Dim LED Display - Old LM8560 Mains Driven Clock

[t1d]

Background: I am a still-learning hobbyist, but I have very nice lab equipment. PSUs, Freq Gens, Freq Counters, Scopes, Mutlimeters, Soldering Stations, etc... So, if you can tell me what to test, I can tell you what I find...

I am wanting to repair this clock for the learning experience. Yes, I know a new clock would be the better investment, so, let's stick with discussing the repair, please and thank you.

This is a very minimal circuit... I have only started my investigations, but I am only finding...
- a 120 to 4.7VAC-0-4.7VAC transformer to a couple of rectifying diodes... No voltage regulator
- the LM8560 takes its timing from the 60Hz mains AC. The is no external oscillator can.
- various other diodes, resistors, caps, inductor, etc., but no transistors. I replaced the 330uF e-cap with a 470uF (what I had in stock,) just as a matter of maintenance. The 330uF tested as good. The LM8560 data sheet spec's a 470uF e-cap. This did not improve the display brightness.
- I do have 10.87=11.00VDC fluctuating on the LM8560 supply pin, with 0.47VAC. This chip runs on -15V to +0.3V, so I probably had my meter probes inverted.
The minimal components seems to be in keeping with the DS common circuit diagram. See attached.

I think this chip is meant to drive a common cathode display. So, what might cause the display to dim, equally, over all the digits?

Thank you for your help.

[retiredfeline]

It's a special kind of common cathode display where half the segments are driven on each half of the mains cycle. Biplexed if you like. So you may not have any luck finding a replacement unless it's for/from a similar kind of clock. So you will have to repair it.

I picked up a similar cheap clock from e-waste once, checked that it was working and gave it away. I think it was deliberately dim because it was for bedside use. Or maybe the displays had aged.

[t1d]

Thanks for your reply!

Here are some pictures. I am still investigating, but it looks like they skimped on everything. Look at page 7 of the DS. There should be transistors on the switches. They skipped those, for example.

Look at the return for LED cathode. It looks like the DS has two switching diodes on the display returns. But, the board only has one glass diode. Maybe they combined the returns into the one diode? And, maybe that diode is leaky? Would that cause dimming?

The glass diode does test as good, but I only checked it in-circuit, which I know is not legitimate. If you think it being leaky would dim the LED, I will pull it and test it out-of-circuit.

Thanks for your help!

[retiredfeline]

I think that glass diode is not for the display current. Maybe the reverse protection for the mains sensing circuit. Those 1N type rectifiers are the display cathode returns. They are very visible on the RHS of the board.

You could check the cathode limiting resistors but it's unlikely that both would rise in value so you would see half the segments dimmer than others. You could reduce them in value to increase the brightness perhaps at some cost to the life of the display.

[james_s]

Are you sure there's anything wrong with it? Old LEDs were a lot dimmer than modern ones, it should be bright enough that you can read it in normal room light but it won't be blinding like a modern LED display.

[t1d]

I think that glass diode is not for the display current. Maybe the reverse protection for the mains sensing circuit. Those 1N type rectifiers are the display cathode returns. They are very visible on the RHS of the board.

The two 1N diodes below the red-black-red wires of the transformer secondary, which are nested between the resistors, are the rectifying diodes. I have not checked the connection of the other 1N, but I will.

You could check the cathode limiting resistors but it's unlikely that both would rise in value so you would see half the segments dimmer than others.

Agreed.

You could reduce them in value to increase the brightness perhaps at some cost to the life of the display.

I fear I am losing the display, so that may be a solution to squeeze a little more life out of it. But, I will check everything else, first.

Thank you so much for your input.

[t1d]

Are you sure there's anything wrong with it? Old LEDs were a lot dimmer than modern ones, it should be bright enough that you can read it in normal room light but it won't be blinding like a modern LED display.

That is a fair question, but I am sure that it has been dimming, significantly, over the last month, or two. I can barely see it at all, in the daytime.

[floobydust]

So far it's not making sense to me. All the LED's shouldn't go dim, B and C segments age the fastest and E or G ages the least, I forgot.
Used to actually repair these things in a shop, for manufacturer's warranty. Flat rate $5 if it had a radio. They are very low cost.
Common failures were an LED here and there for a dead segment. The clock IC does experience mains surges which can damage it, and rural electricity can be high mains voltage which ends up with too much DC for the IC. Most had bad counters, on fast set they would skip and jump 1.2.3...6.9.10.11 kind of thing.

You can try illuminate an LED using a multimeter to judge how their brightness is. Not a multimeter that puts out over say 3V on diode-test, to damage the IC. That will test if the display is to blame.
*The digit bank dropping resistors, at the far end, are chipped and likely damaged. 47R or 470R? Did you nail them with a screwdriver.

I had an alarm clock using this IC with blue LEDs and it was super bright, with no dimmer, I couldn't use it and went back to a 1970's Randix cube clock radio with VFD I still use lol.

[t1d]

So far it's not making sense to me. All the LED's shouldn't go dim, B and C segments age the fastest and E or G ages the least, I forgot.

Not making sense = agreed. For them all to dim, there would have to be a cause that is common across all of them. That lends itself to the common cathode side, I would think...

rural electricity can be high mains voltage which ends up with too much DC for the IC.

I don't think we have those kinds of power problems, were I am. The label states that the power supply is to be 120VAC. Meaning, it is not an older unit that is only rated to 110VAC/etc.

You can try illuminate an LED using a multimeter to judge how their brightness is. Not a multimeter that puts out over say 3V on diode-test, to damage the IC. That will test if the display is to blame.

I like that idea. I have a couple of ways to get at it... I have a Siglent DMM which has an adjustable voltage threshold on the diode function. Or, I could just use my PSU at some voltage and with the current limited. What voltage and current values do you suggest?

*The digit bank dropping resistors, at the far end, are chipped and likely damaged. 47R or 470R? Did you nail them with a screwdriver.

No, I am very confident that I did not nick them with a screwdriver. Best I can tell, those marks are not chips, but are smudges - of what I do not know. I have only tested those two resistors in-circuit, but they are reading just 5R/5 Ohms, each.

[timenutgoblin]

Background: I am a still-learning hobbyist, but I have very nice lab equipment. PSUs, Freq Gens, Freq Counters, Scopes, Mutlimeters, Soldering Stations, etc... So, if you can tell me what to test, I can tell you what I find...

I am wanting to repair this clock for the learning experience. Yes, I know a new clock would be the better investment, so, let's stick with discussing the repair, please and thank you.

This is a very minimal circuit... I have only started my investigations, but I am only finding...
- a 120 to 4.7VAC-0-4.7VAC transformer to a couple of rectifying diodes... No voltage regulator
- the LM8560 takes its timing from the 60Hz mains AC. The is no external oscillator can.
- various other diodes, resistors, caps, inductor, etc., but no transistors. I replaced the 330uF e-cap with a 470uF (what I had in stock,) just as a matter of maintenance. The 330uF tested as good. The LM8560 data sheet spec's a 470uF e-cap. This did not improve the display brightness.
- I do have 10.87=11.00VDC fluctuating on the LM8560 supply pin, with 0.47VAC. This chip runs on -15V to +0.3V, so I probably had my meter probes inverted.
The minimal components seems to be in keeping with the DS common circuit diagram. See attached.

I think this chip is meant to drive a common cathode display. So, what might cause the display to dim, equally, over all the digits?

Thank you for your help.

Is the display consistently dim regardless of the time being displayed? Is the display brighter at 1:11 compared to 10:58 (12 hour time) or 22:58 (24 hour time) ?

The transformer has self-regulation when it's being loaded. The greater the load, the lower the output voltage on the secondary winding.

According to the datasheet for the LM8560 it requires 15VDC nominal, but can operate at lower voltages hence the battery backup feature (9V).

Are the rectifier diodes working properly? Of course they are...

[t1d]

Is the display consistently dim regardless of the time being displayed? Is the display brighter at 1:11 compared to 10:58 (12 hour time) or 22:58 (24 hour time) ?

The unit is disassembled, presently, so operating the switches may prove difficult, but I will give it try, tomorrow.

The transformer has self-regulation when it's being loaded. The greater the load, the lower the output voltage on the secondary winding.

I'm not sure I know about the effect you are speaking of... I do know about voltage "droop."

According to the datasheet for the LM8560 it requires 15VDC nominal, but can operate at lower voltages hence the battery backup feature (9V).

From a previous post = "- I do have 10.87=11.00VDC fluctuating on the LM8560 supply pin, with 0.47VAC. This chip runs on -15V to +0.3V, so I probably had my meter probes inverted."
I did not have my probes inverted; this chip can run on positive voltage, as well. See pg7 of the DS. With the GND Probe on Pin 20 and the Positive Probe on Pin 15, the DMM reads a positive voltage.

Are the rectifier diodes working properly? Of course they are...

Given that we have the expected voltage at the chip and that the diodes test okay in-circuit, I think that they are operating. Secondary = 4.7-0-4.7 = 9.4VAC - 2 x 0.7Diode Drop = 8VAC x 1.414Factor = 11.31VDC
[/quote]

As I mentioned, I checked the Cathode resistors in-circuit. I went back to double-check that the smudges were not breaks; they are not. But, the paint rings did not make any sense to the values I had read in-circuit. So, I pulled them.

I checked them out of circuit, with the Kelvin/Four Wire feature = ~4.7R, each. I looked at the paint rings, under my microscope. The first, yellow ring was not visible to the naked eye. Yellow_4 - Magenta_7 - Gold_/10 - Gold_5% So, they are good.

[retiredfeline]

Given that we have the expected voltage at the chip and that the diodes test okay in-circuit, I think that they are operating. Secondary = 4.7-0-4.7 = 9.4VAC - 2 x 0.7Diode Drop = 8VAC x 1.414Factor = 11.31VDC

Actually you do not have a sqrt(2) factor since the cathode supply is rectified but not smoothed because no capacitor. This is another cost saving measure. Also avoids current peaks tthrough the transformer and rectifiers. Each half bank of LEDs is actually getting half-sine waves and would emit a 60Hz visual hum if only your eyes could see it.

The positive/negative voltage supply thing is just a matter of perspective. With PMOS technology, usually voltages are referenced to the most postitive voltage so are usually stated as negative.

[t1d]

Actually you do not have a sqrt(2) factor since the cathode supply is rectified but not smoothed because no capacitor. This is another cost saving measure. Also avoids current peaks tthrough the transformer and rectifiers. Each half bank of LEDs is actually getting half-sine waves and would emit a 60Hz visual hum if only your eyes could see it.

The positive/negative voltage supply thing is just a matter of perspective. With PMOS technology, usually voltages are referenced to the most postitive voltage so are usually stated as negative.

Thanks, retiredfeline. I haven't slept, so I will have to let this information soak, until I can process it fully. lol

[t1d]

I re-installed the two cathode resistors and reflowed the display ribbon cable, on the controller PCB. Then, I considered the illumination, again, at typical room light levels.

There is a minor difference in brightness, between the characters, left to right, as you would expect, with age. However, it is just too dim to be seen across the room.

I have determined that the display is just plain used up and that's okay. I have checked what there is to check and I learned some things, thanks to you good folks. I think we can call this one "Concluded", but not "Solved."

Time to buy a new clock. (There's a joke in there, somewhere.) Or, maybe I will use the clock drive, from this clock, and build a giant character display with three LEDs per character segment. I guess that would require driving some transistors with the clock to power the multiple LEDs... Or, I have been wanting to DIY a clock with a PIC microcontroller. I think I have all the needed parts. The problem is that I am not much of a coder. Maybe I will find a complete project, on the internet? There are lots of fun possibilities.

Thank you, all, again, for your help and support. The EEVBlog Community is the best forum I know of, regardless of topic. You make it that way!

[retiredfeline]

Instead of discrete LEDs to make a segment, you might consider what I'm thinking, and that is to buy those LEDs in the shape of filaments that are used in retro looking bulbs. You can make quite tall digits with them. Don't know how bright they are but one filament model takes 12V and 45mA. I don't think you want that glowing in your bedroom.

[james_s]

You could just buy some 7 segment LED displays and solder them up on a piece of perfboard to replace that one. Or if you want a challenge, you could scrap the LED dies off the board in the existing display and solder some 0402 SMT LEDs in their place.

[tooki]

Are you sure there's anything wrong with it? Old LEDs were a lot dimmer than modern ones, it should be bright enough that you can read it in normal room light but it won't be blinding like a modern LED display.

The snap domes make me think it’s not that old. The one board could be indicating a 1994 date code. Red LEDs were a perfectly mature technology by then. 1970s LEDs were dim, but not 80s/90s. (Sure, we now have LEDs way brighter than that, but indicator LEDs aren’t meant to be blindingly bright.)

[tooki]

Instead of discrete LEDs to make a segment, you might consider what I'm thinking, and that is to buy those LEDs in the shape of filaments that are used in retro looking bulbs. You can make quite tall digits with them. Don't know how bright they are but one filament model takes 12V and 45mA. I don't think you want that glowing in your bedroom.

I just found a vendor on AliExpress this week that sells 3V LED filaments, from 20mm all the way to 300mm!! I am SO going to build a clock with some of those!!!

[floobydust]

I was going to use some too for a big display thermometer, but I thought all are high voltage series strings 45-70V. For 3V they must run a wire through and all be in parallel. Hmmm.

[tooki]

The last time I’d searched was a few years ago, and indeed all I found then were high voltage filaments.

[Runco990]

I just looked at that LED clock.  I have a few very much from that era, and man.... they are DIM!!!  REALLY DIM.  And work just as they should.  We are spoiled by today's super bright LED's, but you could not see the old one's in bright room light.  They were pretty terrible.

[james_s]

Well remember that they were mostly meant as bedside clocks, you probably don't want a blinding beacon on your nightstand while you're trying to sleep. I have a Heathkit digital clock I built back in the 80s that has a light sensor that automatically dims the display at night but I don't recall ever seeing that feature on a store bought clock. I don't know why because it isn't complicated.

[floobydust]

The LM8560 LED's are multiplexed duplex at mains frequency, so each pair of digits is on 1/2 time driven by sine-wave, and then that pair is further reduced in perceived brightness by whatever this IC mux operates at. The only RC clock is 900Hz supposedly for the AC fail and alarm beeper, I could not see what it uses for muxing the pair.
Point is the LED's may be extra dim due to the oddball (=cheap) multiplexing scheme. The flicker would be annoying.

I added a dimmer to my VFD alarm clock and it did not work out. A front-facing phototransistor can only see what's in front of it, which is my dark furniture/bed so it always registers dark. I'm going to change it and make it point up at the ceiling which is white.

[james_s]

Now that you mention it, I have a Heathkit VFD clock at our cabin and it has a light sensor too. It's a CdS cell pointed forward at a slight upward angle, seems to work really well.

[retiredfeline]

The LM8560 LED's are multiplexed duplex at mains frequency, so each pair of digits is on 1/2 time driven by sine-wave, and then that pair is further reduced in perceived brightness by whatever this IC mux operates at. The only RC clock is 900Hz supposedly for the AC fail and alarm beeper, I could not see what it uses for muxing the pair.
Point is the LED's may be extra dim due to the oddball (=cheap) multiplexing scheme. The flicker would be annoying.

The multiplexing is at mains frequency coordinated by the sense pin which is also the timebase. No mutiplexing on top of that. Nothing particularly odd about the scheme, just increase the current to about 2x (and perhaps a bit more because the conduction angle isn't the full 180°) and it's a standard multiplexing (biplexing) scheme. Maybe the 60Hz flicker might be noticeable in the corner of your eye but obviously they can get away with it.