EMRFD Message Archive 12238
Message Date From Subject 12238 2016-01-29 10:41:32 n1vc Bifilar Torriod independant testing and also SA design question This should probably be a two part question but here it goes: I am currently winding T201 on the August 98 QST SA on page 41. This transformer takes the amplified 10MhZ 2nd IF and feeds it into a 6db Pad and then to the selected video filter.
Question1) Forgetting about the circuit, I want to verify that T201 works on the bench. I have wound many torriods before and tested them at desired RF frequency with my MFJ-259B. However, I have never built a bifilar transformer in my life! How could I test it with the MFJ on the bench? BTW its specs. are 10 bifilar turns of #28 on a FT-37-43 core.
Question 2) Why did the design call for a bifilar transformer? Could we not have used a link coupled transformer to match Q202 collector to the pad input? Please note I understand that I could simply build the circuit, plug it in, and verify that it is working in the lab. However, I will never learn anything doing that and how to implement it in my own design of the future. Any pointers appreciated!.
12239 2016-01-30 09:15:25 vasilyivanenko Re: Bifilar Torriod independant testing and also SA design question
You can use a conventional transformer if you like. For broadband transformers with a Z ratio of 4:1, 9:1 etc., builders tend to ply bifilar/trifilar wound transmission line transformers. The FT37-43 core exhibits tremendous loss: with 10 turns of wire even spaced, QuL @ 1 MHz = 6 - 10, the SRF [very difficult to measure] runs ~ 35 MHz. With all this loss, twisting the wires in a bifilar transformer improves the coupling and gives a reasonably accurate 200 : 50 ohm transformation across its bandwidth.
So in short, transmission line transformers exhibit wider bandwidth and higher efficiency than conventional.
12240 2016-01-30 09:52:15 n1vc Re: Bifilar Torriod independant testing and also SA design question Okay T/V thanks for your reply...that helps. So in other words the designer might use the conventional transformer approach and if he (or she) does not get enough signal, then he might replace it with the proper Z ratio, Bifilar wound, transformer. I suppose that is a sort of the empirical approach, however, if I wanted to simulate the transformer in ARD or RFSIM99 maybe when I get to the coupling factor I might want to use a higher number than the conventional. Any thoughts here? Actually now I think I remember an example simulation in the ARD manual that does do this. Now to find that old manual in the mess..... And unfortunately ARD does not run in Windows 7, I have to use the XP emulator. What a pain!. I wonder if ARD will run under Windows 10? My guess is probably not.
12242 2016-01-30 17:13:53 kerrypwr Re: Bifilar Torriod independant testing and also SA design question RF transformers and the ferrite inside them are very interesting devices.
I think that the best reference to their design is in the Philips application notes ECO6907 & ECO7213 which cover both conventional and transmission-line types; Google will find these app notes.
The references tell us that the tx-line types are better but I like to see things for myself.
I try to emulate the scientist who was travelling through the countryside in a train; his companion looked out the window and remarked "Those sheep have just been shorn" to which the scientist replied "Yes, on this side anyway".
On this quiet Sunday morning I built a transformer on an FT37-43 core; it had two windings, one of 20 turns and one of 10 turns.
I placed the windings on opposite sides of the core; I have previously found that winding the shorter winding over the longer gives passable results but I wanted to see the effect of separating the windings. Here it is, in a test fixture and with a 200-ohm resistor across the larger winding;
Here is a sweep of Real Z, Imag Z and VSWR;
It's very poor; indeed, I didn't expect it to be as bad as this so I made another one and the test results were almost identical.
I then made a tx-line version, using the same FT37-43 for consistency.
The new transformer used 10 turns bifilar connected as a 4:1 device; I didn't measure the line Zo but I wound it a little less tightly than the twist that I have found to give 50 ohms Zo so it's probably somewhere near the desirable 100 ohms. With a pin-connector and a 200-ohm resistor installed as per the previous model, it looked like this;
Here are the test results;
This is a very good transformer; it is wideband well beyond the 50 MHz limit of my tests.
That answers Vince's question 2; the designer, a very smart bloke, knew what he was doing! :)
Re question 1; these devices can be tested with an antenna analyser in the same setup that I used.
Terminate the "non-50-ohm" winding with the appropriate resistance and apply the analyser to the 50-ohm winding to see how good the VSWR is.
Keep connections as short as possible; you can see that I use SIL pin-connectors which are very convenient for this kind of work. In this case, a short SIL strip is soldered to a BNC(f); one SIL pin goes to centre, the next pin is removed and the outer two are soldered to the body. Silver-plated BNCs ease soldering.
If, as occasionally occurs, neither winding is 50-ohms, apply the analyser to that which is closest to 50.
12244 2016-01-30 17:30:41 Bob Thompson Re: Bifilar Torriod independant testing and also SA design question Sent in error. Pls delete
12245 2016-01-30 18:16:12 n1vc Re: Bifilar Torriod independant testing and also SA design question Hi Kerry,
Yes, your test results sure indicate that the designer went with the Bifilar for good reason. I'm very glad you went through this effort because I am going to duplicate it myself and see if I get similar results. It will penetrate my brain better for future use. I will try it with the MFJ-259B as well as PHSNA (Poor Ham Scaler Network Analyzer). I realize the latter can only read VSWR but that should suffice, do you agree? I don't think I really need to look at the Imaginary Z or reactance as if it is out of range, the VSWR should show a problem. Is thi correct? However, I may be missing something here, please advise.
I get more satisfaction from tearing into a design in this fashion and learning rather than actually getting on the air an operating these days. I will admit the SIL connection here looks strange to me but maybe it hasn't sunken in yet. I will take another look as I would not want to have my data skewed by unnecessary lengths.
Finally, my dad gave me some good advice probably some 50 years ago, which I still recall. Besides American English my dad spoke a second language which was Italian. As a kid I would repeat what he said sometimes without having a clue. He impressed on my brain to never to say something to pers
12246 2016-01-30 18:43:04 kerrypwr Re: Bifilar Torriod independant testing and also SA design question G'day Vince.
I try to keep true Australian English alive; see here for "bloke";
Now you know!
SWR is a valid test for these transformers so use what you have.
SIL is just Single-In-Line strip;
Available from electronics places and, of course, on ebay.
Get the machine-pin ones; the square-pin type used in computers etc is not quite the same and will damage round-hole sockets.
I use them in many test fixtures; my AADE meter is an example;
Pushing wire leads into the sockets will damage them; I usually solder a pin-strip to any components to be tested but, in this case, a second strip makes a "double-adaptor" so that the primary pins are not damaged.
For the transformer tests I attached a two-pin strip to the 50-ohm winding; that connects neatly to the matching socket on the BNC.
12247 2016-01-30 21:23:38 kerrypwr Re: Bifilar Torriod independant testing and also SA design question I mentioned earlier that I had obtained "passable" performance from conventional design by winding one winding over the other; that is my memory but I don't have any evidence so I made another transformer.
This one has 10 & 20 turns but the 10-turn winding is wound over, or interleaved with, the approximate centre of the 20-turn winding;
The performance is rather better than the other 10/20 design;
Not, of course, as good as the tx-line design but it would do the job at 10MHz.
It may be that this is a kind of "hybrid"; tx-line action going-on in the centre of the 20-turn winding and flux coupling going-on at the ends of that winding. I'll leave that one to the intelligent ones among us.
I think I've seen reference to a general principle that conventional transformers work up to about 20MHz; it may be in one of those Philips app notes.
12249 2016-01-31 08:38:29 vasilyivanenko Re: Bifilar Torriod independant testing and also SA design question
Thanks Kerry -- good results and big thanks for posting them. It’s refreshing to see some actual experiments posted here.
.....“Yes, your test results sure indicate that the designer went with the Bifilar for good reason” .......
You can rest assured that the author (Wes) performed these sweeps long ago – and that he plied a transmission line transformer for good reason . I’ve swept nearly every RF part and stage in my lab based on the 3 true themes of EMRFD: build and measure – to measure is to know – any measurement is better than no measurement.
EMRFD is not simulated methods , nor theoretical methods in radio design. In this day and time with all the cheap [and expensive] digital-based + analog instruments available to, we can rejoice that most can put together a decent lab without breaking their bank accounts.
Making radios primarily functions as an excuse to ply test equipment.
12280 2016-02-04 10:29:40 n1vc Re: Bifilar Torriod independant testing and also SA design question
Would you mind making a quick pencil schematic of this setup? I am getting some weird results and it has to be my setup. What else can it be?
Thanks Vince N1VC
12281 2016-02-04 13:49:04 kerrypwr Re: Bifilar Torriod independant testing and also SA design question The transformer is the "standard" 4:1 bifilar device;
There are different ways of drawing these but this is the one that I visualise when making connections.
My guess is that you have connected the "centre-tap" wrongly; of course I have never ever done this. :) :)
I wind the device, scrape-off the enamel and use a continuity tester to identify the ends of one of the pair; I then bend each end over to identify it.
Then it's just a matter of connecting one bent end and one unbent end to form the twisted "centre-tap"; it's 50 ohms from the twist to either single-wire end and 200 ohms between the single wires.
This is not the same transformer as I posted earlier but it's the same type;
This is the fixture I used for measurements;
Test fixtures are fun to design; there are many different kinds required to suit different instruments and different DUTs.
Work out something that will transition from the DUT to your analyser; lead lengths aren't too critical at HF but it's good practice to keep them short.
To check that you are using the analyser correctly, make a couple of test loads (I call them "sanity checks").
A 50-ohm resistor and another of around 180 ohms or so, fitted to whatever connector type you are using, will serve; if you don't get the SWR reading that you expect from these, you know that there is a problem somewhere.
12286 2016-02-07 13:21:23 n1vc Re: Bifilar Torriod independant testing and also SA design question Kerry,
Good results here:
Using the MFJ-259B:
Note: readings are low because all I had on hand was a 180 ohm resistor. Because it is a 4 to 1 transformer my results looks like they are right on. I am still going to do SWR sweep with my PHSNA. Thanks for the assist.
Here is my test setup:
12287 2016-02-07 17:27:05 kerrypwr Re: Bifilar Torriod independant testing and also SA design question Great result; gives you a good feeling, doesn't it? :)
12288 2016-02-07 18:36:07 n1vc Re: Bifilar Torriod independant testing and also SA design question It certainly does 12311 2016-02-09 09:41:52 n1vc Re: Bifilar Torriod independant testing and also SA design question Okay here I added a jpg from a PHSNA sweep. It never exceeds 1.2 SWR all they way up to 40 MHz, but has its lowest VSR around 10 MHz. I think it works as designed.