I want to covert an old Dana Lighting Aerolite alarm clock-radio with built in folding halogen lamp to a deaf alarm clock. My plan includes sticking a relay across the lamp switch and tapping the alarm signal at the buzzer position on the switch to trigger the relay coil through a MOSFET.
Can you hunt up the datasheet for the MSM5096B alarm clock chip? I've been unable to find it.

For the clock control chip it would either be some microcontroller with a proprietary firmware, so datasheet won't help you to get the functional description.
Or it would be a some clone of one of the common LED clock chip like LM8560 or similar (they differ in display configuration - the most recent and most commonly used I remember the LM8560 has a two common cathode "duplex" display, the earlier chips I have seen jut in datasheet used single common cathode)
What backup battery it uses? 3V or 9V?
What display it uses? LED or LCD? Is it multiplexed?

A piece of circuit reverse engineered around the buzzer will help, or at least a board photo. The buzzer may be passive, driven by some pulse series from the chip, or active DC driven with internal oscillator. In the case of passive buzzer, some simple peak rectifier circuit needs to be hacked to keep MOSFET or solid state relay or whatever input steady.

Do you have an oscilloscope?

My late father used the Radio Shack kit to connect to my alarm clock to the kit to activate the bed vibrator.  I'm currently using the Sonic Boom as shown at My Nightstand at my new Home.

I dont think the chip matters that much. I'd do the following :

 - With the buzzer either in circuit or removed (choose how you are going to make it in the end), add a rectifier bridge across it, then few uF capacitor on te DC output. To be nice to the chip which is driving it all (and not short its output with the capacitor which pulls high current to charge), add a few 10's Ohms resistor in series with the input of the rectifier bridge

 - Evaluate what is the voltage in the capacitor. Connect a multimeter across the capacitor, and see how high it charges when the alarm goes off

 - Connect an optoisolator to the capacitor (so the alarm will flash the optocoupler), with suitable resistor for the voltage you measured

 - Connect the other side of the optoisolator to a circuit that will shape the flash seen in the optocoupler to a nice timed pulse. It can be a monostable 555 circuit (look it up, adapt the component values to your needs) or any other circuit which does the same thing

 - Connect the output of the timed pulse to a suitable bipolar transistor or FET and then your relay



I expect that the circuit doesn't apply constant DC across the buzzer when it is silent. (It definitely won't with a magnetic buzzer, but it might or might not if the buzzer is a piezo disc. If there is constant DC there, you may have to add another capacitor (few 10's...100's uF) in series with the rectifier bridge input in order to block the DC

Well, this all is over-complicated for a person who probably even does not own an oscilloscope.

As a rough hack, I'd try some reasonable sensitive opto-TRIAC like Sharp S102S01 or S202S01 and just connect it's LED input in series with some suitable current-limiting resistor say, 330 Ohms by luck across the buzzer, be it piezo or electromagnetic. If this does not fire, reverse the polarity. TRIAC output could drive high voltage halogen or HF halogen supply directly, I'd be a bit cautious with magnetic 12V transformer, but probably still OK.

In these things (judging what I have at home) is just one single transformer with one secondary feeding the clock/radio part and then the 12V high current one feeding the halogen. All the switching is then on the low voltage side. So triac would have way too high voltage drop when starting from 12V.

I think the only way to proceed is to trace the buzzer and radio power control circuit and start from there.
To control the radio, there already is some kind of detector/rectifier, as the radio does need a permanent power. And the clock IC very likely provides the same signals all the time (assume one of the clock ICs, not the microcontroller), they are only afterwards blocked by the mode switches.
So depend on how that is arranged, you may either tap on the existing control (if the radio/buzzer switch is after that) or replicate the radio switch circuit to control the relay.

The LM8560 generates only "buzzer" alarm signal by a single PMOS open drain output (it uses PMOS logic technology, so the ground of the clock circuit is positive), so there is a kind of filter circuit which turns that into continuously (when the buzzer is buzzing) available power for the radio.
The is a second "sleep" output, but that works only for the "sleep" function (press the button, it runs the radio for set amount of time and tyen shuts it down), these are combined in the radio power control circuit, but you won't need that (unless you want tye "sleep" function for the lamp asc well). This "sleep" usually goes around the alarm mode switch directly to the radio.

In fact, S102/202 optos I mentioned have forward drop of less than 1.5V @2A, so still may be OK even for flashing 12V halogen directly, especially accounting for somewhat constant-current behavior of small power halogen transformers.

To invent an optimized circuit, some hint of actual clock schematics is needed, @lightsofpahrump was asked for, but he prefers to keep silence. Ok, his choice.

If there is a radio as alarm function, an easiest way will be just to hook a relay (contact rating powerful enough for a halogen) across radio power lines.

forward drop of 1.5V @2A

...means power dissipation of 3W, not that small...

There also exist photo MOSFETs or even SSR gate drivers like VOM1271 (an example, there are ton of similar products, many even in more DIY convenient packages), you take a pair of MOSFETs like IRFZ44 or so, connect Sources together, Gates together, between Gate and Source connect the output of the VOM, the drains then form the power switch. You get about 100mOhm, that means about 0.2V drop, so barely 0.4W total at 2A.

I know. Photovoltaic generators are specialized stuff, hard to get and so unproportionally expensive for what they do. Better to get just a finished 60V high current optoMOS SSR. There is a lot of Chinese ones now for cheap. But optoMOS will need a clean DC control current, while two-stage optoTRIAC like S102/202 will trigger on pretty much any crud above the threshold, including unsmoothed buzzer signal )

S102/202 etc are quite large, easy to dissipate ~3W even without a heatsink, but if doubt could be bolted to anything metal.

In these things (judging what I have at home) is just one single transformer with one secondary feeding the clock/radio part and then the 12V high current one feeding the halogen. All the switching is then on the low voltage side. So triac would have way too high voltage drop when starting from 12V.

I think the only way to proceed is to trace the buzzer and radio power control circuit and start from there.
To control the radio, there already is some kind of detector/rectifier, as the radio does need a permanent power. And the clock IC very likely provides the same signals all the time (assume one of the clock ICs, not the microcontroller), they are only afterwards blocked by the mode switches.
So depend on how that is arranged, you may either tap on the existing control (if the radio/buzzer switch is after that) or replicate the radio switch circuit to control the relay.

The LM8560 generates only "buzzer" alarm signal by a single PMOS open drain output (it uses PMOS logic technology, so the ground of the clock circuit is positive), so there is a kind of filter circuit which turns that into continuously (when the buzzer is buzzing) available power for the radio.
The is a second "sleep" output, but that works only for the "sleep" function (press the button, it runs the radio for set amount of time and tyen shuts it down), these are combined in the radio power control circuit, but you won't need that (unless you want tye "sleep" function for the lamp asc well). This "sleep" usually goes around the alarm mode switch directly to the radio.

You are spot on about the transformer!
I want to turn it on with a relay driven by a MOSFET and I found the alarm trigger pin on the schematic. I could tap that, because judging by the schematic, the alarm is disabled by a link in the alarm/radio mode switch sending a signal to the chip unless the switch is in AUTO(radio alarm) or ALARM(buzzer) position, not in ON(radio on independently of alarm) or OFF(alarm disabled and radio off).

In fact, S102/202 optos I mentioned have forward drop of less than 1.5V @2A, so still may be OK even for flashing 12V halogen directly, especially accounting for somewhat constant-current behavior of small power halogen transformers.

To invent an optimized circuit, some hint of actual clock schematics is needed, @lightsofpahrump was asked for, but he prefers to keep silence. Ok, his choice.

The Alarm Output pin is going through the MODE selector to either the buzzer or the transistor that switches on power to the radio IC depending on position and there is a turning-off-the alarm link in the switch I explained in my last post that shuts down the alarm mode in the chip if the mode switch is not set to an alarm position. I want to tap the output of the chip directly, triggering the relay through a MOSFET to avoid overloading the IC.

It looks like LM8560 style output, so with the output generating the 900Hz signal, gated by 2Hz to generate the beeping...
In the radio control path there will be some kind of filter, turning the 900Hz into a solid steady voltage. You may need something similar for the relay. The time constant depends whether you want the light solid ON (then it needs to cover more than 0.5s) or flashing (cover just the 2ms).
Example is drawn in the page 4 of the LM8560 datasheet (assume it is its clone or something with similar output).

I want to tap the output of the chip directly, triggering the relay through a MOSFET to avoid overloading the IC.

Okay, so we understand your *intentions* but what is the actual question?

Chip that is used seems to be obscure on the internet so no luck. You *have* to do some reverse engineering. Power supply voltage, polarity. Draw the circuit around alarm pin output, the switch, and buzzer. That's all rather simple.