OK, I'm just sure I am thinking wrong so I am asking for your solution to my little circuit analysis problem.

I am connecting my 50 ohm signal generator through a DA121/U matching circuit to my receiver. I am trying to figure out the correction factor from the generator setting to the antenna terminal voltage. The DA121/U is a simple circuit. The input connects to the generator and the output to the receiver. To keep things simple, assume that the receiver input impedance is 125 ohms.

The DA 121/U circuit consists of a 68 ohm resistor across the input terminals and a 100 ohm resistor from the input BNC center terminal to the output BNC center terminal. The input and output BNC outer terminals are tied together through the metal box they are mounted in.

So what could be simpler . My generator is a Boonton 102C and it reads voltage correctly when the generator is driving a 50 ohm load. By "correctly" I mean that the load voltage and the meter reading are one and the same.

The question is, when I set the generator to X volts, what is the voltage at the receiver antenna terminals? Again, assume that the receiver input is 125 ohms.

I have my own answer but I don't like the measurements I am getting so I am looking for confirmation.

Let's ignore reality for the time being and not worry about cable impedance (I am using 50 ohm coax) or loss, whether the resistors in the DA121/U are good or inductive or what have you (assume they are pure resistors), or whether the receiver input is really 125 ohms resistive or not. Kind of a classroom problem. I can say that someone who should know better called the DA121/U an 8.9 db device and I am pretty sure that is bunk, no matter how you interpret it.

PM me with your answer if you don't like taking tests in public. I promise to reveal my answer when the smoke clears and I also promise to be truthful. I have lots of practice at being wrong.

Tony

First, the generator sees 68 ohms in parallel with 225 ohms = ~52 ohms (close enough....)

Once everything is hooked up, the attenuation is 125 / 225 = 0.555 (~5dB)

Thats all right. Looks like nobody's biting anyway. Sometimes I get brain-lock. After I posted I realized that all I had to do was load it with a resistor and measure it to confirm my math. I was close. I calculated .555 as the factor to multiply the generator reading by to get the actual voltage at the receiver. Thats about what I measured with a 130 ohm resistor. That comes out to about -5.1 db.

So I'm left with tying to unravel why my R-390a sensitivity measures substantially worse when I use this matching device as opposed to my other method. That method is to use a 100x (40db) attenuator with a 50 ohm input and a .5 ohm output impedance. With such a low impedance driving the receiver, I am forcing the input voltage to be 40db less than the generator setting, no matter what the input impedance is (I am swamping the input Z). Measured that way, I get excellent numbers. Measured with the matching device the numbers are substantially worse. This is forcing me to think about what is actually going on. I think that part of the issue is that the receiver is sensitive to the voltage at the antenna terminals, not the power. So, when I "peak" the antenna control, I am probably just raising the input impedance to a high value rather than "matching" anything. Higher input Z means more voltage on the antenna terminals. Matching gives maximum power transfer but that will be at a lower input voltage. I haven't sorted it all out yet but thats got to be part of it.

I have tried adjusting the R-390 antenna input for a nearly constant input Z (R actually) of 125 ohms and then measuring the sensitivity (using the DA121/U circuit) without touching the peaking control. It is pretty bad. Some bands went from .5 microvolts up to 2.5 microvolts. All bands are at least 2x worse (I haven't actually finished measuring all of them yet but that's the trend).

Looks like I doubled there. Great that someone else got the same answer and the logic is exactly the same. Thanks for that. The harder I look at this sensitivity measurement stuff, the less I know. I think all we end up measuring is how well the receiver hears a particular signal generator . I don't think that many receivers actually have a constant input Z and the antennas they are connected to (at least for SWLs) are all over the map too. The end result is just a bunch of numbers.

I experimented with a hat full of torrid cores, until I found two that would give me the bandwidth I wanted, to check the sensitivity of my SP600 (95 or 125 ohms, depending on what tech info you believe).

I wound the transformers so that I could, as near as possible, match the 50 ohm output Z of the generator, to impedance of the receiver.

Assuming negligible loss in the transformer, the generator output meter, in dbm, would also be the same dbm at the receiver's antenna input. This also assumes that the impedance is as specified and remains constant at all frequencies checked.

I used a calculator, programmed to convert dbm into micro-volts and with the assumptions mentioned above, the SP 600 measured consistent and between 1-2 micro-volts at the low, mid and high end of every band.

I can't be sure of the accuracy, however, until I devise a way to check the receiver's input impedance, at the test frequencies used. However, the consistency, is encouraging.

Charlie

Charlie,

The method I used for the "input impedance" was to connect my RF millivoltmeter across the receiver input terminals and drive a strong signal (so that I get a meter reading) into the receiver. I do this with the DA121/U attached. For the R-390a there are 5 coil sets. For the antenna coil in each coil set, I adjust the inductance at the low end of the band and the capacitance at the high end of the band. I adjust for a meter reading that is about .555 times the generator setting. I have to iterate back and forth between the top and bottom ends of the band. When I am done, the input Z is fairly constant across the whole band (maybe 30% variation). Then I measure sensitivity with the DA121/U in there. This is the best I can do without an impedance bridge of some kind.

This is not without it's issues either. I found that the best way to set this was with the receiver off. With the receiver on, the strong signal would overdrive the RF stage and the input impedance would drop dramatically. Turning on the AGC brings it back up part way. The smaller I make the signal, the less pronounced this effect is. More opportunities to screw it up.

Tony

I just did a quick search of my test fixtures and could not find the adapter to convert 50 ohm unbalanced to 95 impedance balanced. It was nicely machined but, inside were 2 or 3 resistors, some in the ground side, to do the matching. Might have been alright for a receiver input not tied to ground, but what if the receiver input had a center-tap to chassis?

I didn't like them, and the built in loss meant I would have to read the generator's output meter and add the attenuation of the adapter. So I built TWO identical fixtures, using torrid transformers. Primary isolated from secondary. I connected one adapter at the output port of my 50 ohm signal generator, the other fixture (50 ohm) to my r.f. sampling voltmeter. (the transformer/adapters= 50:95:95:50) This way the voltmeter's 50 termination properly terminated the setup. I swept the frequencies to cover what I needed for the test. The loss was very small and after experimenting with various torrids, flat in response, over the range I was testing.

Naturally, I had to use a signal level that the r.f. meter could read, but that was only in testing the transformers, for loss and flatness.

I ASSUMED that the performance would hold at the micro-volt level.

Resistive matching devices, attenuators and terminations can be made very flat. I just didn't like the way my (original) adapter was designed and being lazy, didn't want to factor the loss at every measurement.

Other things to consider is cable impedance. At 30 Mhz, almost any cable of length will fudge the measurement, if it is not the correct impedance. Also, r.f. leakage past generator attenuator, cabling etc.

I had this problem when I was alining a couple of R391s a few years ago.

The local surplus store has, at times, the twin-ax connectors for the R390 and I see 125 ohm twin conductor cable (meant for old computer network, I think) that might be suitable.

The prices for the connector, LAST TIME I CHECKED, was way cheaper than the store in Lima.

If you have a need, let me know.

Charlie

I try not to do math after 10PM Years ago it might have been due to adult beverages but now I just start falling asleep without any help

Carl

Carl, I can appreciate that. I have to watch out when I think it is "obvious" or "easy". Thats when I get it wrong. I thought this whole sensitivity measurement thing was going to be so much better when I really split the hairs and did it right. It's not going so well. I should just go back to slapping the generator on the radio, maybe with a terminator, and going to town. That gave numbers that I could brag about. Most people are smart enough to stop when they get the answers they like.

Tony

I am sorry if my dissertation was boring. My goal was to be able to, in essence convert my signal generator to an impedance to match my receiver, so that I could read sensitivity in dbm the way I saw it done at NASA tracking stations.

I have the equipment, I think, to successfully measure adequately the input impedance of my receiver, for greater accuracy, but, time, bad back and other projects will delay this from happening for a while.

Nearly all r.f. spectrum analyzers are calibrated and used in primarily in dbm and I think this way has its merits.

Splitting hairs is second nature to a (former) PMEL tech.

Charlie

Not boring to me Charlie. I'm sorry if I said anything to make you think so. I have thought about using an impedance transformer myself. I have the cores to do it. It's still on my list of options, and I will get to it. Right now I am confused by how much variation I am seeing between methods that I thought would be pretty much the same. Obviously I am missing an important piece of understanding somewhere. Probably more than one piece. I appreciate your comments so keep them coming.

Tony

The first problem is that without knowing what the receivers impedance is at the frequency used, the signal generator attenuator setting and meter level won't accurately tell you what signal level is at the antenna terminals.

And as you said, over-driving the receiver, to get an indication on a r.f. milli-voltmeter, changes the receiver's impedance, is not a solution.

Perhaps, with the signal generator connected to the receiver, (30% modulation), with a tee adapter. And switch the avc OFF, monitor the receiver's audio with an a.c.voltmeter. IF changing the signal generator 3db causes a 3db change in the audio output, you will be able to use the receiver itself, as a r.f. voltmeter.

Adding and removing known terminations at the tee (antenna terminals) and noting the signal change, should provide information for a mathematical solution as to what the receiver's impedance is.

Break out the calculator. I don't do math after 2200 hours either.

I am sure there are better (or other) ways of doing this, but this is off the cuff. I plan to make a bridge from the Radio Amateur's Handbook, but I need a second receiver as a null detector and all my radios weigh too much to bring together conveniently. Someday.

Good luck,

Charlie

Charlie, I like that idea. In fact, you can tell if the impedance is low or high by looking at the swing in db at the output of the radio compared to the swing of db at the generator. Seems like that could be turned directly into a correction factor.

I think I figured out one of my issues with the "constant impedance" approach too. If constant impedance does not coincide with resonance, then you lose. So it seems to me that you adjust the antenna coils by the book and then you get what you get for input impedance. You just have to know how to correct the generator reading to get the actual voltage at the antenna terminals. But I think you have provided a way to get a handle on that too.


Tony

Have you looked into field change 5 for shipboard receivers? This allows the
use of the unbalanced input and supposedly provides a better match to 50
Ohm coax.

73, Roger

Thanks Roger. I'm not sure which mod that is but I know of two, just not by that nomenclature. The first one just grounds one side of the balanced input. I have the adapter that does that. It doesn't change the input impedance, it just provides a convenient connection for BNC cables. The second swaps around the connections on the back of the antenna relay and may be the one you are talking about. I've done that before but decided to return the radio to standard.

As for my brilliant extension of Charlies idea, i don't think it works as I thought. Charlies idea of swapping terminations is good though.

FC 5 is actually both of them. Swapping the cables just allows use of the
unbalanced input for coax.

When using the shorting connector on the balanced input, you ARE
changing the input impedance of the receiver because you are only
using half the turns on the input transformer. Therefore, the 125 Ohm
input becomes 31 Ohms....a better match for 50 Ohm cable.

Actually, the input Z wanders around with frequency. The 125 Ohm
figure is a nominal value. I'm not sure just what you are trying to
accomplish but I think you are chasing your tail to some extent.

Here is an article that might interest you:

http://www.jamminpower.com/main/noise.html

Regards, Roger

Hi Roger,

I'm not sure I understand the part about shorting out half of the transformer. The input transformers on the 390a are not center tapped. One side of the input winding goes to one pin of the twinax connector and the other side goes to the the other pin. There is a pair of small caps on there that is used for balance adjustment and you short half of that balance network out when you ground one side. That does have an effect of reducing the input impedance but it's pretty small with the capacitance increase from 3 or 4 pf to 7 pf or thereabouts.

As for what I am trying to accomplish, it is to measure the sensitivity while taking into account the fact that the input Z wanders all over with frequency . The assumption that the input Z is 125 is wildly wrong (I've seen it up at 700) sometimes. The other part is just to decrease my cluelessness factor by a small amount. In the process of doing this I am getting stuck in various blind alleys and making various wrong deductions.

I don't mind a little tail chasing one in a while, if the end result is that I learn something new.

I remember reading about the change you are describing and it made sense to me at the time. I might even have it in my R-390a folder.

Regards to you as well,

Tony

Edit:

I did find this change. What it does for "shipboard use" is to allow the use of J103 (unbalanced connector) instead of J104 (balanced connector). One side of the input transformer primary is grounded by grounding one pin of J104. Then P205 and P206 are swapped on the back of the antenna relay block. That connects the other side of the input transformer to the center pin of J103. In this configuration, the unbalanced connector uses the same path into the radio that the balanced connector used to use. The only effect on impedance should be the small increase in capacitance across the input connector that I mentioned. At 30 Mhz (if my math is right) that is only 700 ohms of reactance, not enough to reduce the input impedance by much.

That means that the balanced input DOES use a center tap to ground and these UG-529/U (53 to 95 ohm) adapters I have, were probably meant to be used with something else.

Roger: I don't think Tony is "chasing his tail" A lot of spectacular claims are made for one receiver or another and might be because of "measurement error".

Tony: To beat a dead horse some more (trying for more accuracy). to minimize 50 ohm cable errors from the generator, to an unknown load, if you have a 20db or even a 10db inline attenuator, put that at the END of the line as close to the receiver as possible and the cable will "see" a better 50 ohm load.

Charlie

edited to add "ohm" after "50"

AARRGGHH! You're correct Tony the transformers are NOT center tapped. That
will teach me to keep my mouth shut until I check the schematic!

The only thing I can suggest at this point is to measure the input Z at the
measurement frequency and cobble up the appropriate matching pad.
Compute the pad loss and use that to figure the actual input voltage.

73, Roger

Thats exactly how I use the "swamping" method that I described up near the beginning of this thread. I really like that method, and not just because it gives nice results for my radio . The 40db attenuator goes on the end of the cable and I have a little 1 inch male to male BNC that connects to the radio. I like it because it pretty much guarantees a known voltage at the antenna terminals. The thing I DON'T like about it is that it hammers the Q of the antenna tuned circuit. It's not as bad as it sounds in the 390a because they redesigned the transformers for more "step up" to try and compensate for the loss of an RF stage. This increases the tolerance of low impedance sources. But .5 ohms is a little much. I still am able to peak it with the antenna trimmer on most bands but I would prefer a better match. That's what led me to the DA121/U which I also use at the end of the cable.

I dug out my selective level meter today. With that in "bridging" mode (HI Z input) I can read the RF voltage at the antenna terminals directly down to about -130db relative to .775 volts. Not exactly a universal method but it should be more accurate than what I have been doing. That gives me a better guess on the input Z without overdriving the receiver.

edit; Your comment about using a separate receiver made me think of the SLM.