Don wrote:

54-215 (full HP part number 1854-0215) is the ubiquitous 2N3904.

53-20 (full HP part number 1853-0020) is 2N3702.  2N4403 is an
equivalent if 3702 is hard to find.

Also, note that the HP schematics of the 10544 have some errors that
were (as far as I can tell) never corrected by HP.  I posted a corrected
and annotated schematic to Didier's site (ko4bb.com).  Here's a link:

http://www.ko4bb.com/getsimple/index.php?id=download&file=HP_Agilent/HP_10544_Crystal_Oven_Oscillator/HP_10544A_schematic_further_corrected_and_annotated.pdf

Best regards,

Charles

Bill wrote:

For the following discussion, you need to refer to the corrected
schematic I mentioned in my last post.  If you are looking at the HP
schematic, you will wonder how the hell it works (and it wouldn't, if HP
actually built them as they drew it).

The 10544 oven control circuit uses pulse-width modulation to control
the heater in a "bang-bang" manner rather than a smooth proportional
manner.  UJT A3Q3 forms a relaxation oscillator (along with A3C1 and
A3R10), with a period of ~250uS (frequency ~4kHz) and a voltage span
from ~0v to ~8v.

This positive-going ramp is applied to the base of Darlington A3Q2
(MPSA12), which is 1/2 of a differential pair current switch along with
A3Q1 (2N3904).

The thermistor and associated op-amp circuitry set a threshold voltage
between ground and about 7v at the base of Q1.  After the relaxation
oscillator resets to ~0v, current flows through A3Q1 and A3R8, pulling
the base of Darlington A3Q4 negative and turning it on to saturation.
The collector of A3Q4 therefore applies essentially the full oven heater
supply voltage from Pin 14 (nominally 24v) to the high side of the heater.

The oscillator voltage ramps positive toward its ~8v maximum (the
trigger point of UJT A3Q3).  When the emitter of A3Q2, which is two
diode drops below the ramp voltage, exceeds the voltage at the emitter
of Q1 (as set by the thermistor and A3U1), Q2 steals the current that
has been flowing in A3Q1 and turns Darlington switch A3Q4 off, which
interrupts the current flowing through the heater.  Some time later
(about 250uS after the previous reset), the oscillator voltage reaches
the trigger point of the UJT and it resets the voltage on A3C1 to ~0v
and the cycle begins again.

Thus, every ~250uS the heater is on for a time (set by the thermistor
circuitry) and off for the remainder of the ~250uS.  This switching
action can be seen at the "Oven Monitor", Pin 11 (but note that the
instrument may have a capacitor to ground on the mother card side of the
oven monitor, to integrate the switching waveform for use by the
instrument's health monitor).

Best regards,

Charles

Thank you, Charles.

What a clever way to minimize the power dissipation in Q4 with the
components of the day.

A switching regulator without the steep (and noisy) transients of
today's switchers.

Bill Hawkins

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Charles
Steinmetz
Sent: Monday, January 08, 2018 10:21 PM
To: time-nuts@febo.com
Subject: Re: [time-nuts] AM vs PM noise of signal sources

Bill wrote:

For the following discussion, you need to refer to the corrected
schematic I mentioned in my last post.  If you are looking at the HP
schematic, you will wonder how the hell it works (and it wouldn't, if HP
actually built them as they drew it).

The 10544 oven control circuit uses pulse-width modulation to control
the heater in a "bang-bang" manner rather than a smooth proportional
manner.  UJT A3Q3 forms a relaxation oscillator (along with A3C1 and
A3R10), with a period of ~250uS (frequency ~4kHz) and a voltage span
from ~0v to ~8v.

This positive-going ramp is applied to the base of Darlington A3Q2
(MPSA12), which is 1/2 of a differential pair current switch along with
A3Q1 (2N3904).

The thermistor and associated op-amp circuitry set a threshold voltage
between ground and about 7v at the base of Q1.  After the relaxation
oscillator resets to ~0v, current flows through A3Q1 and A3R8, pulling
the base of Darlington A3Q4 negative and turning it on to saturation.
The collector of A3Q4 therefore applies essentially the full oven heater
supply voltage from Pin 14 (nominally 24v) to the high side of the
heater.

The oscillator voltage ramps positive toward its ~8v maximum (the
trigger point of UJT A3Q3).  When the emitter of A3Q2, which is two
diode drops below the ramp voltage, exceeds the voltage at the emitter
of Q1 (as set by the thermistor and A3U1), Q2 steals the current that
has been flowing in A3Q1 and turns Darlington switch A3Q4 off, which
interrupts the current flowing through the heater.  Some time later
(about 250uS after the previous reset), the oscillator voltage reaches
the trigger point of the UJT and it resets the voltage on A3C1 to ~0v
and the cycle begins again.

Thus, every ~250uS the heater is on for a time (set by the thermistor
circuitry) and off for the remainder of the ~250uS.  This switching
action can be seen at the "Oven Monitor", Pin 11 (but note that the
instrument may have a capacitor to ground on the mother card side of the
oven monitor, to integrate the switching waveform for use by the
instrument's health monitor).

Best regards,

Charles

Bill wrote:

Well, don't get too giddy about how quiet the scheme is <smile>.  HP
warned that if you use the same 12v (nominal) power supply for the oven
control and crystal oscillator circuits, you need to add extra filtering
to the oscillator power line to avoid injecting switching noise into the
sensitive oscillator circuitry.

I always use separate voltage regulators for the oven control and
crystal oscillator circuits when I build a 10544 or 10811 into a
project, and I have modified HP instruments by adding an additional
regulator for the oscillator section when they had 10544s installed, to
good effect.  Even then, a small amount of switching noise is induced
into the oscillator section internally to the 10544.

That said, it is a simple and robust system that performs quite well in
general.

Best regards,

Charles

Hello to the group. The Stable32 book arrived yesterday from Lulu and its
as expected. Good quality at a reasonable cost.
It also arrived far faster then the web site would have you believe.
Regards
Paul
WB8TSL

On Fri, Jan 5, 2018 at 10:13 AM, Dr. David Kirkby <
drkirkby@kirkbymicrowave.co.uk> wrote:

On the Rb unit I am sending you, I just use female standard jumpers on the pins.  You’ll need +24 at about 1A and +5 at 400ma or less.  Be careful if and when you open it as the caps I used were a little tall and they might get hung up on the case as you slide the board out.  There is a 9-pin header inside but I never saw any activity on it.  Overall the units are pretty stable.

Have fun!

Jerry

Jerry Hancock
jerry@hanler.com
(415) 215-3779