|
Er, yes. Thanks for that suggestion. I'm afraid such micro-surgery is a bit beyond me at my age!
I think maybe these tests would be better suited to a scalar network analyzer - or just a spectrum analyzer with tracking generator. Esp one of the old analog storage analyzers in fact (giving my age away again).
|
I suspect what you have there are RF/IF transformer(s)?with tuned primary and untuned secondary.?
?
If there is a tap on the primary, the center pin of the three pin side, it is used to feed supply voltage to
the active device.?
?
I would remove one side of the primary dog bone cap and take a low frequency?
measure of the primary and secondary inductance values. Get this from a S11 shunt measurement.
?
Then while taking? a measurement of the primary inductance (S11), place a short circuit on the secondary side
and re measure the shift in the primary L value, again from S11. The primary L value should decrease and this?
shift in L provides the coupling coefficient. This value obtained from SQRT [(1-L_primary_sc/L_primary_oc)]. Again,
primary L is the measurement, shunt mode, S11.... while the secondary is OC and then SC.?
?
Given all this one can ascertain the frequency response analytically with the dog bone C value provided and?
an assumption on the secondary termination Z.?
|
It runs fine under Fedora linux using WINE.
--
T. Gerbic
Central California
|
Great, thanks for the clarification.
|
Yes, jumper across your test fixture, center to center, shield to shield.
|
Okay, "Response" and not "One path two-port" then. I'll give that a try tomorrow. One other point, though: you say to choose 'thru' and connect a 'thru' standard. Surely a 'thru' is a basically a 'short' connected in series with the signal path rather than shunting it to ground? So kind of like just connecting a jumper between the two coax inner conductors? Is that what they want?
|
I'm at work so I don't have my VNA with me.
The trick is to activate the S21 trace (press parameters, then S21, make sure it's showing the uncorrected S21 trace) then bring up the Cal Menu.
Then select Response (in the context of S21). You would be present Open, Short and Thru. Connect you thru and press "Thru", ignore the Open and Short.
Again, I assume your VNA is similar to my 8510 in term of Calibration procedure.
However, if all you want is to discern the pass and stop bands, you probably don't need to calibrate. I only calibrate my VNA when I need serious measurements. This is to minimize the wear and tear on the Agilent Calibration Standards cost me an arm and a leg.
Good luck! Calvin
On Wednesday, February 19, 2025 at 07:00:33 AM PST, Jinxie via groups.io <paul666@...> wrote:
Okay, I've hit a bit of a brick wall with the calibration. It was suggested a through calibration for S21 would be sufficient for this test. However, the choices under the calibration menu are causing me some confusion.
I go for the cal menu and get these choices:
?
Response
Response and isolation
S11 one port
S22 one port
Full 2 port
One path 2-port
?
The last choice seems to me the one to go for, so I choose it and then get a request for the following standards to be connected:
?
Reflection
Transmission
Isolation
?
I'm not familiar with these standards at all. I only have a short, an open and a load. So how should I proceed, please?
I note here's also a feature on the extended menu for setting Z0. Not sure if that could assist with the impedance mismatch or not....
?
|
Okay, I've hit a bit of a brick wall with the calibration. It was suggested a through calibration for S21 would be sufficient for this test. However, the choices under the calibration menu are causing me some confusion.
I go for the cal menu and get these choices:
?
Response
Response and isolation
S11 one port
S22 one port
Full 2 port
One path 2-port
?
The last choice seems to me the one to go for, so I choose it and then get a request for the following standards to be connected:
?
Reflection
Transmission
Isolation
?
I'm not familiar with these standards at all. I only have a short, an open and a load. So how should I proceed, please?
I note here's also a feature on the extended menu for setting Z0. Not sure if that could assist with the impedance mismatch or not....
?
|
Indeed. And that is what I found when I used my bare 'rig' to test the yellow can, which came out to have a peak at the very common IF of 7.3Mhz. I'll grant you that whilst the peak was readily discernible, it did not resemble a band-pass filter at all. There was no identifiable -3db shoulder on either side as it was really sharp, almost approaching the curve of a crystal filter. Nevertheless, it was clearly a 7.3Mhz IFT and that's all I really needed to know to list them for sale.
|
On Tue, Feb 18, 2025 at 04:02 PM, Jinxie wrote:
I may be able to run this Windows program under WINE
Yes, that what people use. You know, for your application it might just be easier to use 1k input and output resistors and eyeball the response. From what you said, I gather that you're most interested in simply categorizing the bags of filters/transformers. It doesn't sound like you need to worry about characterizing the stopband response accurately, which is what limited dynamic range will obscure. Even if 1k isn't right, I think it might tell you what the part does and at what frequency. Brian
|
Okay, thanks, Brian. I may be able to run this Windows program under WINE, I'll have to investigate further on the appropriate discussion forums.......
?
J.
|
You'd have to run a Windows emulator.
The program changes the 50 ohm reference impedance to the impedance you specify. For example, if you set the renormalization impedance to 1.5k ohms, the renormalized |S21| response of a filter will be as if had you driven and loaded it with 1.5k. If you measure a device using an impedance it's not designed for, you won't get the response it would exhibit in a circuit that uses the proper impedances. The program lets you see the correct response without actually driving/loading the device with the proper impedance.
The attached plot shows the response of a Murata 10.7 MHz ceramic filter. The blue curve is |S21| for the 50 ohm VNA measurement. The red curve is after the data was renormalized to 330 ohms, which is the source/load impedance the filter was designed for. The renormalized response is as expected.
Brian
|
Brian, I'm sorry. What does this program actually do? I should also say I only have Linux computers, so that adds another barrier to your suggestion, I'm afraid.
|
Ed, yes I can see how it? would be possible to rewind one of these things in practice, but for me I think it would be a task too far. I just don't have the dexterity to do something like that. I did take one completely apart and the actual core was miniscule; I'd estimate the diameter as less than 2mm. I'd really struggle with that!
|
One more thing - you can get some idea of the transformer pinouts and circuit topologies in old transistor radio schematics, from say the 1960s - 1980s.
?
Ed
|
Yes it's well worth taking some apart to see the innards, and how to put them back together. Those cans seem to be a decent size (1/2" and 3/8" square ones are very common and workable) for surgical mods for other uses. You can delete the built-in caps, and change the windings within reason. The ones smaller than that can be quite difficult to work on, especially due to the very fine wire needed to fit.
?
Ed
|
On Mon, Feb 17, 2025 at 03:15 PM, Jinxie wrote:
As for the 50 ohms, that's the 'system impedance' and there's nothing I can do
to change that!
Sure there is. Download the reference impedance renormalization utility near the top of the following page. Download instructions are at the bottom of the page. Try various renormalization impedances until you get a reasonable response. Google AI says the impedance is around 1.5k ohms for typical 455 kHz IF filters. I'm not sure I'd trust that number, but I think you'll know when you've chosen something close to the correct value. Brian
|
That may well be necessary if I encounter one of those 'problem child' devices that Ed described!
|
If you have a lot you can carefully disassemble one of each to get further knowledge.?
toggle quoted message
Show quoted text
On Feb 18, 2025, at 8:34?AM, Jinxie via groups.io <paul666@...> wrote:
? Ed, many thanks for that comprehensive answer. All bar one of these devices have that capacitor you describe tucked away in the base. Can't be more than a few tens of pF at most I'd have thought. The sole exception is one with 6 pins instead of 5 for some reason. The 5 pin ones have 3 on one side and 2 on the other and I *believe* the 3 side is input and the 2 pin side is output. Normally only two of the three input pins are used. I think there's an extra inductor in there if you want to use it in which case you connect to the two outer pins. Seems there's a bit more to these trannies than first appears!
|
Ed, many thanks for that comprehensive answer. All bar one of these devices have that capacitor you describe tucked away in the base. Can't be more than a few tens of pF at most I'd have thought. The sole exception is one with 6 pins instead of 5 for some reason. The 5 pin ones have 3 on one side and 2 on the other and I *believe* the 3 side is input and the 2 pin side is output. Normally only two of the three input pins are used. I think there's an extra inductor in there if you want to use it in which case you connect to the two outer pins. Seems there's a bit more to these trannies than first appears!
|
Jinxie
