[time-nuts] Distribution amps and isolation transformers

Joseph Gray jgray at zianet.com
Mon Nov 24 14:39:01 EST 2014

Although my current project doesn't have the strict requirements that
many of you have, I thought I would mention part of it, due to the
recent talk of distribution amps.

I am putting one of the FE5680A's that I bought a while back into a
permanent home. I have a nice box from a gutted HP microwave mixer
that is just about the perfect size. I also have two surplus RGB video
distribution amps rated for 100 MHz. After changing out the 75R
resistors for 50R, I will mount the bare board in the HP box, along
with the rubidium. Since the video amp BNC's are not isolated, I
thought I'd better do that while I was at it.

Being frugal, I used a heat gun to remove several transformers from
old 3Com Etherlink III ISA cards. As most of you know, there is a
single 1:1 transformer for the 10Base-2 and a dual 1:1 transformer
module with low-pass filters for the 10Base-T.

After removing the parts, I thought I'd characterize them on the
spectrum analyzer. It was a good thing I did. Some of the transformers
were obviously bad and had wacky curves. (the Ethernet cards were
pulled from a scrap pile.)

All of the dual transformer modules were within about 2 dB of each
other. The -3 dB point being around 15 MHz and dropping to about 20 dB
down or more at 30 MHz.

The single transformers were of two types: square and oval with two
flat sides. All of the oval transformers were almost flat out to  50
MHz (I didn't test higher). The square transformers were odd. Their -3
db point was around 8 MHz and then things gently sloped downward about
30 degrees for quite a ways. Being that both styles of transformer
were being used for the 10Base-2 connection, I assumed that they
should behave similarly. Rolling off at 8 MHz doesn't sound good for a
10 Mbit speed.

The finished project isn't going to be anything fancy. I just wanted
to make a semi-portable standard. I will divide things down so I also
have 1, 5 and 10 MHz TTL outputs, in addition to the 10 MHz sine.
Later, I may also incorporate a DDS that is locked to the 10 MHz to
give me programmable sine and TTL outputs.

I also spent all weekend tweaking both of the rubidiums via the serial
port. Both of them had a zero offset programmed. Using a scope and
triggering on my Z3801A, I slowly dialed things in until I got as
little drift as possible. One rubidium needed to go low and the other
high. BTW, using the calibrator program from VK4XV made things much
easier when talking to the rubidiums serially.

Joe Gray

More information about the time-nuts mailing list