Light-sensitive XOSC

[Andrew Holme]

I've just built a 1.5 GHz synthesizer using a 10 MHz CFPT 9301 TCVCXO from
IQD (Farnell order code 127 2080) as the reference. I've just powered it up
for the first time and was disappointed to see 100 Hz sidebands at
around -40 dBc on the 1.5 GHz output. That equates to a peak deviation of 2
Hz at 1.5 GHz or 13 mHz on the reference. I have lots of low noise LDOs on
the board powering all the different parts independently, so I was surprised
that 100 Hz was getting in there; and I was about to try powering off
batteries when, whilst viewing the output in FM Demod mode on my SA, I
realised the XOSC was light-sensitive! I turned off the (halogen bulb) room
light in my lab - and no more 100 Hz sidebands! Anyone heard of that: a
light-sensitive XOSC !?

TIA

 

[Andrew Holme]

"Andrew Holme" wrote in message
I've just built a 1.5 GHz synthesizer using a 10 MHz CFPT 9301 TCVCXO from
IQD (Farnell order code 127 2080) as the reference. I've just powered it up
for the first time and was disappointed to see 100 Hz sidebands at
around -40 dBc on the 1.5 GHz output. That equates to a peak deviation of 2
Hz at 1.5 GHz or 13 mHz on the reference. I have lots of low noise LDOs on
the board powering all the different parts independently, so I was surprised
that 100 Hz was getting in there; and I was about to try powering off
batteries when, whilst viewing the output in FM Demod mode on my SA, I
realised the XOSC was light-sensitive! I turned off the (halogen bulb) room
light in my lab - and no more 100 Hz sidebands! Anyone heard of that: a
light-sensitive XOSC !?

TIA

More info: it's not sensitive to visible light, only infra-red! And I can
detect the TV remote!

 

[Uwe Hercksen]

I've just built a 1.5 GHz synthesizer using a 10 MHz CFPT 9301 TCVCXO
from IQD (Farnell order code 127 2080) as the reference. I've just
powered it up for the first time and was disappointed to see 100 Hz
sidebands at around -40 dBc on the 1.5 GHz output. That equates to a
peak deviation of 2 Hz at 1.5 GHz or 13 mHz on the reference. I have
lots of low noise LDOs on the board powering all the different parts
independently, so I was surprised that 100 Hz was getting in there; and
I was about to try powering off batteries when, whilst viewing the
output in FM Demod mode on my SA, I realised the XOSC was
light-sensitive! I turned off the (halogen bulb) room light in my lab -
and no more 100 Hz sidebands! Anyone heard of that: a light-sensitive
XOSC !?

Hello,

did you try to block the emission from the halogen bulb to the XOSC? Is
there any change of the supply voltage of the XOSC when switching the
bulb on and off?

Bye

 

[Jasen Betts]

I recall that older-style ceramic-package transistors and op amps
could definitely be light sensitive. Even the ceramic frit around a
metal can's leads could admit enough light to cause odd effects,
line-noise hum and harmonics being among them.

This doesn't usually seem to be a problem with plastic-package parts,
but I suppose it's possible that there might be enough IR leakage
through some encapsulation plastics to read the semiconductor
junctions and have some effect.

Glass encapsulated diodes like 1N914, 1N4148, and the SMD variants of
same are light sensitive.

 

[Tauno Voipio]

Hello,

did you try to block the emission from the halogen bulb to the XOSC? Is
there any change of the supply voltage of the XOSC when switching the
bulb on and off?

An incadescent bulb will show the AC component of the supply
as slight undulations in the light output.

50 years ago, I built a speech transfer system using a car bulb
with DC feed and the output transformer of a radio in series as
the transmitter, and a scraped OC71 with an amplifier as the receiver.
To my surprise, the audio came through pretty well.

Without the DC bias, the output contains a full-wave rectified
copy of the line voltage.

 

[[email protected]]

Glass encapsulated diodes like 1N914, 1N4148, and the SMD variants of
same are light sensitive.

I used those diodes for limiting the input voltage range from 0 to Vcc
in a very high impedance circuit. Moving my hand around the circuit
caused a huge output swing due to those protecting diodes not shadowed
by my hand.

On an other case, a microwave source using a low frequency (<100 MHz)
crystal and a few multiplier stages was very stable when left alone.
However, when someone looked at the circuit with an "evil eye", the
frequency started to swing at a few seconds interval. Later on, it
became obvious that the person breathing into the circuit caused some
rapid temperature and hence frequency changes .

 

[Tim Williams]

An incadescent bulb will show the AC component of the supply
as slight undulations in the light output.

50 years ago, I built a speech transfer system using a car bulb
with DC feed and the output transformer of a radio in series as
the transmitter, and a scraped OC71 with an amplifier as the receiver.
To my surprise, the audio came through pretty well.

Without the DC bias, the output contains a full-wave rectified
copy of the line voltage.

Also some pretty dramatic lowpass, I would think -- part of the reason
incandescents work surprisingly well at that is because the filament cools
off by radiation so much faster at high temperatures (thermal conductivity
due to radiation goes as (T_abs)^3!). At yellow hot (say, 1200K), the
temp change per watt is 8 times greater than at invisible heat (say,
600K).

Tim

 

[Tauno Voipio]

Also some pretty dramatic lowpass, I would think -- part of the reason
incandescents work surprisingly well at that is because the filament cools
off by radiation so much faster at high temperatures (thermal conductivity
due to radiation goes as (T_abs)^3!). At yellow hot (say, 1200K), the
temp change per watt is 8 times greater than at invisible heat (say,
600K).

Tim

To my surprise, it did not sound as muffled as a low-pass
should make it. Maybe we were so far in the skirt that
all the frequencies were equally attenuated.