HP 105B frequency standard (00105-6013) oscillator teardown

[ArthurDent]

I thought some of you might be interested in what the inside of the HP 105 frequency standard oscillator looks like. The crystal oven assembly (HP 00105-6013) has been used in a few different HP standards from the mid 1960s onward for several years and it is still a pretty good crystal standard. I picked one up cheap and it seemed to work just fine but when I tied to fine tune it using the EFC there was no change in frequency. After looking at the schematic and double checking everything and even substituting it into a good HP 105B, there was no doubt but that the varicap had failed open.

It took me a while to decide to open the oscillator because I thought I might break something and make matters far worse, but I finally got the tools out. It turned out that it is fairly easy to open it up and get to the innards. I took a number of photos that I’m posting here that may help others who wish to delve inside this oscillator. Having taken it apart once I found there are a few things I would do differently that would make disassembly a lot easier. First, take photos before you disconnect any wires or remove any parts so you won’t have to guess where they go when you reassemble the unit.

On the end with the connectors, remove the 4 screws on the sides and the 3 screws on the end and remove the end plate. After taking photos, remove the wires on the connectors that go directly from connectors to the circuit board and prevent you from removing the circuit board. Loosen the 2 small torx screws on the knob and remove it. There are 2 small purple wires that come up through holes in the board and are soldered to terminal posts- unsolder them or you won’t be able to remove the board.

Carefully lift the board up and move it to one side.  The 2 purple wires go into a hard foam cover for the inner oven and must carefully be pried up to remove. It is a tight fit and may be hard to remove so be careful. This cover is also the holder for the sensor that the 2 purple wires go to. Take another photo of the wiring going to the inner oven. Unsolder the 2 10K resistors and the 2 inductors from the top of the inner oven. Unscrew but do not remove the 2 hex head bolts from inner oven.  Carefully pull on the 2 bolts to lift the oscillator out of the inner oven. The 2 bolts are held in place by 2 clips on the bottom of the assembly.

If you check the photos you will see how the oscillator is assembled in ‘levels’ that each has different parts of the circuits with the crystal on the bottom level. The level separators/circuit boards have yellow tape around them but in the photos I have removed one piece of tape to allow access to the parts. Replacing any parts could be a big problem because there is almost no space between levels. I managed to get the varicap out but it wasn’t easy. Now I have to find a suitable replacement and reassemble the unit. Although the parts list showed an HP part number, the varicap had Motorola markings on it and I haven’t found any info on it yet. Here are the photos.

[doktor pyta]

Thanks.
P.S. what a coincidence: http://bbs.38hot.net/thread-838238-1-1.html

[5065AGuru]

Arthur,

I have several dead disassembled oscillators. If you give me your address I'll remove and send you a varactor for free.

Merry Christmas

Corby Dawson

[ArthurDent]

Thanks.
P.S. what a coincidence: http://bbs.38hot.net/thread-838238-1-1.html

Very interesting! Great minds think along the same lines, right? 

The 2 oscillators I have are serial numbers 1231 and 1271 so are pretty close together in when they were made. The one I opened for repair seems to be different than the one you have in parts layout and the crystal can. It does look like the one you have is in really great shape. One thing is no matter how they look the batteries in your unit are probably questionable (at best) after all these years. The HP 105B  I got was cheap because the original batteries in the unit had shorted out and burned some resistors in the charging circuit. I found that the original charging for the 2500mah batteries I used for replacement was too high so I reworked the charging circuit so the battery pack wouldn't overheat or overcharge. The specs say 8 hours for battery power and the new batteries and charging circuit give me 10.5 hours and the unit remains cool.

I replaced the 32V zener in the power supply with a 30V zener and replaced the 1.6 and 5 ohm resistors in the charging circuit that had burned up with 2 10 ohm resistors mounted on the rear panel. It might take longer to recharge to a lower level but it should last longer and run cooler.

[jfphp]

You can find the schematics diagrams in the service chapter of the 5065A first edition, 5245M, 5248M... manuals and the parts list is in the parts chapter but there is no layout drawings of the different stages. Main problem is the aging of the quartz after about 25 years of continuous running (!) : you will exceed the tuning range of the screwdriver adjust capacitor and playing with the other caps of the oscillator will degrade the phase noise.

[ArthurDent]

You can find the schematics diagrams in the service chapter of the 5065A first edition, 5245M, 5248M... manuals and the parts list is in the parts chapter but there is no layout drawings of the different stages. Main problem is the aging of the quartz after about 25 years of continuous running (!) : you will exceed the tuning range of the screwdriver adjust capacitor and playing with the other caps of the oscillator will degrade the phase noise.

Not necessarily true. The present mechanical adjustment left on my oscillator that was made in 1967 (over 50 years ago) is from +6hz to -9hz so it will take a very very long time for it to age out of range. 

[Leo Bodnar]

What does ADEV for 105A/105B look like for 1 to 10,000 seconds?
Leo

[5065AGuru]

Leo,

I just finished plotting 20 of the 0105-6013 oscillators used in the early 105  as well as a few other instruments.

Most were in the low to mid parts in 10-12th out to 1000 Sec, although a few did  in the high 10-13ths.

I have found that selecting from 10811 oscillators will usually give better performance at least for the really good ones.

My reference Quartz units are a selected 10811, a selected FS1200, and an HP 106A.

Cheers,


Corby

[bob91343]

Generally, a varicap is just a diode that has been measured for capacitance.  I have tested 1N4004 for instance and it has the same characteristic.  I would not hesitate installing a small diode for frequency control.

To be safe, I'd first measure its capacitance with various bias voltages.  If the curve seems about right, I'd use it.

Most good bridges include provision for bias, and also use small enough excitation so as not to cause the diode capacitance to shift much, or ever forward bias.

[Leo Bodnar]

Very interesting, I wonder how phase noise looks like for these.  It would be a good exercise to redesign the sustaining amplifier using modern transistors and techniques that were not available at the time, while keeping the crystal and environmental design intact.
Do you have a way of measuring crystal parameters directly?
What is the working temperature of the oven? 105A/B were designed in late 1960's, while SC-cut crystals were not produced until early-mid 1970's.
Cheers
Leo

Leo,

I just finished plotting 20 of the 0105-6013 oscillators used in the early 105  as well as a few other instruments.

Most were in the low to mid parts in 10-12th out to 1000 Sec, although a few did  in the high 10-13ths.

I have found that selecting from 10811 oscillators will usually give better performance at least for the really good ones.

My reference Quartz units are a selected 10811, a selected FS1200, and an HP 106A.

Cheers,

Corby

[German_EE]

Designed and built when the letters H and P meant something

Thanks for this!

[ArthurDent]

... What is the working temperature of the oven? 105A/B were designed in late 1960's, while SC-cut crystals were not produced until early-mid 1970's. ...

The temperature marked on the crystal is also marked on the outside of the assembly. Here's the one I'm working on.

[Leo Bodnar]

I got HP 105A that has a full scale reading (100+) on the meter in "OVEN" position even after a good hour warm up.

The output frequency is 4.9999981MHz

Does this sound like a faulty thermal sensor and overheating oven/crystal?

Leo

[5065AGuru]

Leo,

I'd give it a couple hours as the coarse oven might be out of whack.

The coarse oven is just used during warmup to shorten the time.

The unit will work fine without it but just takes a while to warm up.

If after extended warmup you can adjust to 5Mhz exactly I'd say it's working.

Then if your oven reading is still 100+ I think there is a resistor to select to put it on scale.

Cheers,

Corby

[Leo Bodnar]

Thanks, Corby,
I am replacing tantalum caps on the heater controller PCB with ceramic X7S ones (they are not capaciatnce-critical anyway.)
The thermal fuse for ~110V fast heater was blown as well.
I'll give it another go.

I don't know the oven temperature, it does not seem to be an SC-cut since it's dated early 1970's.  If it works, I will sweep the oven temperature to find the turning point.

Leo

[5065AGuru]

Leo,

If the coarse heater fuse was blown I'd just leave it out and live with an extended warmup. Since most of these stay on 24/7 that usually is not a problem.

You are right, it in not an SC cut. Most probably AT which you can verify by plotting the warmup curve.

Cheers,

Corby

[Leo Bodnar]

I stuck a thermocouple inside the foam near the Aluminium can containing XO - under the temperature controller.
I am getting 60-61C which is a bit too low I think.  There is no turnover temperature recorded on th outside of the enclosure and I don't want to rip it all apart just yet.

Temperature controller has some weird bridge oscillator design that I have not seen for ages.  I'd like to model it before I do any more modifications.
One of my suspicions is that the heater is possibly damaged. Its resistance is 100R - I am not sure what its nominal resistance is.

Leo

[5065AGuru]

Temp is marked on the crystal. Two that I have opened show 68.6C and 67.5C.
I'll see if I can measure the two heater windings resistance.

Cheers,

Corby

[5065AGuru]

Both oven windings (assuming both fuses are good) can be measured without opening the unit.
The fast warmup heater (two pins on the side) reads around 240 Ohms.
The normal heater (read between the M and +24V pins) reads around 102 Ohms.
So if your normal heater was the one that read 100 Ohms it OK.
There is a small hole on the two pin end the allows you to reach the pot that adjust the temperature.

Cheers,

Corby

[Leo Bodnar]

Thanks, Corby, for taking the time to measure your units.

Looks like the heater is not faulty then. I suspected it could have been two 100R in parallel and one winding open-circuited.

The NTC seems fine - it drops its resistance with increasing temperature.

The problem then might be either in the faulty or misaligned 200R pot (it was set for just 10R) or components drift.
I am still curious to understand why HP used Wien bridge oscillator as a temperature controller.  The only reason I can come up with is safety - if it stops oscillating then heater cuts off (final stage is AC coupled.)

Leo

[tkamiya]

My recently acquired HP105B also stays at full scale in OVEN setting.  I was told by someone the spec for zener diode changed between earlier and later oscillator modules, so that it needs to be checked.  I have not done that.  It's working fine as is.  Sure enough, there are signs someone has replaced components in power supply module.  It's very possible component values are off.

[ArthurDent]

As I mentioned in post #3 I made changes in the charging circuit in my 105B. I replaced the original shorted batteries with smaller sub-C cells. The new batteries powered the 105B nicely but improvements in battery chemistry in the last 50 years created charging problems for me and the new batteries were also overheating because of a too high charge rate. To correct this I replaced the zener diode with a lower voltage one and replaced the two resistors that determine the charging current. It has been working fine since then. Reread my previous post to see what I did.

[tkamiya]

Oh, that was YOU!  I knew I saw it somewhere..... and I didn't realize it was in this thread.

I am thinking of doing exactly the same. 

[uncle_bob]

Leo,

If the coarse heater fuse was blown I'd just leave it out and live with an extended warmup. Since most of these stay on 24/7 that usually is not a problem.

You are right, it in not an SC cut. Most probably AT which you can verify by plotting the warmup curve.

Cheers,

Corby

Hi

Just to keep things straight ... The crystal is most likely a BT cut. That was HP's "secret weapon" back before the SC.

Bob

[Xtremexp]

Hello forum members,

Here I am with HP 105 unit. I picked up as I love to collect such engineering marvels. it was last calibrated in 1998 by hp, the calibration seal was still intact. I powered up, as people reported, the meter goes above on OVEN, 18.7V and 5MHz switch position, the meter shows ~30 on both 1MHz and 100KHz.

Only 5 MHz output is working. the frequency is 5MHz (I do not have precision device to measure it to full accuracy of the 105 device) but 1MHz and 100KHz output is not there.

I opened up the unit, it has ovenized oscillator 10811A (which is newer than the pictures above posted, i think). I am wondering what would be first step to look into to find out the issue on 1MHz and 100KHz output. Any suggestion at this point is helpful.

Regards,
Xtentec