I built a new osc cct:
I seem to be picking up HUGE amounts of hum from the heater . Is the case being earthed from the xtal end properly?
10/05/2008
At the Moorabin Hamfest I picked up an old Hewlett-Packard Oven and Xtal.
These were apparently from a frequency meter. - Must have been a real old
frequency meter as the extract of the cct the seller gave me shows a 6AU6
oscillator! As well as the oven, the xtal and the cct extract, I got the
original transformer from the HP whatever which was useful.
The oven - a Z235 appears to be a simple thermostat type and the target temp is 65 deg C
judging from the thermometer on the side:
The original oscillator was a Pierce cct showing two variables for the caps in
the phase shift network. I suspect these were air variables used for low drift vs
temp? Or perhaps to adjust the cct to just to the point of oscillation.
Anyway , given the xtal was designed for a cct with these values, I figured it
would be a good place to start.
14/05/2008
WHOA!
The oven looks just like the one in the HP 101E on this
page
although this apparently used a 1Mc xtal.
The manual for a HP100c shows a different style oven, but the cct for the
oscillator is quite similar with with 2x100P variables
The HP-100c is rated at 10PPM . it also state several hors are required for the
temperature to stabilize.
HP-100D (once again similar oscillator cct) rated at 2ppm over a week, when operated continuously with a 30 day run
in period.
This page has a
zoom-able view below a 100E so we can see the variable caps . Also the blue blob
there is identical to whats still hanging of my oven...
524D... The internal oscillator of the 524C and 524D is stable within 5 parts in 10e8 per week.
524D uses a 1MHz Oscillator ref http://www.hpmemory.org/wb_pages/wall_b_page_01.htm
17/05/2008
The oven temp doesn't seem to reach 65deg, well at least the thermometer doesn't show that. Although I haven't
operated the oven pins down yet../
What cut is th xtal . possibly CT or BT ?? freq seems to go up with xtal heating from ambient
Need to build AGC cct for oscillator.
18/05/2008
Heres a pic of the 100Kc xtal in the base of the oven. That is an octal base its in :
I built a new osc cct:
I seem to be picking up HUGE amounts of hum from the heater . Is the case being earthed from the
xtal end properly?
19/05/2008
I have thrown together a meacham bridge oscillator using an opamp and one of the 100kc xtals out of the Larkspur military gear:
The meacham bridge oscillator has always fascinated me and an op-amps seems to be the obvious way to implement it rather than a pair of transformers and a pentode. Varying R1
allows the level to be adjusted almost nothing and the lamp provides an AGC function to keep the level constant.
All the parts
values are just what I grabbed at the time apart from the R1,R2 , the lamp and the xtal - and even these were just guesses. The bridge stabilizes when the impedances conform to
Z1/R1 = R2/X1
At resonance the impedance of X1 is at its lowest.
The only problem is that the xtal freq is way low, 99.9407Kc , thats 600ppm out!
20/05/2008
I measured the resistance of a 50mA 12V 'grain of wheat' lamp (DSE P8130) and got these charts:
The last chart should correspond to Fig 5 in "Meacham, L.A The Bridge Stabilized Oscillator" Proceedings of the Ire Oct 1938
Clearly Mr Meahcam's lamp was higher resistance than what I have been able to dig up thus far. However this doesn't seem to be effecting my
oscillator that much. I have, however bought a few 12V 50ma (P8140 from DSE) that seem to have the same characteristics, for when incandescent
lamps are all banned!!
I do note that the amplitude of the oscillator can be effected by heating the bulb of the lamp with a soldering iron. Because a xtals freq changes
with power level, this will probably cause temp related variations if the lamp isn't in the oven. Heating the leads of the bridge resistors with
the iron had a slight effect , but not nearly as much as the lamp, and i suspect the resistors got a LOT bigger temperature change.
While I think of it the lamp doesn't glow at all. It runs in a very low voltage., <200mV RMS.
I just realized that the voltage dividers in the bridge are wrong. 1k2 for R2 results in about 1/2 voltage across the Xtal. Also adjusted R1 for
balance and increased C1 to a 1uF film cap to reduce phase shift. I think the op amp isn't too happy driving the low impedance of the bulb and R1
in series ( must be about 50 ohms? ) I might have to move the bridge balance point back to keep things happy
On the Larkspur xtal being off-freq....
C13 (Larkspur) Calibrator extracted from C13 manuals of ralphk - 'check out the welder of Doom'
Anyway, looks like a electron coupled pierce , so why is the xtal freq out ????
I tried the 2nd one of these I have and it runs at 99.9304 kHz rather than the
99.9407 kHz of the first one...
21/05/2008
I grabbed some old 50V 20mA Bar telephone exchange bulbs, as I figure they will have higher resistance :
The problem with these T6.8 bulbs is that they are a bit 'knock' sensitive. That is there resistance changes subject to slight knocks. I think the filament coils touch or something. This destabilizes the oscillator.
22/05/2008
Heres a closeup of the t6.8 bulb showing the two metal supports:
After a bit more testing, this particular bulb may be faulty? Its resistance changes DRAMATICALLY with small knocks, the changes seems to be
in an upwards direction . Some 60V 3W (50ma) bulbs don't seem to have this problem, althoough their resistance is about half that of the
T6.8
The Larkspur xtals seem to have a motional resistance(?) of about 1.2Khz at resonance. The HP xtal is less, but im'm not sure by how much.
I have tried the HP xtal in this cct and it runs at 99.9559KHz (well thats what the beanch, only slightly stabilised counter says)
23/05/2008 Despite Ref 1 stating the the xtal in a pierce oscillator is in series mode, it appears it isn't. As an xtals parallel resonance is typically slightly higher than its series resonance, the Meacham bridge series oscillation point is too low...
24/05/2008
The final Meacham cct is :
For use with a 32Khz watch xtal C2/C3 are increased to 10nF and R1 to 3.8 k variable (2k trim +1.8K).
The 32.768 xtal oscillats at about 3.2764Khz and its resitance seems to be about 10K
29/05/2008
Adding somewhere between 10 and 20pF in series with the xtal seems to move the oscillation point to 100Khz . BUT the oscillator is inclined to take off at the third
overtone. Interestingly, it seems to oscillate at 100kHz AND the third overtone (slightly more than 3 x fundamental) as the two wave forms arn't harmonically related when
I look on the CRO, ie. they are phase shifting relative to each other. Replacing the 1uF cap with a series 5.6mH and 470pf, resonant at about 100kHz, suppresses the
third overtone
leaving a 100Khz fundamental.
The problem is, has the ~15pF cap in series with the xtal reduced the overall stability, and does not exact tuning of the series trap cause a phase shift that will effect
the freq?. The original meacham article shows tuned ccts either side of the bridge and indicates they have negligble effects when tuned properly . I'll have to check
this out. However, it does seem the meacham may be back on track.....
31/05/2008
I reduced the loaded 'Q' of the series resonant cct in the feedback by going to 1mH and ~ 3000pF. This should make the phase shift
at changes in freq less severe(I'm assuming the tuned ct will drift, not the XTAL!)
The frequency of the third overtone seemed to be closer in percentage terms to 200Khz than the fundamental was to 100Khz. Also I
seem to remember reading overtone oscillators are less effected by circit variations - eg harder to pull (Ref2 3.2.3.3 & 3.2.4.1). So I tought I'd try to
force the third overtone and see how it goes. I changed the series trap to 200Khz - aprox 1000uH in series with 215pF . The xtals C
is probably 30PF judging from the series tuning cap . 30p at 300khz = 17,676ohms, 15p at 100 Khz = 106,060 ohms
1/06/2008 I made a small oven using an old eddystone box and a constatnt current mosfet cct as a heater with 12V it pumps in about 6.6Watts. I will use this to try to find the crytals turning point. With an ambient of 16.6deg C (2.896V on LM336) I can heat up to 46deg C (3.190V) with no insulation. With insulation I get about the same?
2/06/2008 With 24V and some insulation I get 77 Deg( 3.500V). This is hot enough! With full insulation I'm guessing I'd get to 100 deg C easily.
4/06/2008
I did a plot of one of the 100khz Larkspur xtals vs tmp. This is fromthe cooldown cycle, which was slower
This chart is probably not too accurate as the temperature was changing quite fast so the detla-temp effects are probably distorting the results, but the turnover
point is rather low at about 37 degrees.
15/06/2008
I plotted F vs T on cool-down of the HP xtal in th HP oven at 100khz . The temp sensor was in free air inside
the oven.
I think the higest temp graph points on the chart may be out a bit as the max temp of the oven was 76 deg
which is close to where the turnover point appears to be. This turnover is significantly higher than the 65
deg quoted for the oven. Once again this could be rate of temperature change effects as well as the actual
temperature.
Ref 1 Crystal OScillator Circuits, Mathys RJ ,Wiley 1983
Ref 2 Design of Crystal and Other Harmonic Oscillators, Parzen B Wiley
Misc Temp links:
http://www.hpmemory.org/wa_pages/wall_a_page_11.htm
http://www.hparchive.com/