I'd like to design a unified VCXO Carrier Board to these requirements:

1. It can host one of the the following VCXOs:

1.1. HP 10811A-6111 (as from 5370A)

1.2 Morion MV89A

1.3 MTI 260

1.4 CV-950

1.5 Timetech

1.6 Axtal

1.7 Pascall

2. It provides unified tuning: 0V = lowest possible frequency, 3V3 or 5V
   = highest possible frequency, no matter of the VCXO tuning sense and range.

3. provides a 5V tuning voltage reference for those VCXOs that don't
   have one of their own.

4. Frequency can be adjusted from external Vtune input and from a 10
   turn pot.

5.Board has 2 reference frequency inputs with LTC6957 receivers. One of
them can interface the onboard VCXO to the CPLD.

6. Board has a 1pps input 3V3 CMOS level

7. It can lock the VCXO to the reference frequency or the 1pps in.
   Provides LED lock indication.

8. It features a Xilinx Coolrunner2 2C64 CPLD, complexity 64 FlipFlops +
   combin. Logic. Unused Pins are brought out to Testpoints in 100 mil
   grid. The Coolrunner remembers its configuration and can be reprogrammed
   using the standard Xilinx USB dongle. It has a 10 pin 2mm header for
   this purpose. Small circuits can run at 200 MHz. This function exists
   already:
   <
   https://picasaweb.google.com/lh/photo/4Bpcfouj8WH0shNGIyuVUtMTjNZETYmyPJy0liipFm0?feat=directlink

9. The Coolrunner provides a standard 1pps /20us out, maybe 10/100/1000 pps.

10. 2 Monoflops for 1pps LEDs in/out

11. There are 2 output buffers that drive valid 3V3 CMOS into 50 Ohms.
    They can be re-clocked to LTC6957 outputs with 1G74 flipflops.

12. Unbuffered VCXO output is available on SMA connector

13. One additional buffered output (LMH6702/AD8009 or discrete to avoid
    neg. supply). This is not meant to be a distribution amplifier.

14. Regulators for the voltages needed.

15. Requires soldering skills 0603 / sot23-5 / MSOP. No commercial
    interest. Could be TAPR or DIY.

I'm open to suggestions & ideas.

regards,
Gerhard, DK4XP

Hi

Multiple footprints are fine and they don’t generally take up a lot of space. The 10811 is a bit of a hassle in that respect.

That immediately gets you into op amps and feedback resistor stability. With a number of combinations, the required resistors can get pretty expensive. It
also gets you into dual supplies with some OCXO’s.

This gets you into the same sort of “is 1 ppm stability on the tuning good enough” set of questions.There are a number of OCXO’s out there that have odd
reference voltages.(10 V etc)

Sounds good. Consider ground loops and offsets on the external input.

I’m not sure these are needed if the destination is a CPLD.

Hopefully buffered with discrete logic. CPLD’s often are not very rugged in terms of over voltage on the inputs.

Probably not easily with a 64 flip flop CPLD. For a full narrow bandwidth PLL you will need some more “stuff” on the board. With a
wideband loop, you will have a lot of noise on the oscillator.

Given the number of much larger devices in the 10 to 100X larger range that are still under $10, I’d look at another device.

ok.

I would dedicate a couple of discrete (sot-23 logic) buffers to the 1 pps outputs.

With all this stuff on the board, layout will be “interesting”. Keeping it all from cross talking will be a challenge.
That gets even more complex as the number of external connections goes up. The more configurations, the
more things to worry about …

One simple example of the above:

Your OCXO:

1. pulls 3 ppm with a 5V tune.
2. has a stability over -30 to +70C of 1 ppb
3. has an ADEV at 1 second of 1 ppt

That comes out to:

Stability wise, that’s 10 ppt / C
Tune wise you have 0.6 ppm / volt or 0.6 ppt / uV

To maintain the 10 ppt/C your reference / tune voltage needs to be better than that. For instance, 5X better is a fairly
common design goal in many systems. That’s 2 ppt/C from the tune. Roughly 4 uV of stability would do it.

With a 5V reference the same 4uV is a 1 ppm / C reference. The same math applies to the stability of the resistors
in the tune circuits and the offset drift on the op-amps.

Lots to think about !!!

Bob

Gerhard,

Don't know if it would help much in your ultimate plan but the concept has been done before as the 53132-60011 and 53132-60016 Optional Time Base Boards for the 53132A counters.  These boards will accommodate either an Isotemp OCXO33-46 (HP 1813-0931) OCXO for OPT 001, or either of two variants of the 10811 for OPT 010 or 012.  There is a jumper that needs to removed and possibly connected to a power supply for the 10811's.

Might be helpful to take a look at for ideas.

Hope this helps.

Joe

-----Original Message-----
From: time-nuts [mailto:time-nuts-bounces@febo.com] On Behalf Of Gerhard Hoffmann
Sent: Thursday, October 22, 2015 2:41 AM
To: Discussion of precise time and frequency measurement
Subject: [Bulk] [time-nuts] Unified VCXO Carrier Board

I'd like to design a unified VCXO Carrier Board to these requirements:

1. It can host one of the the following VCXOs:

1.1. HP 10811A-6111 (as from 5370A)

1.2 Morion MV89A

1.3 MTI 260

1.4 CV-950

1.5 Timetech

1.6 Axtal

1.7 Pascall

2. It provides unified tuning: 0V = lowest possible frequency, 3V3 or 5V = highest possible frequency, no matter of the VCXO tuning sense and range.

3. provides a 5V tuning voltage reference for those VCXOs that don't have one of their own.

4. Frequency can be adjusted from external Vtune input and from a 10 turn pot.

5.Board has 2 reference frequency inputs with LTC6957 receivers. One of them can interface the onboard VCXO to the CPLD.

6. Board has a 1pps input 3V3 CMOS level

7. It can lock the VCXO to the reference frequency or the 1pps in.
   Provides LED lock indication.

8. It features a Xilinx Coolrunner2 2C64 CPLD, complexity 64 FlipFlops + combin. Logic. Unused Pins are brought out to Testpoints in 100 mil grid. The Coolrunner remembers its configuration and can be reprogrammed using the standard Xilinx USB dongle. It has a 10 pin 2mm header for this purpose. Small circuits can run at 200 MHz. This function exists
   already:
   <
   https://picasaweb.google.com/lh/photo/4Bpcfouj8WH0shNGIyuVUtMTjNZETYmyPJy0liipFm0?feat=directlink

9. The Coolrunner provides a standard 1pps /20us out, maybe 10/100/1000 pps.

10. 2 Monoflops for 1pps LEDs in/out

11. There are 2 output buffers that drive valid 3V3 CMOS into 50 Ohms.
    They can be re-clocked to LTC6957 outputs with 1G74 flipflops.

12. Unbuffered VCXO output is available on SMA connector

13. One additional buffered output (LMH6702/AD8009 or discrete to avoid neg. supply). This is not meant to be a distribution amplifier.

14. Regulators for the voltages needed.

15. Requires soldering skills 0603 / sot23-5 / MSOP. No commercial interest. Could be TAPR or DIY.

I'm open to suggestions & ideas.

regards,
Gerhard, DK4XP

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Gerhard wrote:

As I have said before, there is very little if any advantage to using
an LTC6957 at 10MHz (as opposed to using a run-of-the-mill
comparator), and the LTC6957 is not as good, even with filtering on,
as a circuit with lower inherent jitter such as an optimized
Wenzel-style squarer.

LT published an app note that shows the improvement with filtering
enabled, using a 10MHz input.  The relevant graph is attached below.

The graph compares an optimized 6957 implementation without filtering
and with optimum filtering.  At an input level of -10dBm, the phase
noise floor is 7dB lower with filtering, and at an input of +10dBm,
the improvenent is <2.5dB.  Extrapolating beyond the graph to the
right, at an input level of +13dBm (= 1Vrms, the customary level for
frequency references), the improvement with filtering will be very
near 0dB.  It is not hard to add 10dB or more of gain at 10MHz with a
residual phase noise penalty in the -180dB range.  So, with a 10MHz
input, even if the output level of the source is lower than +13dBm,
you can easily do just as well with a 6957 with the filtering off as
with it on.  And with a squaring circuit that has inherently lower
residual PN than the 6957 (e.g., an optimized Wenzel-type squarer),
you can do better than with any permutation of the 6957.

These data are consistent with testing I have done of the LTC6957 and
other squaring circuits.

I realize that you did not say you expect to use the board only at
10MHz, and the LTC6957 with filtering may provide some improvement at
lower frequencies (compared to the 6957 without filtering).  But even
with filtering on, the 6957 will not outperform an optimized
Wenzel-type squarer until you get well down into the kHz range.  And
for the usual time-nuts case -- 10MHz at ~13dBm -- the 6957 is not
the best solution, even with filtering enabled.

Best regards,

Charles

Am 22.10.2015 um 18:30 schrieb J. L. Trantham:

Thanks, I'll take a look if I find it to harvest ideas, but given the
ebay situation, the Morion and
the MTI-260 are a must. The project will probably require jumpers by the
pound..
I still have a wire-wrap gun. :-)

Gerhard

The price of the resistors is small change
compared to the $8 connector for the 10811.
Then again maybe the plan is not to actually
build one board with all the parts that will
accomodate any of the OCXOs on the list

Dimitri

At 04:44 AM 10/22/2015, Bob Camp wrote:

Charles
Your statement about the PN of comparators conflicts with my measurements. The LTC6957 evaluation board had an 18dBc/Hz lower phase noise floor than a comparator circuit with 10MHz 15dBm inputs. However I only measured a single comparator circuit. The Holzworth sine to CMOS converter had a comparable PN to the LTC6957-4.
I haven't, as yet measured the PN of an optimised Wenzel circuit.My setup for this measurement had a PN floor of around -180dBc/Hz.

Bruce

 On Friday, 23 October 2015 11:01 AM, Gerhard Hoffmann <dk4xp@arcor.de> wrote:

Am 22.10.2015 um 18:30 schrieb J. L. Trantham:

Thanks, I'll take a look if I find it to harvest ideas, but given the
ebay situation, the Morion and
the MTI-260 are a must. The project will probably require jumpers by the
pound..
I still have a wire-wrap gun. :-)

Gerhard

time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.

Am 22.10.2015 um 22:04 schrieb Charles Steinmetz:

What do you consider a run-of-the-mill comparator? LM139, LMV7219,
AD8561, ADCMP580?

What optimizations?  I have seen the "Wenzel" circuit in cheapish
frequency counter inputs in
the late seventies, maybe with a diode bridge added as input protection..

so it shows that one can replace filtering by signal power :-)

But in the KHz range, the filter corner frequency should be much lower
than in the 6957...

The 100 MHz range would be more interesting. The Wenzel squarer would
then need
"real transistors" with higher 1/f corner. The PLL would have a sub-Hz
BW in this application,
so the race for the lowest floor seems not too important. OK, maybe for
the 1pps.

The 6957 takes one square centimeter including transformer and works for
all frequencies.
Any objections against the AD9901 phase comparator? I have a tube of them.

regards, Gerhard

Hi

If you are buying the “right” 0.5 ppm / C resistors, they often run well over $10 each. You have
a lot of them in a circuit….

Bob

Gerhard wrote:

For squaring 1-10MHz sine waves, the LT1719 and 1720 are the best
that I've found.  This is a matter of how much internal hysteresis
the comparator has, how smoothly the internal hysteresis acts, how
much gain the part has and how it is structured, how fast it is (both
risetime and propagation delay), whether it suffers from thermal
feedback from the output stage to the input stage, how much internal
ground bounce it has, and a host of other die-level issues.

The LT1719 uses bipolar input supplies, so ground can be the
reference voltage and also be in the center of the input common-mode
range.  This is always quieter than biasing the inputs to the middle
of a single supply, which is usually done with the 1720 and many
other comparators.  (Like other single-supply comparators, the 1720
will work referenced to ground with only a positive supply -- but (i)
the inputs are then at the very edge of the input common-mode range,
and (ii) you can only drive the input 100mV below ground, so you have
to figure out how to clamp the input signal.  It is much easier to
just use an LT1719 with +/-5v on the input stage, and it works better, too.)

Page 22 of the LT1719 datasheet shows the simplest possible circuit
and discusses its performance.

Use medium-speed transistors, bias both bases from the same low-noise
voltage reference such as an LM329, capacitively couple the emitters,
and use a higher supply voltage, for starters.  I use MPSH81/MMBTH81s
and a power supply of around 20v (see attached schematic) for a
reasonably optimized implementation.  Other transistors can be used,
but I've found that the H81s work better for squaring 1-10MHz sine
waves than anything else I've tried -- they hit the sweet spot of the
bandwidth/gain tradeoff and have a nice flat gain vs. current characteristic.

It's a matter of the slew rate of the input sine wave at
zero-cross.  Once you reach the critical slew rate for any particular
input architecture, the comparator is hard-switched fast enough that
it doesn't spend significant time in the linear region generating noise.

That should work fine.

Best regards,

Charles