Cal. kit for a 4195A & 41951A

[Fred Bartoli]

Hi,

Probably a question for Tom, but others are welcomed to answer too

I've finally got my hands on an HP4195A (0-500MHz and freshly caled) and
also the 41951A impedance set + 16092A fixture. Unfortunately the 41951A
has been badly tempered and the calibration kit (short and 50R load) is
missing.
After some time spent I've fixed the 41951A, and seem to have a fully
functional unit (checked through calibrating it *with* the 16092A
fixture and a pair of selected 100R-0805 as the reference load).
At least the measurements I could do make sense, so far.

Now I'd just like to have all the possible accuracy from this and thus
want to rebuild an APC7 cal kit I can trust.

The original parts aren't available anymore, just 2 50R load left at
agilent at a (cough) $1350 price! and no more short.

I've found agilent has the 909C load which should be OK (given for
SWR=1.005 @ 2GHz).
I've also found an Amphenol short which should be OK since, AIUI, the
short is made by compressing a flat piece against the ref plane of the
mating connector.

I also want to add to this an APC7-SMA adapter for some on board
impedance measurement.

I believe all this would be perfectly OK but since I have not much hands
on experience with VNAs I'd just want to make sure.

Oh, last point, for those knowing it: I don't know if the open load
(04191-85302)is complete or not (it has an opened threading at the
opposite of the mating end) and could I find any picture/doc of it.


Any comment much appreciated.


--
Thanks,
Fred.

 

[K7ITM]

Fred Bartoli wrote:

Hi,

Probably a question for Tom, but others are welcomed to answer too

I've finally got my hands on an HP4195A (0-500MHz and freshly caled) and
also the 41951A impedance set + 16092A fixture. Unfortunately the 41951A
has been badly tempered and the calibration kit (short and 50R load) is
missing.
After some time spent I've fixed the 41951A, and seem to have a fully
functional unit (checked through calibrating it *with* the 16092A
fixture and a pair of selected 100R-0805 as the reference load).
At least the measurements I could do make sense, so far.

Now I'd just like to have all the possible accuracy from this and thus
want to rebuild an APC7 cal kit I can trust.

The original parts aren't available anymore, just 2 50R load left at
agilent at a (cough) $1350 price! and no more short.

I've found agilent has the 909C load which should be OK (given for
SWR=1.005 @ 2GHz).
I've also found an Amphenol short which should be OK since, AIUI, the
short is made by compressing a flat piece against the ref plane of the
mating connector.

I also want to add to this an APC7-SMA adapter for some on board
impedance measurement.

I believe all this would be perfectly OK but since I have not much hands
on experience with VNAs I'd just want to make sure.

Oh, last point, for those knowing it: I don't know if the open load
(04191-85302)is complete or not (it has an opened threading at the
opposite of the mating end) and could I find any picture/doc of it.


Any comment much appreciated.


--
Thanks,
Fred.

Hi Fred,

Unless you are interested in making precision lab-grade measurements, I
think you'll do fine with much less expensive and more available cal
standards. Personally, I'd get a set of adapters from your APC-7s to
SMAs -- and out to only 500MHz, even BNCs should be OK. If you want to
get a bit fancier, adapt the APC-7s to 3.5mm connectors, which mate
with SMAs but are higher precision. I personally don't think it's
worth worrying about. It's a really good idea to always leave adapters
connected to the APC-7s to protect them anyway, assuming you use the
instrument often enough that you don't want to leave them covered most
of the time. But never leave them uncovered! They are far to easy to
damage if left unmated/uncovered.

My experience designing a 3-way splitter that needed to work well to
6GHz is that inexpensive 0805 SMT resistors in the 25 to 100 ohm range
work quite well to several GHz, if you're careful about the environment
around them (e.g., the size and shape of the holes bored in the
grounded block that holds the connectors). I expect that if you keep
lead length to essentially zero, you can turn an SMB thru-hole jack
into a 50 ohm load that will have at least 40dB return loss out to
500MHz: cut the center pin to the height of the resistors, and solder
two 100 ohm 0.1% 0805's radially symmetrically out from the center pin
to the outer shell. If possible, get someone with a known-good cal
standard set to compare it with that set. An open and a short can be
created in the same way. I suppose that one of the free 2-d fields
solver applications could tell you if you need to make any adjustment
for length, but in any event, just using the same jacks for all three
should yield lengths within a degree out to 500MHz. Maybe if I get
some time to play next week, I can make some measurements of hacked
"standards" against my known-good ones. I'm pretty sure I still have
some of the 0.1% 50 ohm parts around.

Cheers,
Tom