[R-1051] Frequency Standard NTC Thermistor Retrofit FINAL

Mon Oct 29 19:49:39 EDT 2012

INTRODUCTION

The R-1051 series Frequency Standard is a high-failure item.  There are a

lot of dead ones around.  Usually they have a bad thermistor.  Leaded PTC

thermistors like the original are hard to find, but an NTC thermistor can be

used instead.  It's easy and cheap, and it works fine.

If anyone expresses interest, I'll post photos and sketches on SkyDrive.

Warning: This mod is not applicable to the early R-1051 plain frequency

standard, P/N 666231-006, which uses an oscillator and rectifier in the

oven control instead of a DC amplifier.  If you have one of these, you're

on your own, and good luck!

DISCUSSION

This mod replaces fixed resistor R13 with an NTC thermistor, and replaces

PTC thermistor R2 with a fixed resistor.

I used a 100K thermistor.  At the time I wrote this, October 2012, eBay members

"tayda2009" and "taishopetc" were selling a pack of 5 for $0.99 including shipping.

It's made by a Taiwanese company called Thinking; it's part number TTC05104.

The TTC05 datasheet is available at http://www.thinking.com.tw/documents/en-TTC05.pdf .

There is nothing special about this part; it has about the same R-T curve as most

TC thermistors.  They don't put the curve on their website, but they'll send you a

pdf if you ask.  I have it too.

You have to modify circuit board A2A5A1, but it's easy and reversible.

While working, refer to the A1 schematic, Figure 5-6 in NAVELEX 0967-LP-970-9010.

NTC thermistors are much more sensitive than PTC.  An NTC thermistor placed

directly into the old bridge circuit will provide too much gain, causing the control

circuit to oscillate.  We work around this by adding a resistor in series, to reduce

the percent change of the total.  This also makes the rate of change vary with

thermistor resistance Rt, but a second fixed resistor in parallel with Rt makes

the total approximately linear near operating temperature.

Temperature trimpot R15's adjustment range is (5.3K/10K - 4.3K/9K)*10V or about 500mV.

The original thermistor together with upper leg R13 puts out about 6mV/degree.

That's about 80 degrees of adjustment range, or 5 degrees per turn.  (It's a 15-turn pot.)

This is really twitchy, and the slightest drift will throw it way off.

It's possible to eliminate this problem.  If you increase your thermistor output,

the degrees per turn go down correspondingly, but you have to change

R23 to keep the loop stable.  4.7K/15K/11K makes about 30mV/degree instead of

6mV/degree, which makes R15 about 1 degree per turn.

This resistor set is for R16 = 4.3K .  If you have an R-1051H module (3.9K),

use 4.7K/18K/10K instead.

My calculations said I had to change R23 to about 240K if it was 1.2M before,

but testing showed that 330K is optimal for my 4013399-0701 with mains regulation

upgrade (see below).  This should be good for all versions where R16 is 4.3K* .

For the R-1051H, I'd try 220K or 270K.

* The NAVELEX schematic labels R2 as 3.9K but that's at 25 degrees.  It's around

4.7K at 85.  The R-1051H schematic also labels it 3.9K but that can't be right

because R16 is also 3.9K; it must be a different thermistor, one that's about

4.3K at 85.

PROCEDURE

0. Determine your crystal's turnover temperature.  The manual says 85 degrees but

    mine was 82.

1. Connect a 3" to 5" cable to your new thermistor, which I call R913c.

    Try to arrange for the cable to be away from the crystal and oscillator area when

    A1 is reinstalled in the oven.  This will make sense after you have read the entire

    procedure.

2. At the other end of the cable, put a connector, which I call J92.  I used a 2-contact

    socket that mates with the popular 0.1"-spaced square pins.

3. Arrange thermistor R913c in the bottom of the oven near the abandoned R2.

    Make it lie flat and out of the way.  The connector must reach the upper left

    corner of board A1 when viewing the component side.  Seal it in place.

    I used JB Weld.

4. Cut a piece of double-sided plated-through padboard, 1.0" long and 0.3" wide;

    10 pads by 3 pads.  (You can use single-sided if you're careful soldering.

    Arrange it with the pads up.)

    Solder square pins P92 at one end, extending off the board and parallel to it.

    P92 will point toward the bottom of the oven after installation, and thermistor

    connector J92 will mate to it.

5. On the padboard, tack three resistors, R913a (4.7K), R913b (15K), and R902 (11K).

    Arrange R913a near A1 centerline, R902 near A1 edge, and R913b in the middle.

    R913b goes across P92.

    One end of R902 goes to P92 pin 2 with the other end uncommitted for now.

    One end of R913a goes to P92 pin 1 with the other end uncommitted for now.

    R902 replaces thermistor R2 in the lower-left leg of the bridge, and the R913 trio

    replaces fixed resistor R13 in the upper-left leg, with R913a the series element,

    R913b the shunt element, and R913c the thermistor itself.  R913b and R913c

    are in parallel, and R913a is in series with them.

6. Prepare the top surface of R15, and glue the padboard to it.  I used super glue.

7. Remove R13 (4.7K) and P2 pin 3 (which goes to old thermistor R2).

8. Run a wire from R902/P92 to the left pad where R13 used to be.

    This pad also goes to where P2 pin 3 used to be.  This is the temperature

    sensing node.

9. Run a wire from the uncommitted end of R902 to ground, which is available

    at an unused pad next to the top pin of R15.

10. Run a wire from the uncommitted end of R913a to the right pad where R13
     used to be.  This is +10V.

11. Remove R23 and replace with 330K.  (220K-270K if R-1051H.)

12. Slide A1 partway into the oven, connect J92, then slide A1 in the rest of the way.

13. Adjust R15 for the proper temperature.

TESTING; USING A DIFFERENT NTC THERMISTOR

Transient response is important.  General Dynamics selected R23 to control this.

R23 determines the gain of the amplifier, excluding the output transistor.  In older

A1's where Q7's (2N697) collector load resistor R24 is 2.2K, the nominal value was

1.2M, which gives a transconductance of about 230mA/V .  The latest A1 for the

R-1051H is 1.2K and 820K, giving about 280mA/V,making up for the lower output

of their 3900-ohm thermistor.  Larger values tighten control but increase ringing,

while smaller values reduce ringing but degrade regulation.

The resistor values I provided above are okay for the Thinking TTC05104.

Other parts may need different values.  There are three goals:

 1. Control loop is stable,

 2. Setpoint is achievable within R15's trim range, and

 3. Bridge voltage vs T is linear around the setpoint.

Ask the manufacturer for your thermistor's R-T table; they don't usually put

them on their websites.  With that and a programmable calculator, you can

come up with usable values.  Although you can assess stability with just a

VOM, a storage or digitizing scope or a chart recorder makes it MUCH easier.

You need an extremely slow sweep rate.  I used a triangle-wave function

generator set to .00002Hz as X input.  That's 250s/div .

Monitor Q7 collector during the transient you get when switching from

operate to standby.  (It's less stable in standby, because the higher

supply voltage increases system gain.)  Your amplifier may have more

gain than mine, or less.  Watch the scope trace and you'll be able to tell

what's right for you.  Keep the insulating lid of the oven in place; otherwise

you'll choose too much gain.

MAINS REGULATION UPGRADE - 4013399-0701 ONLY

What a fascinating journey.  My particular Frequency Standard (P/N 4013399-0701)

does not have a preregulator, and R24 is referred directly to unregulated +28V.

This gets away from the R-1051B's embarrassing "dead oven in standby" problem,

but it makes the oven sensitive to variations in the power source.  If the mains voltage

rises, so will the temperature.  However, I discovered that, on a shorter time scale,

this ciruit actually creates negative feedback, above and beyond what's obvious in the

schematic.  Let's say the oven temperature is rising.  The amplifier will pass less current

to the heater.  With less load, the +28V line rises, increasing the current through R23,

which increases heater current.  I discovered this when I tried to improve the temperature

regulation by reducing mains sensitivity.  A simple way to accomplish that is to add a

resistor from +28 to the thermistor side of the bridge, so the supply appears equally in

both bridge arms and thus cancels out.  When I did this, the oven went unstable, because

I had eliminated a hidden source of negative feedback.  I could have reduced thermistor

gain, but why bother since I'm already changing R23?

With 330K in R23, 820K is appropriate for compensation.

PRESERVING ORIGINAL R23

If you want to leave R23 alone, you can scale your new thermistor's output back

down to the original level with a different choice of scaling/linearizing resistors.

The disadvantage is a twitchy R15, same as it was originally.  I don't know why

you'd do this just to preserve one 1/4W carbon comp resistor, but...

I calculated some combinations to save you the work.  They're for TTC05104,

with R16 = 4.3K .  If you are doing an R-1051H with 3.9K, bump R902 down 10%.

Resistors  mV/degree

a/b/902

22/10/27   8.5

24/9.1/30  7.2

27/9.1/33  6.5

30/9.1/36  6      <--- worked for me

33/9.1/39  5.5

SUMMARY

Thinking TTC05104, scaled and linearized by 4.7K/15K/11K, with 330K feedback,

and (4013399-0701 only) 820K compensation.  This should work for all modules

originally using a 4700-ohm thermistor.  You can even reuse your old R13

for R913a.

For the R-1051H, which uses a 3900-ohm thermistor, I'd try 220K or 270K, but I

don't have hardware to try it.

Dave Wise

SWL in Hillsboro, Oregon