Saturday, April 30, 2016

RG174 or CAT 5?

 20 Feet of RG174 vs 20' of CAT 5 Cable

(A back to back insertion loss test)

So my debate about RG174 as a viable TL for my portable 20/15m sloper came to a screeching halt when compared to my CAT5 TL.

I set up the KX3 with my 112 ohm baluns and measured the matched insertion loss.

First performing the CAL THRU measurement in the back to back mode. (The Red to Black convention is actually phased correctly. One balun is wired backwards).



Then I stuck the CAT5 in there. The CAT 5 suspended on plastic camp tables 18” up off the floor. And of course I can't get the photo rotated correctly...


20 feet of  RG174 yields about 0.8 dB loss at 14.060 MHz and the CAT5 has 0.3dB loss for the same 20 foot run.

A delta of 0.5dB!

Welcome CAT5 to the last ½ dB Club!

Saturday, March 19, 2016

The Park Portable Doublet System Revisited (Wrap Up)


How Much Heat Can You Generate With A 9:1 Unun And The Same Load At 3 Minutes?

I know, I know this was supposed to be about Baluns and ununs but I have taken a respite from this topic for a little while. At least until I can gather more data.


This really wouldn't be a fair comparison if we didn't find out how much loss there is in a 9:1 unun fed directly from the KX3 as if we were setup outdoors feeding an end fed halfwave antenna.

But First, A Reference

Let's get an idea of how much power is lost if you were to attach the Halfwave wire directly to the KX3, hit tune and go.





At a 9:1 SWR at 3 minutes we barely get the load to heat up. (You're gonna burn up your finals!) There is another misconception that won't go away.






Well, there you go. Not too efficient!


Now Through The UNUN




I took my best, most efficient unun and attached it directly to the rig. I have wound a few. Well, there seems to be a bit more heating there but it still seems a bit low.




Okay. Now your talking, or rather saying, "Say again, over!" We just need a politician to say that we now have a 6-dB improvement!

But What Does This Mean?

Say you are set up with an end fed halfwave over poor dirt and running 5 Watts from your QRP rig out in the park. Next, attach an antenna system that produces a -2dB gain (because the antenna itself is 0-ish dB gain.)


3.2 Watts Radiated










Now, take an efficient doublet system and couple that to the same 5 Watts.







What you will get is an effective radiated power of something greater than 5 Watts. But how much greater?

First, convert all the power levels to dBm and then add up the dB levels both plus and minus to get the result. Remember the braodside antenna gain was 7.7dB and the BLT had 0.68dB loss for a grand total of...

"WHAT?"

25 Watts




The guy on the other end doesn't know if you are running 25 Watts or not. There is no way to tell.

What this demonstrates is the ability to focus RF like a flashlight beam and aim it towards your intended target. Out here in Colorado country with poor ground, setting up a doublet in the North/South plane to favor an East/West direction is more than intended it's desirable.

What's not acceptable to me however is to throw away 22 Watts of available power when you don't have to.


72/73
WV0H Myron
Printed on Recycled Data

Friday, February 19, 2016

The Park Portable Doublet System Revisited (Part 6)

BLT Loss Verification By Calories


Using A Calorimetric Approach To Measure The MB-BLT Loss

I've always wanted to measure the power lost in the MiniBalun BLT under mismatched conditions
or the power delivered to a complex load but never really came up with a good way to do it.

Measuring the actual power dissipated in this type of load is quite hard. You can't measure it directly with a diode detector because you will alter the impedance with your attached components, clip leads and other stray stuff that will change the impedance. Likewise a 50-ohm measurement system is out. You could try a back to back measurement but when you set up two tuners to match the same resistor you will end up with an unknown power split between the load and the downstream BLT/load. The I thought why not try measuring temperature rise due to power dissipation in the resistor?

The biggest advantage of the Calorimetric approach is that nothing can change the load impedance because nothing is touching it. As well, if there is any stray anything, pointing the IR probe directly in the middle of the resistor won't upset any balance in the load. If there is a downside to all this you have to keep the resistor thermally isolated from its environment (or at least constant for a relative measurement) and attention needs to be paid to how the leads of the resistor can conduct heat away from the resistor element itself. Keeping the resistor at the desk out of air currents and with added longer wire extensions on the leads of the resistor is enough to normalize the temp rise in this case.

Recall that the predicted total mismatch loss is around 0.7dB using the MiniBalun BLT on the PPD on 20m.


The predicted loss in the BLT is 0.12 + 0.172dB = 0.29dB loss with a 4154-j2314 ohm load.

Finding A 4154-j2314 Ohm Load...Um, Yeah

I have been a collector of all things electronic for quite some time now and I found a 7W 4.7k ohm resistor in my junk box. On the AA-600 it comes in at around 4500-j1600 ohms. We can find out how close it is by using a percent error formula.


where the expected value = 4154-j2314
and the measured value = 4500-j1600
we obtain % error of about 17% using vector math on my HP-48G. Good enough given the difficulty of obtaining high impedance measurements from an AA-600 in the first place.

Power Dissipation In The Load - A Baseline

Since the real power dissipated in any load will only occur in the resistive portion we can compare the power dissipated at DC versus RF. And how do we do that? By using an IR temperature probe to measure the temp rise of the resistor.

I darkened the side of the resistor with a Sharpie to get the emissivity more inline with the fixed emissivity setting of the IR probe. I am not sure what it is exactly but since I use the same probe for both DC and RF measurements it won't matter.

Out comes the electronics trainer, glad I didn't throw it away. Also, the 200V 6W DC power supply is a postage stamp sized SIP PCB designed for Nixie tube clocks and is an eBay special (preassembled) from NS6A for about $15. Link

Setting up the IR probe and DC power supply we can observe the temperature rise at DC.






We can compare that to the dissipation at RF using the same resistor.





Temperature on left, applied DC Voltage on right.


Observe that the ratio of the powers dissipated for a given load and time are equated by the following formula. (Temperatures must be in Kelvin or Celsius, I'll pick Celsius).


Since the time is the same for both it can drop out of the equation and the amount of power dissipated either with DC or RF has the same relationship. Inserting the delta temp from both the DC and RF measurements into the Log power formula will yield the power lost due to the insertion loss of the BLT.



Results

 So what do we get when we put the power to it after 3 minutes for each test?



Inserting the numbers into the loss formula we get


So there it is, the measured loss in the BLT is 0.22dB compared to the predicted loss of 0.29dB loss, a 0.07dB difference. (It's silly reporting to hundredths of a dB, but shown for completeness, oh well).


Error Analysis

With the KX3 set to 5W it is actually closer to 5.3 Watts which make the ratios between DC and RF a bit farther apart than reported. If there is more power being applied in the RF test it makes the denominator of the loss formula smaller and the calculated loss increases slightly.

Also, the resistor has a positive temperature coefficient and up around 160C the resistance is about 5000 ohms. Given that the DC source is a constant Voltage source, the dissipated power will decrease as the resistance increases. The KX3 however will continue to deliver 5.3W unaffected by the slight change in load impedance so the power in the load should remain constant but the tuning solution of the tuner will change ever so slightly.

So given all of these pluses and minuses I am satisfied that the measurements are close enough to verify the SimSmith model.


Coming up:

The Functions Of A Balun or Unun


72/73
WV0H Myron
Printed on Recycled Data



Friday, February 12, 2016

The Park Portable Doublet System Revisited (Part 5)

Using All We Know To Predict PPD System Efficiency

Transmission Line Mismatch Loss

Since the matched loss of the balun is known and we will never use it in a mismatched condition behind the L/C matching circuit we can determine the total TL efficiency on each band by running an EZNEC model of the antenna. Then use the complex feed point impedance from the LastZ.txt file swept from 3.5 to 28.5 MHz and inserting the measured TL and selecting the matched generator using "XMtch" setting in SimSmith. The xMtch generator setting uses a matched generator for all loads without any reflections to combine with the incident wave and skew the results.

Restated, the graph below is the sum total of all the transmission line mismatch loss from 3.5 to 28.5MHz in the PPD system. It generally hangs out at around -0.3 dB with peaks and valleys occurring at -0.15dB for 40m and -0.47dB for 17m. Since we obtained the k coefficients from an open/short method, the application of these coefficients in SimSmith creates the equally valid mismatch loss model in which we can show the performance over all HF frequencies. Remember the k0 coefficient is the loss due to DC resistance, the k1 coefficient is the AC (or RF) resistive loss due to skin effect and k2 is the loss of the air between the TL wires, the dielectric loss.

Notice a pretty flat response from 40-10m except on 80m where the decreasing input resistance (14ohms) and increasing reactance (-970 ohms) causes the increased mismatch loss in the transmission line.


In SimSmith the power entering the load (the antenna) is represented by dBW and comes from the right. The "Pwr" nomenclature is misleading in the graph below because it is really the dBW level, the dot by the 14.030 MHz arrow head shows -0.33 dB/power same as the dBW reading in the load block.



The SimSmith calculation is based on the decibel reference to one Watt it also represents the loss in dB to the load. Clicking the <-dBW button will scroll through the power dissipated and power entering in from the source both in W and dBW in the block. 

This indicates the power level dissipated in each block.
In this case the load block dissipates the same amount of power as it looses because it is at the end of the line and the system is referenced to 1 Watt.


Low input impedance of the antenna on 80m makes for increased MML in the TL

Taking an inventory of our loss budget line items we can categorize them into known and analyzed buckets. Meaning we have two buckets of knowledge here: one is measured loss and is known through direct measurements. The other is loss we can calculate from a known good model we have some confidence in because of its alignment with measured results.

Measured Loss Entities:

MiniBalun
T106-6 coil/capacitor
Elecraft BL2
B&W coil/capacitor
TL loss when matched

Modeled Loss Entities:

TL loss when mismatched

What we don't have to do a complete characterization of the PPD system over the entire HF band is the loss of the BLTs or more specifically the loss of the reactive components (coil/capacitors) over all load impedances. Since I'm fresh out of an HF load pull measurement system, we could probably do a back to back measurement of each band and see what we come up with. Given the loss scenario for 20m one could guess that you would get better on the low bands and worse the high bands. What escapes the model for now is how the varying Q of the coil/cap would affect the power transfer of all HF for all the load impedances presented by the antenna itself. Maybe this summer I could try to accomplish this test.


Coming up next week:

BLT Loss Verification By Calories (Part 6)



72/73
WV0H Myron
Printed on Recycled Data

Friday, February 5, 2016

The Park Portable Doublet System Revisited (Part 4)

"Yeah, But How Do You Know The Tuners Are Efficient?"

Or better yet, "How do you know the whole system is efficient?"

Do a back to back test.

Whenever I encounter a complex matching network or system, I at least consider the possibility of connecting them back to back and measure the power difference between the source and when the devices are inline. Both the MiniBalun and the Elecraft BL2 baluns are dual core Guanella types with switchable impedance ratios, either 1:1 or 4:1.


But first, the BLTs are adjusted to a resistor that has the characteristic impedance of the transmission line. Then they connected to the TL strung up in the air.



Balanced Line Tuners (BLTs)

I adjusted the BLTs separately to give less than a 1.3:1 reading on each attached to the 765 ohm load.


Where do you get a 765 ohm load?

The 765 ohm load is actually a 1000 ohm multi-turn pot adjusted to 765 ohms. One of those blue 10-turn jobbies while using the Sark-110 to tell me where 765 ohms was at 14 MHz.  There was a very small capacitive reactance component to it but it doesn't matter. Good enough for this experiment anyway.

Sark-110 screenshot of an HF sweep, Rs left axis and Xs right axis

I attached each BLT to the open-wire line, one at each end. The MiniBalun BLT was at the rig end and the Elecraft BL2-BLT was at the far end with the DL1 attached to the voltmeter. I used 5 Watts for the test.

MiniBalun BLT with AA-600

I set up 36 feet of open-wire line using the #26 wire I use in the field. I stretched it between the deck pole and the fence in the back yard, up about 4 feet off the frozen Colorado ground. It is January after all.

This 36 feet of wire is the wire I had left over from cutting off 66 feet from the 102 foot spools ordered from Amazon. I guess 102 feet is the 'bakers dozen' of 100 foot wire spools.


BL2 BLT setup at far end with DL1 and voltmeter

The above photo is my setup at the far end with the power detector, the Elecraft DL1 attached to the unbalanced end of the BL2-BLT with a double BNC make adapter then the DC output is attached to my Fluke 89 set to auto-hold so when I key the transmitter I don't have to be at both ends at the same time to take a reading.

The wire is white. The string is white and in the picture you can't see where one ends and the other begins but suffice it to say I clipped the clip leads to the wire ends. I used the spreader and anchor as if I were out in the field to give the most representative example I could.

DL1 attached to unbalanced side of BLT

I rechecked the SWR and all is good. To establish the reference I attached the DL1 directly to the KX3 set at 5 Watts, the input level if you will. I obtained 11.367V which equates to 5.389 Watts at the rig end. (TUNE PWR menu set to 5W).

The other thing I noticed while performing this test is the lack of receive signals heard on the band. another indication that there is good balance not allowing common mode signals to enter the receiver. If it was horrible and the line radiated then I would have been able to hear many signals and quite possible work somebody. That wasn't the case, the band was really quiet giving me some indication that the TL was working as designed not allowing signals to leave or enter the transmission line.



Reference power level setup with rig into DL1

Now we relocate the DL1/voltmeter down to the other end and re-measure. Key the transmitter and wait for the beep of the voltmeter. Run down there and see what we have. (Later I got my binoculars out).

Hit Tune, read Voltage after beep

Above is the photo of the far end measurement, 10.547V which equates to 4.654 Watts!

"What? A loss of only three-quarters of a Watt?


Running the numbers of the ratio of voltages instead of converting to power and trying to decipher the diode loss and square root 2 factors, I will just use the ratio of the two voltages and use the 20*LOG(V1/V2) formula.

End to End Loss = 20*LOG(11.367/10.547) = 0.65 dB


How can that be?

That transmission line will radiate?

It will be lossy.

It's not supposed to work!


This is fun. What this tells us is that the aggregate matched system loss including the two BLTs and my lossy open-wire transmission line has only 0.7dB matched loss.

Doing a little check to see where the loss may exist let's take a look at more measured data from an earlier post. I attached two Elecraft BL2s back to back and ran 100 Watts through them, measured the loss and this is what I came up with. The columns at the right are divided by 2 to obtain the loss of one balun.

So for 20 meters, in the 1:1 position, there is approximately 0.14 dB loss for one BL2.

So we have 0.65 dB to account for: (at 14MHz)



Here are the results of the insertion loss test of two back to back MiniBaluns with the KX3 set to 5W and the DL1 doing our power measurement.

The loss formula in the spreadsheet below is =20*LOG10(D17/C17), 20 instead of 10 because we have voltage and not power to ratio.



There is 0.12dB loss in the MiniBalun alone and 0.14dB loss in the Elecraft BL2 leaving 0.246dB lost in the TL and reactive components of the BLTs, A & B.

0.12 + A + 0.144 + B + 0.14 = 0.65 dB

A + B = 0.246 dB
(the loss of the L & C components of the BLTs)


Assuming a 70/30 split because of high Q air coil versus T106-6 wound coil, we can ratio the remaining 0.246dB loss. 

A = 0.7(0.246) = 0.172 dB

B = 0.3(0.246) = 0.074 dB

So, this leaves a total matched loss of the MiniBalun BLT and the transmission line of 0.436dB on 20 meters.


Using the 20m broadside gain 7.54dBi max gain and the EZNEC impedance of 4154-j2314 ohms I put that into SimSmith for a resulting 0.56dB mismatched loss.



Adjustments to the model

Notice there is no power lost in the C block (the Minibalun) in the SimSmith model below. We know it has 0.12dB matched loss so we can add it to the total.

0.559 + 0.12 = 0.68dB





Since the balun is not seeing a load impedance away from its design impedance there will be no mismatch loss. Restated, anytime you attach a muli-band antenna with open wire/balanced transmission line directly to the output of a balun you introduce loss beyond the rated insertion loss rating. This is an important distinction between the end fed halfwave system using a 9:1 unun over a broad frequency range hooked up to either a bit of coax or an internal antenna tuner such as the internal antenna tuner in the KX3. You would be better off to ditch the balun (any ferrite device) and attach the wire directly to the output of the tuner. The tuner has iron powder cores in a tuned circuit whereas baluns and ununs use ferrite material in broadband circuit.

It is very difficult to analyze the balun mismatch loss due to the complex load impedance interaction between the antenna and the unun and the sub-atomic nature of the ferrimagnetic material. You can come close with reflective open/short methods to characterize the unun but cannot determine how it interacts with the load (at least I haven't come up with any way to do this).


Error Analysis

So none of this would be complete unless you step back from the numbers and took a look at the source of errors. The power is measured across the lower 25 ohms of the DL1 with a 1N5711 diode and a smoothing capacitor. Some error is removed from the process by not converting to and from power but using the voltage readings directly to obtain the dB ratios from input to output. That eliminates guessing what the diode loss will be for a given power level. It is convenient that the readings around the 5 Watt power level are almost exactly in step with the dBm readings, meaning 5.0 Watts = 36.99 dBm and removing 0.1 dB from that power level leaves you with 36.89 dBm = 4.886 Watts. A delta of 0.11 Watts. So to have accurate readings within 0.1 Watt or 0.1 dB seems like a tall order for the class of test equipment that I have.

Power Measurement

To help in the verification process I do have an HP-432A with a 478A thermistor mount to do some comparisons. My DL1 and HP power meter are pretty close but are limited by the needles widths, parallax (although it does have a mirrored scale), and resolving to tenths of dB is difficult at best, whereas the Fluke 89 resolves to 3 decimal places. But again, I am stretching the limits of accuracy here and have come to realize that I have done the best I can to characterize the situation given the test equipment constraints at my disposal.

In this post, part 4 highlights the efficiency of the PPD's open-wire transmission line in a matched condition. It is important to have measured results align with predicted results because that correlation is needed to gain confidence when we analyze the mismatched case where I cannot measure the complete system efficiency directly and have to rely on predicted results.

Other Thoughts

Through the course of power measurement and insertion loss characterization of the back to back baluns I discovered that since the AA-600 generates a square wave signal at its output it produces inaccurate power readings at the other end of what every you are measuring. For instance, when measuring the 2 MiniBaluns back to back, the power loss was measured to be 0.8dB. This for a while was my insertion loss baseline. It wasn't until I characterized the loss using my KX3 as a source and I couldn't reconcile the results between the MiniBalun and the BL2.



After running the AA-600 through my 20 meter bandpass filter, all readings fell into the same range as the BL2 power test. Both MiniBalun and BL2 baluns are about dead equal.


Coming up:

The Park Portable Doublet System Revisited (Part 5)

Using all we know to predict efficiency when an antenna is attached


72/73
WV0H Myron
Printed on Recycled Data

Saturday, January 30, 2016

The Park Portable Doublet System Revisited (Part 3)

"Yeah, But That Wide-Spaced Transmission Line Radiates!"

No. Balanced transmission lines don't radiate enough to contribute to the far field pattern. True, they do radiate a little and in practice at levels way down from anything destructive to the pattern or constructive to the far-field radiation. At most, they can upset the pattern due to coupling of the TL wire adjacent to the antenna wire thus skewing the pattern but they don't contribute to the total radiation.

Background

Electromagnetic radiation results from a charge acceleration (AC current) imposed on a conductor. A balanced transmission line (TL) does not radiate because of the equal an opposite current that exists along the TL. It only radiates if the load impedance at the other end is not balanced creating an imbalance of current along the TL. This imbalance is not cancelled out at any point along the TL where is produces a net charge acceleration and it radiates.

Modeling Difficulties

I ran an EZNEC model of the Park Portable Doublet antenna's TL, 33 feet, 825 ohm, #26 copper wire, balanced transmission line. The results show that the radiation from the structure is a result of the small pieces of horizontal wires used to anchor each end of the TL, these little horizontal wires radiate. So, EZNEC may not be the best way to show what could happen. I just showed the results for grins below. The worst case radiation occurred at 6m. (-37dBi at the green dot)


Spreadsheet

I found this spreadsheet on the internet that has a formula in it to show the amount of radiated power you get from an open wire line written by KM9O, Dennis and published in QST. I am not quite sure the origin of the fancy formula used to calculate the power radiated by the TL, but I'll take it.

P(rad)=160*(((SQRT(E4/(276*(LOG10(2*E6/E7))))^2)*((3.14*((E6/12)/(968/E5))))^2))

The radiated signal on 6m is 26dB down from the main antenna radiation and that's 6m. On 20m your mainline radiation will be at least 35dB down. In any case, hardly a contributor to the far field radiation pattern.


The bottom line about open wire transmission lines is this: They don't radiate if you keep the currents balanced and that is not hard to do if you pay reasonable attention to the deployment of the antenna and design/construction of the matching circuit.

On-Air Testing

I intended to perform a field test using the Reverse Beacon Network to establish a baseline SNR with the PPD. Then switch quickly to another setup of just the transmission line without the antenna attached, to see if I could still radiate a signal with just the balanced open-wire feeder.

I have to work out the matching scheme to do a quick A/B switchover while still presenting the same power level to each setup. I did find out that my 100-Watt Kenwood TS-2000 will put out 5 Watts (when set to 5W) at any load impedance, (except into a short or an open) so I may not have to mess with retuning for the experiment.

Software Model Verification

No analysis would be complete without looking at where the errors can occur in a modeled system, after all, garbage in garbage out.

One way to know if an EZNEC model is accurate is to do a convergence test. Set all losses equal to zero and look at the "Average Gain" at the bottom of the main window. Cebik outlined this (and other methods) in his multi-article series in QST while back. If there a greater than 0.1dB discrepancy, set the "Ground Type" to "Free Space" and the "Wire Loss" to "Zero" then adjust the wire segmentation until it converges on a zero dB gain. You have to re-run the FF Plot each time you adjust the segment length to re-compute the result.


Here the average gain is displayed in both linear and decibel units. You would ideally like it to be zero but in practice there is a trade between the number of segments and processing speed, you have to settle on something that is close enough and go with it. Since I try to stay within 0.1dB, my modeled results come within 5-10% of my measured results out in the field.

825 Ohm Transmission Line Using #26 Wire

Establishing transmission line loss is tricky, but with the right tools you can obtain reasonably accurate results using any Rigexpert AA-54 or AA-600 series or Sark-110 analyzer to measure and store the open and short circuit impedances. But first you have to obtain a reasonably long run of TL, around 100 feet using HF frequencies. For me that meant I needed two runs of 100 feet #26 wire spaced 8.25 inches apart. I say 8.25 inches apart because that is the spacing of the PVC spreaders I cut to hold the TL in the back yard on some photographic light-stands.

805 Ohms Maybe?

I wondered about the origin of my "825 ohm TL" and remembered using Sevick's Two-Wire graph in his hardcover balun book (and quite possibly the textbook formula for balanced line) and thought about rerunning the numbers using KM9O's Two-Wire calculator spreadsheet. I got answers closer to 800 ohms than 825 as the the spacing and wire diameter just won't support a higher Z line result. I'll have to dig up the Zplots file to double check, but it seemed to me that 800 sounded about right now thinking back on this. I guess I got the spacing numbers confused with the impedance of the line, and things like that seem to stick in my brain.




The wire I use has more strands then the "normal" 7 stands of #34 wire, I use (19/38), 19 strands of #38 wire I got on Amazon in two 100 foot spools for $26. The increased stranding makes it more resistant to wear out failures due to repeated stressing and flexing and the insulation of PVC seems to hold up when reeled and unreeled from the chalk line reels. Since air fills the gap between the wires, there is no need for Teflon or fancy low loss wire insulation.

The secret sauce in the original spreadsheet results cell (not pictured) is Zo=276*(LOG10(2*F32/F33)) where, F32=8.25 and F33=0.02, which is the text book formula for characteristic impedance of two wire line with a dielectric constant of 1. To account for velocity factor of 0.95 of the PPD TL, we need to modify the formula where the vf = 1/ sqrt(dk) is fit into the formula, =F34*276*(LOG10(2*F32/F33)) where, F34 is the vf = 0.95.



So there it is, 765 ohms characteristic impedance. I've been using 825 and I will need to adjust my future calculations to reflect this change. It may not matter much but we'll see.

Okay, Now Breakout SimSmith

In order to correlate measured results with modeled results we need to establish a baseline about the matched loss of the TL. This line has matched line loss of around 0.12dB for 31 feet at 14MHz as shown in the screenshot below.


Since TL is slightly reactive, an adjustment of the X in the load will tweak the SWR a bit but it's hardly worth the trouble. So, for completeness I adjusted the reactance of the load to -j2 ohms. The SWR (blue line) came down just a tad.


Coming up:

The Park Portable Doublet System Revisited (Part 4)

"Yeah, But How Do You Know The Tuners Are Efficient."

72/73
WV0H Myron
Printed On Recycled Data