EEVblog Electronics Community Forum
Products => Test Equipment => Topic started by: mtwieg on May 29, 2023, 03:04:51 pm
-
Recently my company has come into possession of an Agilent 4294A impedance analyzer. It came with a 42941A Impedance probe kit, and a 100 ohm load calibration. All seems to be in very good condition. But I primarily would like to use it for measuring balanced circuit impedances, in the 10KHz-10MHz frequency range.
Agilent/Keysight did make balun adaptors (16315A, 16316A) but it's unlikely I'll find one for a decent price. And they have banana terminals, which seems silly for measurement frequencies beyond 100kHz or so. So I'm looking at DIY solutions.
What I'd like to do is make the balun a little coaxial module which I can add to the 42941A probe, just before its pin probe tip (it uses an SMA connector for its tips, surprisingly convenient). That should make corrections easier, since the analyzer already knows the corrections for the impedance probe.
I'm wondering if anyone has recommendations on COTS transformers for the balun itself (that I could hack SMA connectors onto), or cores/materials if I wind one myself. Hoping to cover 10KHz-10MHz with one balun, but open to breaking it up if necessary. 1:1 impedance ratio is fine, and I don't plan on using this with any significant voltages between primary and secondary.
-
Thirty years ago, at work, we had two -hp- devices for coaxial impedance measurements (an 8753 network analyzer and a 4191A), and a full set of -hp- impedance standards.
For measuring balanced networks (resonant circuits with several k\$\Omega\$ resistance at resonance), we made simple adapters using MCL plastic-cased transformers with appropriate frequency and ratio specifications for the specific network under test, in the 5 to 80 MHz range.
We had a simple coaxial connection to the 50 ohm winding on the transformer, and a two-socket, 0.2" spacing cheap female connector on the balanced high-Z side.
The important thing was to do the calibration with a mating two-pin male header: we used an open, short, and a 1206 SMT resistor equal to the nominal impedance on the high-Z winding.
The 8753 firmware did not support a calibration impedance value that high, so we lied to it and divided the characteristic impedance setting by a factor of 10.
We could use a reasonable length of good coax between the port on the -hp- equipment and the 50 ohm winding of the transformer.
We had a small collection of 1206 resistors on male headers, of different values for different transformers, and verified the calibration by measuring them with another transformer ratio.
An example of MCL's current production, similar to the models we used, is https://www.minicircuits.com/pdfs/ADT8-1T+.pdf (https://www.minicircuits.com/pdfs/ADT8-1T+.pdf) with an 1:8 impedance ratio from 50\$\Omega\$ and the relevant output pin spacing of 0.2 inch.
-
Thirty years ago, at work, we had two -hp- devices for coaxial impedance measurements (an 8753 network analyzer and a 4191A), and a full set of -hp- impedance standards.
For measuring balanced networks (resonant circuits with several k\$\Omega\$ resistance at resonance), we made simple adapters using MCL plastic-cased transformers with appropriate frequency and ratio specifications for the specific network under test, in the 5 to 80 MHz range.
We had a simple coaxial connection to the 50 ohm winding on the transformer, and a two-socket, 0.2" spacing cheap female connector on the balanced high-Z side.
The important thing was to do the calibration with a mating two-pin male header: we used an open, short, and a 1206 SMT resistor equal to the nominal impedance on the high-Z winding.
The 8753 firmware did not support a calibration impedance value that high, so we lied to it and divided the characteristic impedance setting by a factor of 10.
We could use a reasonable length of good coax between the port on the -hp- equipment and the 50 ohm winding of the transformer.
We had a small collection of 1206 resistors on male headers, of different values for different transformers, and verified the calibration by measuring them with another transformer ratio.
An example of MCL's current production, similar to the models we used, is https://www.minicircuits.com/pdfs/ADT8-1T+.pdf (https://www.minicircuits.com/pdfs/ADT8-1T+.pdf) with an 1:8 impedance ratio from 50\$\Omega\$ and the relevant output pin spacing of 0.2 inch.
Thanks for the input. Minicircuits was definitely the first place I looked for COTS parts, though I don't think their stuff would work well below 1MHz (or rather, would require a lot of correction).
When using a balun with an impedance ratio other than 1:1, is it possible to get the instrument to display correct results? Or did you have to manually convert the primary-referred measurement to the secondary-referred impedance?
-
On the 8753, we could set the parameter for Z0 to a value other than 50\$\Omega\$, but sometimes our calibration value was too high for the firmware and we cheated by dividing it by 10 to keep on scale.
The important thing is to put the calibration artifacts (resistor, open, short) in a location that corresponds to where you are going to measure.
For lower frequencies, you could try audio transformers, such as from Jensen or Hammond or Lehle.
So long as you do not overdrive the transformer into non-linear behavior, its loss (etc.) should calibrate out with the open/short/resistor method, to the computation accuracy of the software.
If I remember correctly, we tried to use MCL transformers that were within their -1 dB bandwidth and matched the expected real value of the impedance (roughly) to 50\$\Omega\$.