On Fri, 19 Sep 2008 09:41:51 -0700, Jeff Liebermann
put finger to keyboard and composed:

On Fri, 19 Sep 2008 17:34:40 +1000, Franc Zabkar
wrote:

On Thu, 18 Sep 2008 13:59:55 -0700, Jeff Liebermann
put finger to keyboard and composed:

On Thu, 18 Sep 2008 22:15:27 +0300, "Joe"
wrote:

To clarify the circuit, the postive side of the speaker receives 17 volts
thru one diode and a resistor and the negative side is connected via
transistor to ground and the transistor is driven by 4001 IC.

What value resitor?
What's the part number on the speaker?

I just happen to have the manual for a similar Heathkit GC-1005
"Electronic Clock" handy. No specs on the speaker (401-163). The
circuit is similar in that the speaker has one lead going to the
collector of an MPS-A20 and the other to 18.5VAC (not DC) through a
diode and 1200 uf to ground for a DC voltage of 25.7VDC. No resistor
in series.

So, let's do the math. My guess(tm) is that 1 watt will be
sufficiently loud to wake the dead. Yours has about a 15VDC swing.
Assuming a 50% duty cycle and a pure sine wave (yeah sure):
Power = E^2 / R
1 watt = 15^2 / R
R = 225 ohms.
So, it's probably a high impedance speaker of some sorts. I have the
digital clock and an LRC meter and could probably measure the
impedance. If you can't get the info any other way, bug me and I'll
rip it apart.

The GC-1107 supplies the speaker via a rectified 13VAC source and 150
ohm 1/2W resistor. That's a DC supply of 18V.

I believe maximum power will be transferred to the speaker if it has a
resistance equivalent to that of the series resistor, ie 150 ohm. In
this case, when the transistor is turned on, the current will be
18/300 = 60mA. Assuming a square wave signal with a duty cycle of 50%,
the power dissipated in the speaker will then be 9V x 60mA x 0.5 =
270mW.

Therefore I'm guessing that the speaker has an impedance/resistance of
at least 150 ohms and a power rating of at least 0.5W. If the
speaker's impedance were any less, then the dissipation in the
resistor would increase.

If we accept that the speaker should dissipate less than 270mW in both
clock circuits, then in in the GC-1005 case we have ...

Power(max) = 0.27 = 25 x 25 x 0.5 / R(min)

So R(min) = 1157 ohms

- Franc Zabkar

I like your calcs better than mine. That suggests that Heathkit would
have used something like a high impedance earphone "speaker" in the
design. That's possible and probably would work quite well.

However, I tore apart a similar Heathkit clock, with the identical p/n
speaker, and measured 41.5 ohms DC resistance. Adding the inductance,
that I didn't bother measuring, will produce about 50 to 60 ohms
impedance. Obviously, this is not the optimum power transfer design,
but that's what Heathkit apparently used. The important thing is that
a common 4, 8, or 16 ohm speaker will NOT work.

Neither of the clock circuits makes any sense to me.

Your clock has a 25VDC supply which, at a 50% duty cycle, would cause
a 41.5 ohm speaker to dissipate 7.5W.

In the OP's clock circuit, a 41.5 ohm speaker would cause the 150 ohm
1/2W resistor to dissipate ...

(18/191.5 x 0.5) x (18 x 150/191.5) = 0.66W

I can only assume that the speaker's impedance at the operating
frequency of the alarm is *much* higher than one would expect. For
example, at 1kHz an impedance of 100 ohms would require an inductance
of 16mH. I measured the inductance of an 8 ohm 1W 3" speaker on my
DMM's 2mH scale as 0.08mH and about 0.5mH on the 2mH and 20mH scales.
I could hear a high pitched tone on the 2mH range (1kHz ?) and a low
pitch on the 20mH range (100Hz ?).

This site appears to be dedicated to saving and restoring old Heathkit
clocks:
http://www.decodesystems.com/heathkit-clocks.html

Here is some info on the MK5017 clock chip that was used in the
GC-1005:
http://www.decodesystems.com/mk5017.html

The MK5017's Tone output is shown driving a 2N3904 transistor
connected to a 17VDC supply through a transformer-coupled 8 ohm
speaker:
http://www.decodesystems.com/mk5017-2.gif

The transformer is spec'ed as "2K/8R".

I'm really clutching at straws now, but is it possible that the
Heathkit speaker has a built-in 2K/8R transformer ??? Does it have the
usual permanent magnet? Would it make sense to have a stationery 2K
winding and an 8R moving coil on a soft iron former ???

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

"Franc Zabkar" wrote in message
news
On Fri, 19 Sep 2008 09:41:51 -0700, Jeff Liebermann
put finger to keyboard and composed:

On Fri, 19 Sep 2008 17:34:40 +1000, Franc Zabkar
wrote:

On Thu, 18 Sep 2008 13:59:55 -0700, Jeff Liebermann
put finger to keyboard and composed:

On Thu, 18 Sep 2008 22:15:27 +0300, "Joe"
wrote:

To clarify the circuit, the postive side of the speaker receives 17
volts
thru one diode and a resistor and the negative side is connected via
transistor to ground and the transistor is driven by 4001 IC.

What value resitor?
What's the part number on the speaker?

I just happen to have the manual for a similar Heathkit GC-1005
"Electronic Clock" handy. No specs on the speaker (401-163). The
circuit is similar in that the speaker has one lead going to the
collector of an MPS-A20 and the other to 18.5VAC (not DC) through a
diode and 1200 uf to ground for a DC voltage of 25.7VDC. No resistor
in series.

So, let's do the math. My guess(tm) is that 1 watt will be
sufficiently loud to wake the dead. Yours has about a 15VDC swing.
Assuming a 50% duty cycle and a pure sine wave (yeah sure):
Power = E^2 / R
1 watt = 15^2 / R
R = 225 ohms.
So, it's probably a high impedance speaker of some sorts. I have the
digital clock and an LRC meter and could probably measure the
impedance. If you can't get the info any other way, bug me and I'll
rip it apart.

The GC-1107 supplies the speaker via a rectified 13VAC source and 150
ohm 1/2W resistor. That's a DC supply of 18V.

I believe maximum power will be transferred to the speaker if it has a
resistance equivalent to that of the series resistor, ie 150 ohm. In
this case, when the transistor is turned on, the current will be
18/300 = 60mA. Assuming a square wave signal with a duty cycle of 50%,
the power dissipated in the speaker will then be 9V x 60mA x 0.5 =
270mW.

Therefore I'm guessing that the speaker has an impedance/resistance of
at least 150 ohms and a power rating of at least 0.5W. If the
speaker's impedance were any less, then the dissipation in the
resistor would increase.

If we accept that the speaker should dissipate less than 270mW in both
clock circuits, then in in the GC-1005 case we have ...

Power(max) = 0.27 = 25 x 25 x 0.5 / R(min)

So R(min) = 1157 ohms

- Franc Zabkar

I like your calcs better than mine. That suggests that Heathkit would
have used something like a high impedance earphone "speaker" in the
design. That's possible and probably would work quite well.

However, I tore apart a similar Heathkit clock, with the identical p/n
speaker, and measured 41.5 ohms DC resistance. Adding the inductance,
that I didn't bother measuring, will produce about 50 to 60 ohms
impedance. Obviously, this is not the optimum power transfer design,
but that's what Heathkit apparently used. The important thing is that
a common 4, 8, or 16 ohm speaker will NOT work.

Neither of the clock circuits makes any sense to me.

Your clock has a 25VDC supply which, at a 50% duty cycle, would cause
a 41.5 ohm speaker to dissipate 7.5W.

In the OP's clock circuit, a 41.5 ohm speaker would cause the 150 ohm
1/2W resistor to dissipate ...

(18/191.5 x 0.5) x (18 x 150/191.5) = 0.66W

I can only assume that the speaker's impedance at the operating
frequency of the alarm is *much* higher than one would expect. For
example, at 1kHz an impedance of 100 ohms would require an inductance
of 16mH. I measured the inductance of an 8 ohm 1W 3" speaker on my
DMM's 2mH scale as 0.08mH and about 0.5mH on the 2mH and 20mH scales.
I could hear a high pitched tone on the 2mH range (1kHz ?) and a low
pitch on the 20mH range (100Hz ?).

This site appears to be dedicated to saving and restoring old Heathkit
clocks:
http://www.decodesystems.com/heathkit-clocks.html

Here is some info on the MK5017 clock chip that was used in the
GC-1005:
http://www.decodesystems.com/mk5017.html

The MK5017's Tone output is shown driving a 2N3904 transistor
connected to a 17VDC supply through a transformer-coupled 8 ohm
speaker:
http://www.decodesystems.com/mk5017-2.gif

The transformer is spec'ed as "2K/8R".

I'm really clutching at straws now, but is it possible that the
Heathkit speaker has a built-in 2K/8R transformer ??? Does it have the
usual permanent magnet? Would it make sense to have a stationery 2K
winding and an 8R moving coil on a soft iron former ???

Finding a way to reduce the duty cycle might be a means to get away with a
more commonly available lower impedance speaker.

On Sun, 21 Sep 2008 15:54:26 +0100, "ian field"
put finger to keyboard and composed:

Finding a way to reduce the duty cycle might be a means to get away with a
more commonly available lower impedance speaker.

AFAICS, at 25VDC a 40 ohm speaker would draw 600mA when the transistor
switches on, regardless of duty cycle, assuming the transistor has a
high enough gain at whatever base drive the IC provides. An MPS-A20
transistor is only rated for 100mA.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

"Franc Zabkar" wrote in message
...
On Sun, 21 Sep 2008 15:54:26 +0100, "ian field"
put finger to keyboard and composed:

Finding a way to reduce the duty cycle might be a means to get away with a
more commonly available lower impedance speaker.

AFAICS, at 25VDC a 40 ohm speaker would draw 600mA when the transistor
switches on, regardless of duty cycle, assuming the transistor has a
high enough gain at whatever base drive the IC provides. An MPS-A20
transistor is only rated for 100mA.

As the speech coil is an inductor it will exhibit a linear rising current
waveform from the initial application of voltage, obviously that will be
shorter for a lower impedance/inductance speech coil so the pulse width must
be made shorter to ensure it terminates before the inductor saturates.

On Sun, 21 Sep 2008 21:24:09 +0100, "ian field"
put finger to keyboard and composed:


"Franc Zabkar" wrote in message
.. .
On Sun, 21 Sep 2008 15:54:26 +0100, "ian field"
put finger to keyboard and composed:

Finding a way to reduce the duty cycle might be a means to get away with a
more commonly available lower impedance speaker.

AFAICS, at 25VDC a 40 ohm speaker would draw 600mA when the transistor
switches on, regardless of duty cycle, assuming the transistor has a
high enough gain at whatever base drive the IC provides. An MPS-A20
transistor is only rated for 100mA.

As the speech coil is an inductor it will exhibit a linear rising current
waveform from the initial application of voltage, obviously that will be
shorter for a lower impedance/inductance speech coil so the pulse width must
be made shorter to ensure it terminates before the inductor saturates.

If my inductance measurements make any sense, then the time constant
of the speaker would be L/R = 0.5mH/40R = 12us, or 80uH/40R = 2us.
That's a very short pulse.

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.

On Sun, 21 Sep 2008 21:24:09 +0100, "ian field"
put finger to keyboard and composed:


"Franc Zabkar" wrote in message
.. .
On Sun, 21 Sep 2008 15:54:26 +0100, "ian field"
put finger to keyboard and composed:

Finding a way to reduce the duty cycle might be a means to get away with a
more commonly available lower impedance speaker.

AFAICS, at 25VDC a 40 ohm speaker would draw 600mA when the transistor
switches on, regardless of duty cycle, assuming the transistor has a
high enough gain at whatever base drive the IC provides. An MPS-A20
transistor is only rated for 100mA.

As the speech coil is an inductor it will exhibit a linear rising current
waveform from the initial application of voltage, obviously that will be
shorter for a lower impedance/inductance speech coil so the pulse width must
be made shorter to ensure it terminates before the inductor saturates.

I used Bob Parker's ESR meter to measure various 8 ohm speakers. AIUI,
the meter delivers 8us pulses to the device being tested.

FWIW, here are my results:

10W/8R - 35 ohms
2W/8R - 22 ohms
0.5W/8R - 15 ohms
20W/8R/20kHz tweeter - 18 ohms

- Franc Zabkar
--
Please remove one 'i' from my address when replying by email.