Page 2 of 2

Posted: Fri May 05, 2006 12:56 am
by MrAl
Hi there Robert,

In my counter, i never intended to go through the trouble of
lighting the decimal segment even though i intended to use it with
a 10 second time base (as well as a 1 second time base) to
get readings with resolution of 0.1 Hz (20MHz was my goal using
a count of 200000000 and 'implied' dp to the left of the right-most
digit). For one application i wanted to look at the stability of
various kinds of oscillators.

But, if i was going to light the decimal point, i would imagine i would
have to multiplex that as an 8th segment, maybe using 3 bits of a
4 bit latch to hold the current position of the dp (zero through seven).
The latch outputs would then run to a 'one of eight' decoder
and the decoder outputs could be scanned with NAND or NOR gates,
the output of these gates driving the dp 'segment'. The scan pulse
comes from the scan generator that scans the digits and occurs
at the correct point in time because the two counts are equal only
when the scan count is the same as the dp count.
Alternately, we could use a comparator logic package to detect 'equals'
between the dp latch outputs and the scan count in binary. When the
two are equal a pulse outputs from the comparator.
Also, an neat alternative to using a dp latch is to use instead another
4-bit counter, where you can set the dp position by pulsing this counter.
That would be neat i think :grin: as you could move the dp up up up
and then it would reset back to the first digit after 8 pulses (maybe from
a toggle switch or from logic). If the counter was a presettable type
you could also load it from other logic or whatever.

The only trick is the dp has to be a pulse that occurs for the same
time the scan pulse occurs, and only for that digit.

There may be simpler ways too ... if you want we can give this
some more thought.

Let's see, the above would take the following logic packages:

Circuit #1
1. DP counter or latch
2. 1 of 8 decoder
3. one NOR gate package for placing 4 decimal positions, or two NOR
packs for placing the full 8 decimal positions.

Circuit #2
1. DP counter or latch
2. 4-bit logic comparator package
3. 1 of 8 decoder

Circuit #3
1. DP counter or latch
2. For placing the dp in up to 4 positions, program a 4 bit ROM to detect
equality between the scan count and the dp count and generate a 4 bit
dp scan pattern, or, for placing the dp in up to 8 different positions
program an 8 bit ROM to do same only generate an 8 bit dp scan pattern.
(The rom chip takes care of all the logic necessary to generate the
scan pattern to drive the decimal points, provided we have access to
the scan generator binary output. If we dont have access to this,
we can still do it but we'll require a larger ROM memory).

So circuit 1 takes either 3 or 4 packages, while circuit 2 always
takes 3 packages. Circuit 3 only takes 2 packages but requires
programming a rom chip (4 bit or 8 bit, with 256 memory cells),
and the rom program pattern is very easy to generate on a
computer.

Sound ok, or sound too complex for what it does?

Either way, sounds quite interesting to me now :grin:

Posted: Fri May 05, 2006 8:53 am
by Robert Reed
Mr Al
Yes, there is a lot of complexity involved for just lighting simple decimal points in these situations. The main problem I have had here, is I am usually running out of board real estate at that point . Wouln't it be nice if the manufacturer would simply draw out a few connection points on the back of their displays just for this purpose. I think I have come up with the simplest way to do this with my home brew "decoder". It requires one 2N3904 and one resistor per decimal point. I have also used a variation of one 'AND' gate ,1 diode and 1 resistor. If interested, I could go through some of my designs and elaborate further on this.But the bottom line is its still an annoyance to have to go through any of this for one tiny little segment.

Posted: Fri May 05, 2006 10:33 pm
by MrAl
Hi Robert,

The transistor solution does sound interesting.
Yes, it's a pain and it would have been nice if they
brought out a separate lead for the dp's. Too bad though.

Posted: Sat May 06, 2006 4:19 pm
by Robert Reed
Mr Al
I thoght I would give you more insight on the circuits I have used regarding Mx'ed decimal points.Maybe in a pinch it may help you with some similar situation down the road. I have only used this in situations where moving the decimal point was required when switching ranges and done so throgh the same range switch.This particular example uses only two moves of the DP, but more could be easily added. The emitters of two 2N3904's were tied together and connected to the DP common terminal of the optics.The collectors were connected to the range switch which placed 5 volts on either at the proper time. Each base was driven through a 10K resistor from the active common terminal of the digit desired, so that when that digit was "live" the transistor base also was. If the range switch had put 5 volts on the collector atr that time, then only that particular DP would be lit. Sorry I cannot seem to transport the print, but if you look at the capacitance meter article in this month's N&V (in the display end of the shcematic), you will see a clearer picture. I have also used leftover AND gates on other projects to accomplish the same purpose. AND gate outputs were each connected to it's own isolating diode of which the other end of these diodes were parrelled up and fed a resistor of the proper value (current limiter) to the DP common terminal. One input of each AND gate was taken hi by the range switch at the proper time as in the transistor version(these need pulldown resistors), and the other inputs were switched hi by the "live " digit driver as was done in the transistor example. The mx'ed displays were LITE-ON LTC-4724J7 in both cases. Obvously, resistor values and polarities would have to be juggled for any individual case. Just thoght this may help you at some point when board space is getting critical--or at least until we can come up with something better. :grin:

Posted: Sat May 06, 2006 7:17 pm
by Ken10
Hello Mr Al,
I'm new here and this is my first post.

I noticed you had allowed for a 0.5 volt drop for the 7447. I looked at the data sheet and didn't see anything concerning a voltage drop on the output. Please let me know where this is stated in the data sheet.
Thanks in advance,
Ken

P.S. If anyone can tell me how this 0.5 volt drop was arrived at, it would be much appreciated.

Posted: Sun May 07, 2006 6:39 pm
by bruceb
It is a given, since when you look at the internal schematic of the 7447 or
most other IC's you will see that the Outputs are thru an onchip transistor.

Normal Collector to Emitter diode junction drop is about .5V

Side note about needing 7 resistors per 7 segment led units
There are plenty of suppliers such as Digikey & others that
see Resistor Packs with 1 common lead & multiple output leads
This way when you lay out your pcb things get much simpler
and it takes up a lot less room .. also all the resistors are then
very close in value.

Posted: Mon May 08, 2006 3:55 am
by MrAl
Hello again,

Ken, most transistors have between 0.2 and 0.5 volts drop
when they are turned 'on'.

Robert:
Oh ok, sounds cool.
BTW, i made a mistake in assuming the way the dp segments were
brought out from the display LED package.
If there is only one dp lead on the LED package (tie multiples together)
then we only need one logic comparator chip. The 'equals' output
drives the dp segment (though a resistor). Depending on what
kind of switch you use you may or may not need another 8 line
to 3 line encoder ic chip for the switch.