|
Components and Circuits For discussions about component types, alternatives and availability, circuit configurations and modifications etc. Discussions here should be of a general nature and not about specific sets. |
|
Thread Tools |
14th Oct 2016, 7:19 pm | #61 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
Sorry Jeremy I'm totally confused.
Are you trying to model the transformer as a transmission line. Or is transmission line the coax cable feeding the Transformer. please can you Confirm what is Zo . I had assumed that Zo was the Input impedance of the primary and had expected this to be the same as output impedance. 50 ohm Or is Zo the characteristic impedance of the transformer made up by the complex summation of the R XL And XC of the primary With R being the DC resistance XL the inductive reactance of the primary (9.2uH I calculated) XC the capacitive reactance of the primary Chris
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww |
14th Oct 2016, 7:41 pm | #62 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
Please can we start again and take the design process step by step. Hopefully we can design it model it and test it to see if the theory works in practice. I have become lost over the last 61 posts.
Specification. Bandwidth 3.5 - 30 MHz. The transformer will be fed from a 50 ohm source and will be terminated in a 50 ohm load. We want to aim for flattest swr over the band. So we need to consider how this affects core choice and winding style and lead lengths etc. Im going to just do the first step and we can comment on any error or adjustments required as we develop the design step by step. 1) For optimum transformer efficiency primary reactance should be 4-5 the impedance being transformed at the lowest frequency. Impedance of the Primary will be same a secondary as the transformers 1:1 = 50 Ohm Primary reactance = 4 x 50 = 200 OHM So L = XL / (2*Pi*f). If XL is 200 Ohms and f is 3.5MHz, L = 200 / (2*Pi*3.5*10^6), or 200/(2*Pi*3.5) uH. L = 9.09uH. The inductance of the primary winding will be 9uH Do we agree ? best regards Chris
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww Last edited by ChristianFletcher; 14th Oct 2016 at 7:50 pm. |
14th Oct 2016, 8:37 pm | #63 |
Guest
Posts: n/a
|
Re: 1:1 RF Transformer Design SWR problems
|
14th Oct 2016, 10:08 pm | #64 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
To clarify, I was trying to give a real world (and very simple to understand?) example showing where it can be beneficial to parallel up transmission lines to reduce Zo. It was a separate example involving a 4:1 quarterwave matching transformer from 12.5 to 50R. The ratio is 4:1 and the OP isn't looking a for a 4:1 transformer he's looking for a 1:1. The fact that it is a narrowband 4:1 design is irrelevant to the fact that if you parallel up two 50R tlines you can mimic a 25R tline. The advantage in using the optimal Zo for the line is that you get better performance.
__________________
Regards, Jeremy G0HZU |
14th Oct 2016, 11:40 pm | #65 |
Dekatron
Join Date: Aug 2003
Location: near Reading (and sometimes Torquay)
Posts: 3,086
|
Re: 1:1 RF Transformer Design SWR problems
I just found this thread and have not yet had time to read it all carefully, but I thought I should quickly mention this...
If I am reading the photo correctly, maybe it is wound wrong. Not sure how big an effect this makes but it is not good. The classic mistake with toroids is to wind them round the toroid until you get back to the start and bring the wire out near the start point. This creates a hidden single turn that is wound round the outside of the whole thing. It means the field is not contained and thus will interact more with the surroundings than expected and also may generate more dead field in the core material which could create more loss. There are various ways to fix it. One is to do as above then take the wires back round the core to the opposite side, thereby hopefully cancelling out the big loop. Another is to work the windings round the toroid then reverse back to the start. There are others. Just a thought. |
15th Oct 2016, 8:48 am | #66 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
Thanks
GMB: Yes I did start and stop my primary and secondary from the same end of the transformer. I guess this may create a spot where the winding spacing is a bit irregular but I would assume this only cause very minor issues with the distribution of the flux. I could be wrong.... Did you mean to link in another picture when you refered to the above or was this my post. Please could you add to your previous comments as this is something I had not considered but as you said could be important and was interesting. Thanks
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww |
15th Oct 2016, 10:57 am | #67 | ||
Rest in Peace
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,872
|
Re: 1:1 RF Transformer Design SWR problems
Quote:
For the classic current-mode 1:1 balun it is obvious that the windings form a transmission line as they simply continue the transmission line either side of them. The winding characteristic impedance should match the system impedance - usually 50 ohms - because the winding length may be a significant fraction of a wavelength. Using twisted pair it is possible to approximate 50 ohms. For a 1:1 voltage balun it is far less obvious (to me, at least) what, if anything, should be the characteristic impedance of the windings. Essentially, you have exactly the same component as the current-mode balun but wired into the system differently. Note that the idea of characteristic impedance for the windings may cease to have much meaning (for both types of balun) if each turn has significant coupling to nearby turns. Note that in the usual current-mode case of tight twisting and spaced bifilar winding, each turn is tightly coupled to its neighbour and only loosely coupled to others for differential mode (i.e. transmission line mode) currents; it is of course tightly coupled via the core to all other turns for common-mode currents. Your parallel winding technique means that this is not true, so I am not certain that it is helpful to speak of a characteristic impedance. My guess is that for a wideband voltage transformer you need sufficiently high core permeability so that you can get both sufficiently short winding length that the length can be ignored at the highest frequency and sufficiently high inductance that the shunt impedance can be ignored at the lowest frequency. This is not easy to achieve, which may be why wideband baluns often seem to use current mode (so winding length is not an issue). Quote:
|
||
15th Oct 2016, 11:25 am | #68 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
Thanks Dave much food for thought. I am hoping to stay away from transmission line theory as much as possible as I think it probably requires a more in depth knowledge and also better maths than I have. I was rather hoping that there were some rules of thumb that I could apply to get somewhere near. Its very possible as other have said that in my ignorance I have just not understood the complexity of the task I have set. However I'm not going going to let that stand in my way just yet.
Reading through the post it does appear that opinion differs quite greatly regarding how this task should be approached if at all. I assumed winding a rf transformer to be some common place activity as could be widely agreed. Obviously not. Without wanting to miss quote Jeremy I think he was saying in his earlier post that one of the problems with my transformer design is that the physical length of my winding was causing issues. I had also thought that wire lengths etc were not significant at HF frequency as the wavelength were physically much greater i.e 60cm wire Vs 80m. Please could Jeremy clarify this for me again. If I want to reduce the length of the winding I need a more inductive material ? I have had a look at the data sheets on the spectrum communication website and I have noted that they have a TYPE 43 ferrite material this looks to have a AL figure an order of magnitude better than the Type 2 dust core I have been using. I have also received a PM regarding this material and having done a bit more reading I'm inclined to agree. Apologies if some of these questions are becoming repetitive. For me this is a complicated subject so it takes a long time for the information to sink in. Regards All and thanks for all the contributions. Regards Chris See Torroid datasheet attached.
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww |
15th Oct 2016, 11:48 am | #69 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
I bought a larger core when I was at Newark the other week this has a 60mm od. I am not sure what material its made from but it was being used in 9:1 1K matching for driving end fed wires. I have done the following test of turns Vs inductance. My question here is are the inductance measurements valid as these are done at 200Khz. Does the inductance of the core material change with frequency ?
In the past I have always bought cores however I feel bit foolish not knowing the specification of whats in my junk box. 25 Turns 800 uH 10 Turns 143 uH 9 Turns 116 uH 8 Turns 92 uH 7 Turns 71 uH 6 Turns 52 uH 5 Turns 37 uH 4 Turns 23 uH 3 Turns 13 uH 2 Turns 6.0uH 1 Turns 1.55 uH
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww |
15th Oct 2016, 12:09 pm | #70 |
Heptode
Join Date: Dec 2014
Location: Limerick, Ireland.
Posts: 901
|
Re: 1:1 RF Transformer Design SWR problems
"Does the inductance of the core material change with frequency?"
It can. Also the Q can change a lot (not just inherently) Different core materials are suitable for different bands. There is a table somewhere. If there is 100% coupling, then inductance is proportional to turns squared. At higher numbers of turns the capacitance and self resonance can be an issue. |
15th Oct 2016, 5:15 pm | #71 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
Quote:
To give you an example, I've downloaded MiniCircuits' 4 port s parameter model for the ADT1-1 transformer. http://194.75.38.69/pdfs/ADT1-1.pdf This model works up to 700MHz because this is the limit of their s parameter model. This model treats the transformer as a 4 port 'black box' and produces an accurate 'steady state' model for the transformer that can be used on a simulator. This isn't my model, it's theirs taken with very expensive test gear in a precision test jig in their laboratory. You can see how much data is in their model in the notepad text window in the video and it covers 10kHz to 700MHz. Note that they have NOT used a tline model here, this is a completely different type of model using 4 port s parameters. But have a look in my hasty youtube comparision video below. I've tried to produce my own replica model of their transformer using nothing more than a short transmission line. I've even toyed with changing the complex impedance at the output ports but you can see that the two simulations agree very well. Could this just be chance? The equivalent of a lottery win? Or maybe it is possible to crudely predict the behaviour of transformers using a transmission line model. Note that I don't have the real Minicircuits transformer here, I've replicated a crude working tline model of their transformer in just a few seconds and you can see that it does pretty well https://youtu.be/o-UrkRlBtS8
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 15th Oct 2016 at 5:25 pm. |
|
15th Oct 2016, 6:05 pm | #72 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
In case anyone thinks that things are different at lower frequencies with bigger cores and windings etc then have a look at the second youtube video below.
Here at home I'm lucky to have a full 4 port VNA and 4 port Ecal allowing me to take very accurate 4 port models of components in a similar way to Minicircuits in their laboratory. So in the youtube video below you can see me compare a 4 port model of my big ferrite core transformer that has windings about 1.4m long against a crude tline model with a Zo of 50R up to about 50MHz. It looks good to me. Am I doing something wrong here and am I just lucky with my models every time? https://youtu.be/xgovVxtFDU0
__________________
Regards, Jeremy G0HZU |
15th Oct 2016, 6:11 pm | #73 |
Guest
Posts: n/a
|
Re: 1:1 RF Transformer Design SWR problems
Stacking two or more cores is one way to increase the al, without having much more wire. I would still go for the current balance method.
|
15th Oct 2016, 6:23 pm | #74 |
Rest in Peace
Join Date: Sep 2008
Location: Solihull, West Midlands, UK.
Posts: 4,872
|
Re: 1:1 RF Transformer Design SWR problems
My basic point was not that you can't use a 'sideways' transmission line model for a voltage mode balun - I don't know whether you can or not. You may well be right in saying that you can. Something I need to think about some more.
My point was that a transmission line model for a balun assumes that at a particular point in a wire there is only 'transmission line' type coupling to the one and only one corresponding point in the other wire. Winding using coax guarantees this; bifilar winding approximates this quite well; a pair of separate windings almost ensures that this can't be true. Consider a 'transmission line' in which at each point in the line there is reasonably strong coupling not only to the return current in the other conductor but also to points a few inches ahead and a few inches behind the point in question. Does that still behave like a simple transmission line (characterised by impedance and time delay alone) or does it have more complex behaviour? The Minicircuits examples you give do not bear on this, as I am sure they have wound them properly. |
15th Oct 2016, 6:39 pm | #75 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
I agree with the points you make about the subtle issues with inter turn coupling etc. What really goes on deep inside the transformer will be very complicated. But my (our?) goal here is to make a behavioural model and I think that the crude tline model with a set Zo is sufficient to achieve this up to a quarter wavelength. It might be possible to find cases where this won't be correct, but I think that as long as the transformer windings are wound sensibly it is probably going to be a useful model
What it can't do is model what happens in the murky world where the transformer winding becomes longer than about a quarter wavelength. Up here I think all the stuff you describe will mean that my simple model will collapse hopelessly. But luckily it isn't that important here. The 4 port model will cope much better but models like this don't say much about the physics inside the 'black box' that they model.
__________________
Regards, Jeremy G0HZU |
15th Oct 2016, 6:51 pm | #76 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
Quote:
The easiest (steady state) case for me to understand is the classic split load as below. The best frequency to analyse it at is at the QW frequency where it will have an easy to calculate amplitude imbalance at the 25R loads if the tline Zo is not ideal. This imbalance can be predicted with a calculator in a few minutes or in a few seconds with an excel spreadsheet. Obviously, if the transformer was behaving as a lumped transformer then it should remain balanced. But it is the transmission line behaviour that causes the problems and I think it can be modelled and predicted just using a few simple sums
__________________
Regards, Jeremy G0HZU |
|
15th Oct 2016, 7:10 pm | #77 |
Heptode
Join Date: Dec 2003
Location: South Yorkshire, UK.
Posts: 900
|
Re: 1:1 RF Transformer Design SWR problems
Well Thanks too everyone. The impossible does now looks quite possible.
I have wound a new larger toroid and I think it could be type 43 material given the much larger inductance over the type 2 that was measured. I have wound this using 7 Turns RG174 coax. I used the inner core as the primary and the outer shield as the secondary. (Thanks) In terms of SWR and usable bandwidth it looks much much better. I may have to lower my expectation on usable bandwidth but given that I don't work the HF bands above 18 Mhz it looks good to go. I would need to consider the power handling and the electrical isolation but I think this certainly looks so much better than my first effort. This was just a quick dirty test but I think this proves its possible and no doubt I could optimise the windings and material to do even better. Im still not really sure why this works so much better. I can only assume as said in one of the earlier posts that the mechanical length of the winding is an issue. I did also shoot for an reactance of 5 x the impedance rather than 4 as was my first core design. Chris
__________________
Every Silver Lining Has Its Cloud https://youtube.com/channel/UCvBpiuUUnErJlNBm6DWb3Ww |
15th Oct 2016, 7:36 pm | #78 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
If you measure the length of the line you used you can model it as a transmission line just as in my simulations. The VSWR degradation you can see as you go up in frequency should be predictable in the model
It works better mainly because you have shortened the line length but also the change from a Zo of maybe 80R to 50R has helped too. The transformer that David described back in post #52? using RG58 might be a better choice for high power. Obviously, you also need a core with a higher permeability than your original transformer that used the T200-2 and it looks like you have done this too.
__________________
Regards, Jeremy G0HZU |
15th Oct 2016, 7:39 pm | #79 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,799
|
Re: 1:1 RF Transformer Design SWR problems
You have an ATU downstream of the transformer, so a relatively small adjustment of tha ought to put the impedance at your transceiver close enough to an ideal 50 Ohms for the higher bands to be useable.
The ferrite core keeps the turns count down and that helps with stray-C and self-resonance, and it keeps the line length down. I think you have a solution. David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
15th Oct 2016, 9:08 pm | #80 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: 1:1 RF Transformer Design SWR problems
I'm not sure how Remote ATUs operate, (never used one) but I'm going to guess that the (auto) sensing and tuning all happens near the antenna. If so then I can't see how it can tune out the mismatch in the transformer because it's in the wrong place? However, I may be missing the obvious here because I still don't know what the overall system really looks like. If you have manual control of the ATU then you could do it?
I'll also comment that the VNA plot looks a bit odd on the Rs response. The rising Rs with frequency looks wrong to me. I'm going to guess that a short test cable has been added inline after the cal at the VNA. Otherwise, I'd expect to see Rs stay close to 50 ohm in this test?
__________________
Regards, Jeremy G0HZU |