digital clock with variable speed

[Michael Schwab]

Hi,
I am looking for a circuit (or any other ideas) for building a digital clock
(hh:mm:ss) where I can manually adjust the speed of the clock. (from 24h in
1 s to 1s in 24h). Everybody I have talked to so far seems to think that
digital clocks come in components where it is impossible to change the
speed.
Thanks,
Michael

 

[Active8]

Simply move the clock at a velocity near the speed of light.
it will slow down.

but he wants it to speed up, also. perhaps a velocity gyrator would do
the trick.

mike

 

[Frank Bemelman]

BTW, i still don't know why anyone would do this except as a prank or to
get rid of someone fast. i saw a clock at a guesthouse in germany that
ran backwards. that was pretty cool. hmmm... i could've swore it was ten
minutes later last time i looked. actually, i think it ran forward, i.e.
counterclockwise with the roman numerals increasing in the same
direction.

That's the hairdressers' clock. When viewed in the mirror, it looks
okay.

I have been thinking about a clock that runs a bit faster between
09.00AM and 05.00 PM. Lets say 10 minutes. During the night it slows
down so 24 hours is still 24 hours.

Think about that. The perfect clock to give your boss as a present.

 

[Spehro Pefhany]

Hi,
I am looking for a circuit (or any other ideas) for building a digital clock
(hh:mm:ss) where I can manually adjust the speed of the clock. (from 24h in
1 s to 1s in 24h). Everybody I have talked to so far seems to think that
digital clocks come in components where it is impossible to change the
speed.

One could, of course, build such a "clock" with a microcontroller, but
I don't think that it will prove possible to modify a commercial clock
to run that far off the design speed.


Best regards,
Spehro Pefhany

 

[Spehro Pefhany]

What about those cheap LED alarm clock that use the 50Hz or 60Hz as
a timebase... those would be relative easy to modify!

Don't they mux (IIRC, some biplex) the digits? Imagine them running at
1/(24*3600) the multiplex rate... at the other end, the poor chip
might not accept 5 MHz+ input.

Best regards,
Spehro Pefhany

 

[Frank Bemelman]

Don't they mux (IIRC, some biplex) the digits? Imagine them running at
1/(24*3600) the multiplex rate... at the other end, the poor chip
might not accept 5 MHz+ input.

I would think the multiplexing runs on some independant free running
RC oscillator, and the clock itself uses 50Hz or 60Hz through a chain
of dividers, but I could be all wrong.

 

[Spehro Pefhany]

I would think the multiplexing runs on some independant free running
RC oscillator, and the clock itself uses 50Hz or 60Hz through a chain
of dividers, but I could be all wrong.

The ones that have biplexed displays use a center-tapped transformer
amd two diodes to multiplex the display without any anode drivers.
They switch the segment drivers based on the timebase input (driven
from one phase).

I checked a modern (Sanyo) clock chip and they still work this way,
though they do have an internal free-running oscillator to handle
other things.

Best regards,
Spehro Pefhany

 

[Gary Tait]

BTW, i still don't know why anyone would do this except as a prank or to
get rid of someone fast. i saw a clock at a guesthouse in germany that
ran backwards. that was pretty cool. hmmm... i could've swore it was ten
minutes later last time i looked. actually, i think it ran forward, i.e.
counterclockwise with the roman numerals increasing in the same
direction.

br,
mike

IMO, it might be as easy as wiring the coil backwards, or flipping the
magnet over, in the quartz mechanism.

Back in '85 or so, I saw a promo clock for Back To The Future, which
ran backwards (was in one of the local video rental emporia).
It

 

[Gary Tait]

I would think the multiplexing runs on some independant free running
RC oscillator, and the clock itself uses 50Hz or 60Hz through a chain
of dividers, but I could be all wrong.

They use the AC line to mux. With a CT transformer, whent the one side
goes pos, a diode supplies common for one half the display, the same
when the other side goes pos for the other side of the display.

 

[Frank Bemelman]

really? you jest, no? the babe that cuts my hair (and all the other
places i've been) need to know this. i thought it was just a novelty.
what's wrong with turning around to face the clock or just asking, you
know?

Impress the babe and buy her one:
http://www.fakecrap.com/products/backward_clock.html

imagine the wives whining. "er, honey. we need one of those hairdresser
clocks so I don't have to turn around when i'm doing my makeup."

now we have a practical application! but it also needs to be slow in the
morning so you can come in late

That's the idea

 

[Active8]

thanks. i already impressed her once, but she's married. i'm sure i can
use a backward clock or some fake parking tickets, though, or maybe i'll
find someone else that i want to impress. the one i saw in Germany was a
real nice wooden piece of work. no cuckoo though, IIRC. the clock, that
is. the girls were flesh ;-)

br,
mike

 

[Active8]

note the spec.

If you can still get hold of one of the single-chip clock circuits they have
a 50/60Hz clock input, which is mains derived. Instead of taking that from a
tap on the transformer, use a clock generated from a 555 timer and have it
variable. I did one of these many years ago as a trainee for an instructor
who was a model train enthusiast - his club wanted to scale time to run
their choo-choo's between stations to schedule.

The MM5314 is one such chip, it appears to still be available here and
there, such as at:
http://safe.spsp.net/cgi-bin/debco/MM5314.html

However if you have problems getting one of these chips it may be easier to
use a PIC.

now all he has to do to make the thing register 24hrs/sec (one end of
his spec) is get that chip to run at 5.184MHz and at the other extreme
(register 1s / 24hr) 694.45 uHz or a period of 1.44 ks or 24min. the
latter is doable with a one shot or mv, but 5.184MHz? eek! i just
checked the limit for the TLC555 as bench tested. it stopped at about
2.4Mhz and that was the fastest one tested. the original 555 topped out
at 180kHz. figure in component tolerances and such... forget about
trying to do the whole range with one cap and pot...

i think the PIC approach is best.

br,
mike

 

[Active8]

The ones that have biplexed displays use a center-tapped transformer
amd two diodes to multiplex the display without any anode drivers.
They switch the segment drivers based on the timebase input (driven
from one phase).

like gary said. i couldn't understand why you'd want to do that. but i
see now that if you output 14 segments and biplex the readout, it works.
you said "biplex" and that cleared it all up. thanks.

mike

 

[Active8]

[snip]

They use the AC line to mux. With a CT transformer, whent the one side
goes pos, a diode supplies common for one half the display, the same
when the other side goes pos for the other side of the display.

i'm not sure that's the same kind of MUX we've been thinking - i.e.,
output the required segments for one digit and energize the common for
that digit. change the segment outs for the next digit and pulse that
digit's common. that way you only need outputs for 7 segments and the
commons for the digits.

why in hell would anyone design the MUX scheme you just described? it's
totally gone.

mike

never mind. i see the utility of "biplexing" now. i was thinking 7 seg
outs, not 14. but if that's the case, they're trading 4 anode drivers
for 7 more segment drivers. so i'm back to "why in hell..."
something i'm not visualizing. maybe a spec sheet will help. i'm just
curious.

br,
mike

 

[Spehro Pefhany]

never mind. i see the utility of "biplexing" now. i was thinking 7 seg
outs, not 14. but if that's the case, they're trading 4 anode drivers

We'd better substitute "digit" for "anode" here to be more general and
accurate.

for 7 more segment drivers. so i'm back to "why in hell..."

It's actually 13 outputs for the segments and zero for the digits vs.
11, so two more outputs total. (plus colon, alarm announciator(s) and
AM/PM LEDs). 3.5 drivers per digit except for the MS digit which
requires only 2.5 if you parallel segments A and D and leave out F
(the MS digit only has to display blank, 1 or 2 for 24-hour time).

Okay, those segments might be a little less evenly lit due to Vf
variations, but only between 10:00 PM and midnight for those who
choose 24-hour time.

something i'm not visualizing. maybe a spec sheet will help. i'm just
curious.

Do a web search for a clock chip datasheet if you're really curious.

16 pins total (out of a 30-pin shrink-DIP) to drive a 4-digit time
display plus two colon LEDs (paralleled), two separate alarm
annunciators, and AM and PM indicators (6 additional LEDs total).

The biplex input signal comes for free since it's required for the
timebase anyhow.

If you drove that conventionally with standard multiplexed displays
you'd need 7 segment drivers plus 5 digit drivers or 8 segment drivers
plus 4 digit drivers. And an extra center-tapped transformer winding.
So far it doesn't look good, right?

With the conventional way, the digit drivers would each have to be
powerful enough to handle as many as 7 or 8 segments on simultaneously
at 25% duty cycle (maybe 150mA each) vs. the biplex segment drivers
which only need handle around 10mA each for the same brightness. But
the cool thing is that you don't need a bulky filter capacitor for the
150mA power supply, and you don't need any regulation for any of the
LED current (which roughly halves the size and weight of the power
supply transformer) and you don't need to worry about visually
annoying beat frequencies between power supply ripple and the mux
frequency, since it's synchronous. AND since clocks are often combined
with radios, you don't need to worry about LARGE fast-switching
currents to the display causing RFI on the AM frequencies. This design
transitions gently with the mains waveform. Of course the chip pinout
is designed so that a single-sided board is possible/easy.

In reference to the OP's question, the min/max frequency at the
50/60Hz pin on the part I'm looking at is rated at 1Hz ~ 2000Hz. Below
1Hz it switches to the RC oscillator (external R & C) as backup. The
2000Hz limit may be created by the 4800Hz (nominal) RC oscillator and
some logic as a method of further filtering the timebase input from
transients on the mains (they suggest an RC input filter and there's a
ST input).

Best regards,
Spehro Pefhany

 

[Active8]

True, I ignored the extremes of the spec coz I can't see an application,

me either. i sometimes overlook specs cuz i skim the post too quickly.
like that deal where i suggested a dishwasher could be turned into a
cement mixer. read the words, followed the thread, made my comment, and
finally woke up and realized/remembered they were NOT talking about a
clothes washer, duh! i'm having a clothes washer issue and i guess it
was a predominant thought. they got me thinking of the maytag man
commercial and all those boy's uniforms just make them look like another
delivery boy.

heh, heh... "i don't care WHAT it's called. just fix it!"

but hey, thanks. i like checking out possibilities and you got me to
check out the 555's ( a limited use pos, IMHO ) limits. nice to know my
database is half way useful.

br,
mike