EMRFD Message Archive 7828
Message Date From Subject 7828 2012-10-05 06:00:07 g3xaq Output transformer for single ended PA I want to increase the output power from my Elecraft K1 QRP rig. I have an ex-laptop 15V/4A supply to hand so a 3:1 turns ratio collector transformer looks like it will offer a useful 20W RF output.
As a first step in this exercise I am trying to understand the design of the original K1 single ended PA transformer. It is made with 5 bifilar turns on an FT50-43 core so is optimised for around 6W output with a 12V supply.
By calculation from the basic physics or cheating with an online calculator such as
I estimate the AC peak flux to be 87 Gauss at 3.5MHz on an FT50 size core, which is well within the recommend 20% of Bsat (I see values for Bsat in the range 2800 to 4000 Gauss for type 43 ferrite, probably depending on how hard into saturation the material is pushed).
So far so good, but I lose the plot is when I add in the effect of around 1A average current flowing in the 11uH collector winding from the DC supply rail. Does this add a standing flux of 1650 Gauss, as calculations based on pure DC without RF suggest? If so, the core is operating way beyond the linear part of the B/H hysteresis loop and ought to get very hot.
In fact the core runs cool so I suspect the effect of the pulsing DC current bias is not the same as pure DC. Please can someone enlighten me about what is really happening, and ideally provide something numerical so I can get on with designing my own 20W PA transformer?
7829 2012-10-05 07:08:44 ha5rxz Re: Output transformer for single ended PA Taking a step backwards here, how do you know that 5 bifilar turns 7830 2012-10-05 07:52:16 g3xaq Re: Output transformer for single ended PA Hello Peter. Sorry, I was not expressing myself clearly. I only meant that a bifilar winding can give a 4:1 impedance transformation, so 12.5 ohms from the 50 ohm output load. A crude approximation (sine waves, Vce(sat)=0v) then gives 5.76W output with a 12V supply.
73, Alan G3XAQ
7831 2012-10-05 15:43:57 email@example.com Re: Output transformer for single ended PA Hello Alan,
How are you planning to obtain 20W from the K1?
72 Terry WAØITP
----- Original Message -----
7833 2012-10-06 05:56:15 g3xaq Re: Output transformer for single ended PA As a follow-up, today I breadboarded a copy of the K1 PA output transformer. It is 5 bifilar turns on an FT50-43 core. With the primary terminated in 12 ohms, plus a 180pF to compensate for leakage inductance, the response looking into the 50 ohm port was flat from 3.5MHz to 21MHz, with a return loss better than 20dB. It wasn't as good at higher frequencies but the K1 only goes up to 21MHz so that's not an issue.
When I injected 1A DC into the transformer primary via an RF choke the RL was degraded from 23dB to 16dB at 3.5MHz. The effect became smaller above 10MHz, with 2dB degradation at 21MHz. This is probably to be expected seeing as the core doesn't have much permeability up there.
So it seems to me that there is a clear effect from DC magnetisation. The question remains as to whether this static test is representative of the current pulses in the real PA and whether it matters much anyway.
I'm a bit surprised nobody has been able to shed any light
7834 2012-10-06 12:58:41 in3otd Re: Output transformer for single ended PA Hello Alan,
I'm not an expert on the subject, so I might be wrong here....
I agree with the analysis in your first message up to the part were you wrote "the core is operating way beyond the linear part of the B/H hysteresis loop and ought to get very hot".
The fact that the core is on the verge of saturation due to the DC current does not imply it has higher losses and has to get hotter than without DC. Losses are due to the BH loop hysteresis and the applied DC simply shifts the operating BH loop but I'm not sure it increases its area (=losses). Sure, this can be a problem for the PA, since it operates with a lower effective inductance value, as your experiment shows, but if the change is relatively small it might still work ok.
73 de Claudio, IN3OTD
7835 2012-10-06 15:17:26 kb1gmx Re: Output transformer for single ended PA Two things..
The final on the K1 is class C in most cases and there is no
standing DC. This means core saturation from DC bias is not
However the final used will likely push the final to expire at 20W output as even Class C 70% eff that means it going to heat the final and be far less tolerant of SWR issues.
To get more power the load impedance will need to be lower that means a 6:1 ot 9:1 (impedance) transformer with more complex assmble of the output core to get bandwidth.
The likely load impedance is in the 4-5ohm range and you need
a SQRT(50/4)= 3.5 or 3.5:1 turns ratio for 12.5:1 impedance transformation. You might try a 2:7 turn transformer with
the primary side of 2 turns. Use heavy wire and larger core
like a two hole or pair of ferrite cylinders side by side.
The problem is that as you push the load current the gain of the device falls and the saturati
7836 2012-10-06 16:43:53 Pat Bunn Re: Output transformer for single ended PA Someone finally said something sensible in this thread.
kb1gmx <firstname.lastname@example.org> wrote:
>The final on the K1 is class C in most cases and there is no
>standing DC. This means core saturation from DC bias is not
>However the final used will likely push the final to expire at 20W output as even Class C 70% eff that means it going to heat the final and be far less tolerant of SWR issues.
>To get more power the load impedance will need to be lower that means a 6:1 ot 9:1 (impedance) transformer with more complex assmble of the output core to get bandwidth.
>The likely load impedance is in the 4-5ohm range and you need
>a SQRT(50/4)= 3.5 or 3.5:1 turns ratio for 12.5:1 impedance transformation. You might try a 2:7 turn transformer with
>the primary side of 2 turns. Use heavy wire and larger core
>like a two hole or pair of ferrite cylinders side by side.
>The problem is that as you push the load current the gain of the device falls and the saturati
7837 2012-10-07 11:27:38 Claudio Girardi Re: Output transformer for single ended PA 7838 2012-10-07 14:25:14 kb1gmx Re: Output transformer for single ended PA 7839 2012-10-07 15:05:52 William Carver Re: Output transformer for single ended PA Who cares that there is no DC current when not generating RF?
Put a DC ammeter in series with the Vcc/Vdd. There is DC flowing when it
is driven. With a single ended otput transformer that is DC that does
contribute to saturation. Period
7840 2012-10-07 16:19:23 kb1gmx Re: Output transformer for single ended PA 7841 2012-10-08 03:17:59 g3xaq Re: Output transformer for single ended PA I am standing on the sidelines being educated by the discussion about my question on single ended PA transformers. To keep things on-topic I should point out that I said
> I want to increase the output power from my Elecraft K1 QRP rig.I should emphasise I said "from" and not "of". I was thinking more of an outboard PA than modifying the K1, but I wanted to know how the original K1 was designed before trying to design my own PA.
I may or may not choose to take a chance with modifying my K1 but that is a different topic, one which I would not care to defend here for all the reasons Allison has described.
To move forward with the analysis of the K1 as designed by Elecraft... If the peak collector current is around 1.5A (class C, 12V, 6W output), I think the peak magnetising field strength in a size 50 toroid with 5 turns is around 8.75 Oersted (path length is about 10mm, 1 Oe = 80t/m at 1A).
Looking at the BH curve on the bottom right of
type 43 ferrite is way into saturation at this field strength. So I am still wondering why the core doesn't get hot. This BH curve is at 10KHz, so is it not applicable at HF? I am still looking for something better than "build it, stick your pinkie on it, and if you get burned the core is too small". Maybe I am being naive?
73, Alan G3XAQ
7842 2012-10-08 08:03:21 g3xaq Re: Output transformer for single ended PA No, the tabulated data I looked at for a size 50 core had the wrong value for magnetic path length. It looks like it was out by a factor of pi. The actual path length is not 10mm but about 30mm so the Hmax in the K1 PA transformer is about 3 Oe. Type 43 ferrite is beginning to saturate at this field strength but looks like it is probably OK.
It's all beginning to make sense now.
73, Alan G3XAQ
7843 2012-10-08 15:48:02 kb1gmx Re: Output transformer for single ended PA Yes, the devil is in the details.
Generally a FT50-43 is not my first choice for more than 5W. I'd maybe try 4 stacked as single core. However there are type 43
tubes with an ID and OD in the same range as FT50-43 but nearly an inch long that are candidates. The alternate is the Ferrite blocks (two hole) like the BN43-3312 that would give you enough for up to 50W or more.
With that said I have a 40M amp running 12W @13.8V using a simple
bifilar twisted pair
7844 2012-10-09 15:20:42 Alan Ibbetson Re: Output transformer for single ended PA > What you haven't mentioned is the possible device and frequencyI was looking at the stud mount MRF450A, simply because it's quite
cheap. It has a bit more voltage rating than most of its ilk so I would
hope it will be safe with my 15V PSU, particularity if I include a
rectifier diode in series as reverse polarity protection. It has plenty
of dissipation and bags of gain, actually probably a bit too much gain
so I may need to use some collector/base resistive feedback to stabilise it.
I'd like it to cover the same range as the K1: 3.5MHz - 21MHz.
> Generally a FT50-43 is not my first choice for more than 5W.Yes, it does seem to be sailing close to the wind for anything beyond
the K1's 7W maximum output.
I have been looking at type 61 ferrite for the output transformer of
this 20W design. The Fair-rite data sheets show a core of type 61 can
carry maybe five times the amp-turns of a similar sized type 43 core.
The lower permeability of type 61 shouldn't be a problem at the 5.6 ohm
collector load I'd aim for. I had hoped a 4 turn trifilar winding on an
FT50-61 core would have enough inductance, but I couldn't get it to give
me a decent 50 ohm return loss at 3.5MHz looking into all 3 windings in
series with the primary terminated in two paralleled 12 ohm resistors,
no matter what values of compensating capacitor I tried. When I
increased it to 5 turns with 390pF across the low resistance load I saw
over 20dB RL 3.5-21MHz. I can't really explain this, unless maybe 4
turns makes the transmission line too short to be effective. Or perhaps
I just messed up making that transformer. I tried injecting 2A DC into
the primary of the 5 turn transformer via an RF choke and the RL was
degraded by only a dB or so. I don't have a higher current variable
supply to hand but the paper predictions say it will saturate less with
the 3.5A current peaks in my 20W design that the FT50-43 at 5W in the K1.
I am still a bit puzzled why most of these transformers at the 20W level
use two or more FT50 cores. Is it just to get more inductance for the
number of turns allowed before saturation? There seems to be plenty of
inductance with just an FT50-61 so that doesn't seem to be the reason.
Or is it to make the transmission line longer? Or something else that
I've not even thought of?
On a side note I've made a 16:1 base matching transformer using five
turns of two lightly twisted sets of 4-filar 28 swg wires on an FT37-43
core. With a test load of 3.3 ohms and a 1000pF compensating capacitor
it shows 22dB RL at 3.5MHz reducing to 19dB at 21MHz. I'm expecting this
to be OK without the added capacitance when driving the PA transistor.
Making this transformer was about the limit of my patience and only
became feasible when I discovered copper wire with lots of different
coloured enamels in a local craft store. Maybe two separate 4:1
transformers is the more sane approach.
73, Alan G3XAQ
7845 2012-10-09 18:43:59 kb1gmx Re: Output transformer for single ended PA 7846 2012-10-09 19:37:22 Reginald Beardsle... Re: Output transformer for single ended PA Alan,
Chapter 5 on powdered iron core transformers in "Transformer & Inductor Design Handbook" by Wm. T. McLyman might help.
It's a bit power supply oriented, but appears to cover your questions. I've been meaning to get the books that Amidon sells by DeMaw & Sevick but have gotten distracted from the joys of radio.
McLyman does note that estimating temperature cannot be done very precisely. This is primarily because of problems posed by unknowns in the heat transfer equation. However, he seems to give a pretty detailed procedure for calculating the energy losses in a transformer.
Hope this helps. I would have responded sooner, but the thread had run into the weeds before I read it.