EMRFD Message Archive 6563

Message Date From Subject
6563 2011-08-22 23:35:16 Kerry Measurement Receiver Sensitivity
I have been building a Measurement Receiver similar in principle to the EMRFD one but a little more elaborate. Here is the block diagram;


I have built & tested the modules shown and have now been testing the assembly. I am using my stand-alone AD8307 power meter for testing but will build a module to include in the receiver; it will have "offsets" so that it will display the input power.

Both filters were used in the tests and the results described below were the same irrespective of which filter was "in".

Typical signal levels are shown on the diagram; it looks good with 40 dB of overall gain (the "round" figure was obtained by adjusting the gain of the AD603 amplifier).

So, say, -50 dBm in produces -10 dBm out and so on.

The problem I have is that it will not respond to input signals below about -85 dBm; reducing the input level does not reduce the output level which "stalls" at around -50 dBm. A constant output reading of about -50 dBm is obtained with no RF input (but with the LO running).

I was hoping to be able to measure input levels of -100 dBm or less.

Is my problem caused by noise? Will pre-mixer selectivity be required to achieve my goal?

Or am I missing something else?

6566 2011-08-23 10:18:52 Tim Re: Measurement Receiver Sensitivity
6567 2011-08-23 10:21:33 Wes Re: Measurement Receiver Sensitivity
Hi Kerry, and gang,

I suspect that the problem is noise, but preselection ahead of the mixer won't make any difference. The problem is the noise from the stages after the filters.

The AD8307 has a bandwidth of about 500 MHz. The AD603s are 100 MHz anyway and when they do drop off, it is slow. So if you want to calculate the noise power you have to do it over this wide bandwidth. The answer is to put another filter in the system that is between the AD603 and the AD8307.

I have a partially finished measurement receiver sitting on the shelf that uses a 6 MHz IF. (WA7MLH contributed some nice, high Qu crystals.) The ring mixer drives a bipolar post amp just like your system. The crystal filter is N=5 with a BW of 300 Hz or so. Then I have an IF amplifier with a gain of about 40 or 50 dB. This will then drive a N=2 crystal filter, also with a 300 Hz BW. I had to do some extra tuning to get the N=2 filter to line up with the main filter. This will then be applied to the AD8307. This much of the system is working, but no measurements have been done. Now to build some swept oscillators to do some experiments with the system, and to characterize the overall receiver.

The intended applicati
6569 2011-08-23 14:13:46 Kerry Re: Measurement Receiver Sensitivity
Thanks Tim & Wes.

Tim makes a good point about WWV; that had never occurred to me.

I used 5 MHz because they were the best crystals I could find; I designed for 12 MHz initially but couldn't get decent crystals from the hundred or more that I bought. We don't have the luxury of cheap hamfest bags of older & better crystals that I hear about over your way.

WWV is not overpoweringly-strong here in VK; WWVH may be stronger but I don't think that's the problem. All modules are sealed in PCB boxes except the AD603 which is still open; I will put a cover on it and check though.

I suspected noise as I implied in my post; I remembered seeing Wes' calculations showing why an AD8307 meter always has a small "standing" reading.

The 300 kHz filter is included because I want to use this receiver to measure noise figure; I built an L/C Gaussian-to-6 unit that has a very good shape. I will try a filter before the AD8307 as Wes suggested.

It may be possible to find a compromise bandwidth for this filter; if not I will have to build another switched pair of filters.

It's a good illustration of how sensitivity and bandwidth are related.

Thanks for the feedback fellas.

Regards, Kerry.
6570 2011-08-23 16:06:26 kb1gmx Re: Measurement Receiver Sensitivity
6571 2011-08-23 16:59:26 Kerry Re: Measurement Receiver Sensitivity
Good points Allison.

I have had little luck with crystals from various suppliers; the best I have, a batch of 5 MHz ones, have Q of about 150 000 (and that's only about 10% of them).

Other batches show Q of less than 100 000 (often much less).

Wes, for instance, mentions crystal Q of 240 000 in EMRFD and I think elsewhere; I can only envy that!

It can get expensive & frustrating buying crystals in batches of 50 or 100 only to find that they're "duds".

I like your comment that " ... 300khz is wide but not that wide"; that's why I chose it for noise measurement.

In fact I erred in my description; 300 kHz was an early design but I forgot that I re-designed it for 500 kHz (it measures at 483 kHz).

Just as an " in my head" calculation, a 1 MHz (noise) bandwidth should be capable of getting down to 174 - 60 = 114 dBm so 500 kHz should do 3 dB better.

I will look at the items you suggest; in particular I have realised that there are a couple of AM broadcast transmitters not too far away; I've only been at this address for about a year and that fact hadn't sunk-in until you mentioned it.

Unfortunately my faithful 141T spectrum analyser has died; I've been saving my pennies for a more modern one (an 8594E I hope) but I haven't found a suitable one yet.

6572 2011-08-23 18:15:24 Russell Shaw Re: Measurement Receiver Sensitivity
6573 2011-08-23 18:28:15 Kerry Re: Measurement Receiver Sensitivity
> I've got a 141T. I've fixed them a few times.

I have two mainframes and three RF units; 8553, 8554 & 8555.

I have repaired the power supply (a known problem) in each 141T.

I have two 8552 IF units; one failed a couple of years ago and the other one failed recently. I don't have the extender cables and I don't really have the skill to repair them.

8594Es have come-down in price in recent times and, provided I can find a good one, I'd like to get one of those; it will also occupy rather less space than the 141.
6574 2011-08-23 18:42:14 Russell Shaw Re: Measurement Receiver Sensitivity
6576 2011-08-23 22:06:48 Kerry Re: Measurement Receiver Sensitivity
So as not to get off-topic I have PM'd Russell about 141s etc.

I'm beginning to think that Allison is on the right track.

I increased the AD603 gain to 50 dB (the maximum) and I can now get to -97 dBm before the RF input ceases to "track" the power meter reading linearly.

But I have realised that, whatever the IF gain, I am "bottoming" on the -50 dBm reading shown on the power meter with no RF or LO input to the mixer.

As I said, the modules are well-shielded except for the AD603 amplifier which is still open over its ground plane.

I think that boxing it up might be the next step.

6577 2011-08-23 22:41:00 victor Re: Measurement Receiver Sensitivity
6578 2011-08-23 23:22:20 Kerry Re: Measurement Receiver Sensitivity
> If we follow your calculation of thermal noise power at 500kHz banwidth which comes out as -117dBm, than add the noise generated by the 2N3866 which has a noise figure around 5dB (-117+5=-112dBm), now convert it to the input (attenuation from circuit input to the amplifier input is around 7dB) then we get an equivalent circuit input noise of -105dBm. If the rest of the circuit adds some more noise (and maybe the amplifier has a worse noise figure) then it gets closer to your result. May be you can improve by some more dB's by using the 2 pole filter at the AD8307 circuit as Wes suggested.
> Victor - 4Z4ME

That was my general thinking Victor; I expected to be able to measure -100 dBm or less.

But I think there are two things going-on at once here.

Increasing the IF gain by 10 dB allowed me to measure inputs 10 dB lower, ie -97 dB input plus 50 dB system gain gives -47 dBm meter reading.

The readings "track" as the input changes; I can change the input by, say, 10 dB steps and the meter reading changes by 10 dB steps down to the point where the input is -97 dBm.

The reason is that there is a meter reading of about -50 dBm even when the RF input to the mixer is off or terminated with a 50-ohm termination; that implies that -50 dBm at some frequency within the wide passband of the AD8307 is reaching the meter.

As Allison suggested this power may originate from some kind of pick-up; the 50-dB-gain AD 603 amplifier is unshielded so it is the first suspect.

It might also be oscillation; that's why I lamented my lack of a working spectrum analyser (although I've remembered that I have an ancient HP180/8558 SA that was OK before it went to the store-room some years ago :) ).

Once I have this nailed I will try some filtering as you suggest; I think that that will extract a few more dB from the system.

Thanks to you and the others for input; probably like many/most home-based experimenters I have no-one with whom I can work-through ideas on a face-to-face basis as they arise so discussions like this are of tremendous value to me.

Perhaps we should set-up an EMRFD retirement home so we could all be neighbours! :)

6579 2011-08-24 00:02:04 victor Re: Measurement Receiver Sensitivity
You could put the circuit and batteries in a metal cookies box (after eating the cookies...), connecting 50 Ohm resistor as input, bypassing the meter wires and having only the meter outside. If your problem is signal pickup, this test will show it.
Victor - 4Z4ME

6580 2011-08-24 09:06:38 Tim Re: Measurement Receiver Sensitivity
I'm not sure what your local RF environment is but if I put a couple of inches of wire on the end of a scope probe, I see about a volt at 630kHz from the radio station a couple blocks from my house, and in the 90-100MHz ballpark at the hundred mV level I see FM stations that are 3 miles away.

It's the broadband stages that get hit the most by this sort of stuff (which I think describes your AD603). Obviously any wideband detector or silicon junction that can act as a detector, pick up the AM signal. Narrowband circuits for HF seem quite unaffected, and I don't have to shield them at all.

The local 630kHz stati
6581 2011-08-24 10:57:19 Roelof Bakker Re: Measurement Receiver Sensitivity
Hello Kerry,

I have a Wandel & Goltermann selective level meter, which can measure down
to -120 dBm.
The frequency range is 200 Hz to 1620 kHz.

It is build in a modular way and all modules are mounted inside heavily
screened boxes.
So there is certainly much to be said for thorough screening measures.

Roelof Bakker, pa0rdt
6582 2011-08-24 15:30:18 Kerry Re: Measurement Receiver Sensitivity
I have no doubt that the problem is pickup from nearby AM transmitters.

I haven't been at this address for very long and the presence of the b/c stations didn't really register until now; my previous home was in a country town with no such stations.

I have found that clipping a length of wire to the power meter input gives a reading of -35 dBm!

So shielding is the answer.

Just for interest, this is the test setup;


The signal generator that provides the RF is out of the photo. The power meter and the LO are at the rear; the two PCB-material boxes on the right contain the mixer, amplifier & filters whilst the "breadboard" construction of the AD603 IF amplifier is obvious.

I will enclose it and see what happens; once this part of the problem is solved I will try some filtering.

It's been an interesting and educational experience!

Thanks to all who gave valuable advice.