[R-1051] Bendix Frequency Standard (was RE: Tools, and first assessment)

[David Wise]

[Regarding my R-1051 plain's Bendix Frequency Standard,
which does not match NAVELEX 0967-LP-970-9010.]

I looked at it some more and sketched out the S1 wiring.
I think the A2 circuit board is the same as NAVELEX, with
only the switch wired differently.  The 5MHz output amp Q9 is
powered all the time, while the COMP lamp mixer/amp Q10/Q11
are powered only in COMP/INT.  EXT(NORMAL) powers down the oven,
while EXT(OVEN STBY) leaves it on.  To me, this makes more sense
than NAVELEX.  If most installations normally used an external
standard, there's no sense in keeping the oven hot, unless you're
preparing for switchover to internal.

I thought some more about the mechanical aspects of thermistor rework,
and I think I'll just epoxy it inside the oven, with wire leads to
a 2-pin connector mating to a 2-pin header on the board.  This will
not require a cut trace, just unsolder and set aside P2 pin 3.

Just for fun, I zapped the old thermistor with a few hundred volts,
current-limited through a 100K resistor, and lo and behold, it
closed up again.  I don't trust it for a moment, and its resistance
is lower than normal, but it let me test the control loop, which
seems fine although I don't know what temperature it's at yet.
(My thermistor-based home thermometer overflows, have to borrow
a thermocouple from work.)

Now that I've adjusted C3 (coarse frequency), the various outputs are
right on.  (Looks like the PO set it for room temp after the oven died.)
I am going to put in 5MHz in EXT, and "bend" it until one of the dividers
fails, so I can determine the error bars.  The trimmers still have their
original paint spots locking them, and I don't want to break them loose
unless I have to.

I have made my first adaptor cable, just a single BNC to hybrid-D female.
(The sex of hybrid-D coax contacts is not intuitive; the "male" has a
female signal contact, but the contact as a whole goes inside the other
when they're mated.)

I discovered that R15, the temperature-set trimmer, was noisy.  I thought
to drill a small hole to admit cleaner/lube, but as soon as I touched
it it broke right apart.  Wirewound trimpots in this form factor
(Bourns 3007P) are out of production.  The recommended replacement
is the cermet 3006P, which is still common on eBay, but it's only
rated to 125C, while the 150C square-shaped 3296 is abundant.
I'll see how much rework it would take to use one, but I have both coming.

Since NTC thermistors are so much easier to find than PTC, I decided to try one.
It's on its way, just a buck at eBay.  10k cold, about 1.1k at 85C.
This requires changes in the other bridge resistors, not just because
of the different value but also because the rate of change is higher.
We want to keep the same servo loop gain as before, or it might oscillate.
It looks like it will come out right with 8.2k on top and a 6.8k in
series with the thermistor on the bottom.  At 85C, most 10k thermistors
change about 35 ohms per degree, yielding about 0.44%/C versus 0.43 original.
Then two traces must be cut, to reverse the error polarity.
By the way, 1.2Meg feedback resistor R23 is a selected part.
Selected for what?  They don't say.  My guess is critical damping.

Since my old thermistor went open, I am concerned about NTC thermistor
failure modes.  If they also like to go open, there's a risk of runaway
unless I add a supervisory transistor.  If I do this, another resistor
has to change and another trace has to be cut.  I need to study the
board layout to assess the difficulty.  Does anybody have experience
with NTC failures?

I tried to attach a sketch but it didn't work out.  Contact me
off-list to see it.  Or does this list have a place to post photos?

Dave Wise (david<underscore>wise<at>phoenix<dot>com)

>-----Original Message-----
>From: r-1051-bounces at mailman.qth.net 
>[mailto:r-1051-bounces at mailman.qth.net] On Behalf Of David Wise
>Sent: Friday, June 29, 2012 5:11 PM
>To: R-1051 Discussion Group
>Subject: Re: [R-1051] Tools, and first assessment (was RE: 
>Intro and WTB: R-1051 cabinet)
>
>I found a PTC thermistor at Mouser.  Vishay part number 
>TFPT1206L3901JM,
>$0.75 each and hundreds in stock.  TC is 0.41% vs 0.43% original.
>The only downside is that it's a surface-mount part (1206 size),
>which makes it a little harder to use.  I am thinking of mounting
>it _inside_ the oven, against the bottom.  Solder it to a little
>piece of FR-4, squash it onto a dab of heatsink grease then a mica
>sheet then another dab of grease, flip face-down and JB Weld the
>whole thing to the bottom of the well.  Solder the lead wires directly
>onto J2 pins 1 and 3.  No oven disassembly.  No circuit modification.