403), to select the data to display. I forced the inputs to 1, 2 , 3, 4, 5 and 6 BDC numbers to test. Here is a picture of the display showing the data.

[Reelman]

Good job Miguel, well done!
Peter

[Miguel Lopez]

Thanks Peter. I would like to remember that that 7-segment display is multiplexed, so those digits are in fact "sweeping" at 1kHz approx.

The "6" is not closed as I'm using an SN7447 for 7-seg decoding. I don't like open "6" and "9" so I will soon replace it by a D147 from the former GDR.

"sweeping" at 1kHz approx.

Good job too, a lot of 'professional' stuff I see is scanned at less than a 100Hz or so making for an awful flickering display.

[gingpeakin]

Elektor published a way of closing the 6 and 9 from a 7447 with a single transistor. I have that issue fro 1975 I think - I'll have to rummage around a bit to find it.

Ging

[Miguel Lopez]

Originally Posted by merlinmaxwell

a lot of 'professional' stuff I see is scanned at less than a 100Hz or so making for an awful flickering display.

We talked about it in the original thread:
https://www.vintage-radio.net/forum/...d.php?t=127307

Originally Posted by gingpeakin

Elektor published a way of closing the 6 and 9 from a 7447 with a single transistor

I have done it with a aditional 7400.

Now analyzing this, I noticed that if you supply the base of an NPN transistor from the B bit of the BCD data, and its collector is wired to the "a" output of the 7447, the "6" gets closed. This transistor will also power that segment in the "2", "3", and "7", but in those states, that segment is also ON, so there is no problem.

The same can be done with the "D" bit of the data and the "d" output of the 7447, to close the "9". It will power the "d" segment in the "8" state, but again there is no problem as that segment will be powered too.

Anyway, I will use the D147

We talked about it in the original thread:

So we did, glad you did it! I must get a memory chip installed.

[Brian R Pateman]

I have re-opened the original thread and merged the new posts into it.

That way we can all see the history of this interesting project.

[Miguel Lopez]

Thanks Brian

This time I will use 4518 and 4011 for the minutes-seconds section, and 4027, 4011 and 4081 for the hours section. Here are the circuits. They were simulated in Multisim 7.

[Herald1360]

Does that simulator actually "light" the LED segments to show the status? Neat if it does!

[Miguel Lopez]

Yes, it does. All of the simulators that I have used show the 7 segment digit.

[Miguel Lopez]

The PCBs for the clock itself and the PSU-time base are waiting for etching. I will do that this week posibly. I would like to share another design that I did, while I was trying several chips to be used. I tried 74192 (too many milliamps) and CD40192. But I also tried the 7490 and it worked. I tested both in the simulator and the breadboard. It worked nicely!!!!.