Of course you have to wait until the DDS clock oscillator stabilizes, perhaps 10-15 minutes, before checking for any IF strip drift.The Elecraft NE602 product detector and simple AGC did not work for me. All the static crashes kept the gain severely reduced, and it took 5 seconds or more to very slowly recover each time. Seemed like a good circuit to try, though. 

But the main IF strip problem is the grainy, staticky sound the IF strip makes--all 3 of mine have the same sound that sounds like low level static picked up by the antenna, but with nothing connected, it's still there. That's the main problem I see with this ML IF strip idea. Using my reference setup, the R75 with 10Hz 599zx DSP, which is very quiet and has no low level of anything heard, the ML IF strip is much noisier and quite annoying to listen to for any length of time.

After awhile you imagine you're hearing idents in the static, just like when listening to real antenna-received static. I've tried 2 different circuit variations now, from the original one. All have the same problem of the grainy, low level static noise constantly heard. 

Unless that and the drifty passband gets fixed, this ML idea probably is not very reproducible for mass-construction. I've built up 3 different IF strips now, with no parts reused in any of them.

Originally I thought maybe I had some bad parts, but all 3 sound the same.  One can certainly hear beacons with the narrow passband, but it would be much nicer to have a quiet receiver, limited by the external signals.

Steve

*********************

MY REPLY:

Good evening Steve,
I was thinking about writing you and asking how your receiver was coming along.  After extensive use, here are my observations on the set.  I hope this answers your questions.

1.  In "real" use I haven't noticed any problem of changes with temperature, though a response check like the one I did in preliminary testing may be worthwhile when I have time.  My radio room has it's own window air conditioner now, and while "slight" changes in temperature as it cycles drastically affects the Hallicrafters SX-71 when listening to SSB or CW, it doesn't appear to affect the ML set in any way.  Of course the ML set with all it's mass of copper shielding, lack of ventilation etc isn't likely to change temperature much, unless the change is large and sustained.  (It's actually rather heavy for it's size).

2.  The cabinets: They come from Radio Shack and are/were their "deluxe steel cabinet", same as the one your preamp you got from me has.  They were a little more expensive than the plastic stuff, but I got these for 9.95 each.  One of them had a slight amount of damage in one corner I was able to straighten out ok.  They must have had poor quality control though.  Of the three cabinets, all three are slightly different shades of paint, either the ivory or the black.

3.  The Static or grainy sound that you hear "fake CW" in.  I call that "CW Ghosties" as it sounds like Ghost CW.  It is due to random noise causing the extremely narrow analog filters to "Ring" slightly, and I don't think there is ANY way you can get around it, not with analog anyway.  I have heard EXACTLY the same thing when I've used VERY narrow analog filters like the Autek QF-1A cranked way down on conventional receivers. 

Bill Hohenstein once explained it to me that it's an indication you're working very near the theoretical limits of sensitivity/selectivity.  Apparently the "ring time" is affected somewhat by the actual narrowness of the analog filter.  Some notes:

     A.  IF I turn off the RF pre-amp, switch the antenna off etc so that I'm listening to ONLY the IF strip noise, I have to turn the Audio Gain up VERY high and then I just barely hear "Ghosties" down in the AF hiss. The RF gain setting (input potentiometer between mixer and input) naturally has NO effect on this.  To me this indicates the IF strip is actually pretty quiet!

     B.  Turning the RF preamp on, the Ghosties rise some, indicating that the RF pre-amp raises the noise floor of the receiver.  The RF gain pot DOES have an effect on this. This I would expect.

     C.  Connecting the Antenna with no signal again raises the level of the Ghosties, and again I expect this, all indicating that the INCOMING noise is the limiting factor, not IF strip noise!

     D.  If you have the antenna connected and the RF gain set TOO HIGH the "Ghosties" can get pretty bad!  There is probably more front end gain in my receiver than is really needed.  I find that OPTIMUM setting for the RF gain varies across the band and with conditions, but typically doesn't run more than about 50% up.  That's with a 5K pot at the output of the RF amplifier and an "assumed" 50 Ohm load in the mixer. 

I find that there is better sensitivity when the LPF at the input to the mixer (RF amplifier has one by default) is switched OUT, but on very weak signals the S/N seems a little better with it switched IN.  I have to presume that's because any 2 MHz noise generated in the wideband RF amplifier is filtered out before hitting the mixer.

4.  The NE602 product detector/AGC: I've not had any problem with the AGC "blocking" the set.  In fact, I'm not at all sure the AGC is really doing anything in my set.  I say this because using a Voltmeter I haven't been able to see the transistor "turn on".  Possibly the diode I used for AF rectification wasn't suitable.  Next time I get in the set I'm going to examine that area closer.  Perhaps a germanium diode is called for there.

5.  AF filtering: I have a brute force RC filter between the Product detector's AF chip and the following output stage (volume control between).  This to eliminate any AF "hiss".   This was chosen empirically and was about .1uF.  It is connected where the 1 Hz Notch filter will be tied in.  In the Notch filter I've added an active 425 Hz LPF BEFORE the Notch filter to remove all above 425 Hz (I'm tuned for 400 Hz beatnote).  Following the Notch filter will be another buffer with a 425 Hz active LPF to take out any AF Hiss generated in the filter or earlier stages.  I find any AF hiss is FAR more distracting than the "Ghosties" ever are.

6.  THE TEST FOR GHOSTIES: You probably know all this, but I'll reprise it for you.  Here is what I am constantly doing ANYTIME I'm using very narrow bandwidths as "Ghosties" can emulate nearly any callsign in your mind.

     A.  FIRST, TUNE up and down a little.  If it's REAL it will change pitch.  A "Ghostie" NEVER changes pitch.  Note, I said "Pitch", not volume.  Ghosties can occasionally be aggravated by nearby strong signals, especially DGPS, but will never have their pitch changed.  The pitch of the Ghostie is determined ONLY by the center frequency of the filter in relation to the BFO and Ghosties get louder in relation to band noise.

     B.  Change Bandwidth: Ghosties are afraid of wide bandwidths and will usually disappear into the noise when things are "opened up".  For example, in my ML receiver, the gains are about equal, but switching in the WIDER bandwidth make the Ghosties nearly disappear, though if you listen closely there is a hint that they're there, much weaker and slightly lower in "pitch" (the wide Bandwidth is offset slightly from 2 MC).  A REAL signal won't change pitch with bandwidth (assuming you have more than one to select).

    C.  You probably won’t hear Ghosties on a receiver like the R-75 with outboard DSP simply because the IF bandwidth is too wide and the DSP even when cranked down narrow generally doesn't "ring".  Now, if you had a "cranked down" Autek QF-1A on the R-75 they’d probably be really bad.

PROPER USE OF THE SET: I've found that this is one set that requires a LOT more finesse to use than the average receiver on which you just crank the tuning until you hear what you're looking for!   Turn the RF gain up too high and the Ghosties and noise will drown out any weak signals.  In fact, turn the RF gain up too high and the set produces some "VERY RUDE" noises in the headset on the least little noise burst.  I believe this is due to the NE602 being overdriven, though I haven't dug into it much yet.   Turn the RF gain too low and you're not going to hear that weak one.  I leave the AF gain set about "midway" and "ride" the RF gain. 

HOWEVER, properly used I believe that it is very difficult to find ANY receiver that will outperform it, especially considering the cost.   Not counting the DDS (which can be used elsewhere too since it's in a separate cabinet) the set cost me less than a GOOD quality CW filter for any name brand receiver.

IT PROBABLY TOOK ME AT LEAST A WEEK OF STEADY USE TO GET THE IDEA, LESS RF GAIN IS "BETTER" in the use of this receiver.  You mentioned I'd probably not need the RF amplifier.  Not true. Without it I only hear the stronger beacons, most likely because my antenna isn't a good match for 50 Ohms.  This has proven to be true in every receiver I've had here.

This set is probably better suited for taking time to ferret out the tougher beacons than it is "cruising".  Punching in for different tuning steps and the narrow bandwidth slows "Cruising" down greatly.  But I've NEVER found a receiver that was able to pull ADK out of the BCB junk on 530 KC.  A DSP trying to process the audio after it's detected has never been able to do it, at least not here where there are 3 Highway Info stations on 530 all the time.  This one did it!

I'm also considering tapping the 2 MC IF prior to the high selectivity stages so that the Hallicrafters SX-71 can be connected for when I want a much wider bandwidth, such as for things like DGPS, high T-storm noise conditions etc.  That way I'd still have my digital accuracy and front end.  Probably would tap it after the first Xtal stage as that way it's "buffered" from the mixer. That first stage is common to both bandwidths and is centered on 2MC.

I guess you would have to say that my receiver is only "partially" a Mitch Lee receiver, the IF section (which of course is it's "core").  The rest is my own creation or robbed from other circuits.

Well, guess it's time to go look for beacons.  Have fun.

73 de Phil,  KO6BB

*********************

STEVE’S REPLY BACK:

Hi Phil,
Thanks for the detailed reply. I've read it a couple times now. Very good information from your use of the ML radio. I now think I'm hearing what's normal for this type of circuit in my own IF strip. I may still have a drift problem. I'm using plastic trim caps; I notice from the pictures you have ceramic ones; Mitch says he has ceramic too. My Mouser order said ceramic on the ad, but when they came they were plastic. I'll have to get some ceramics and try them.

It sounds like the gain distribution is a very carefully adjusted parameter in this circuit, and I don't seem to have quite gotten it adjusted quite right yet. Mitch says the Mouser ECS crystals are simple room temp microprocessor crystals, not spec'd for operation at elevated temps (like in an oven), thus should be fairly stable at room temp.

Now I think I'm trying to make too much out of this circuit. It does amazingly well, for what it is, as you've found out with your ADK reception out of the TIS/HAR jumble. I'm used to hearing a quieter receiving situation, dominated by antenna noise. I need to properly adjust the gain after the IF strip plus the gain ahead of it, now.

That original DDS article you sent me on the JPEG format, I stumbled across FAR circuits, the guy who does all the PCBs for the ham magazine articles--he has done a full PCB for that article/DDS. It's also been upgraded to the v.3a PIC software that you're using, plus he offers programmed PICs with the PCB if you want. The AD9850 plus the PIC are all on one small PCB--complete DDS and Controller--no need to wait and wait and wait for amqrp.org to come out with a new batch of DDS Daughtercards and hope you're among the lucky to get in on the batch. So I've ordered the PCB and the PIC from FAR. (The PIC should be the same as the two you programmed for me, but I bought his just to make sure, too.) It just arrived--it's a beautiful board, about 3 x 3 inches.  Course the AD9850 still has to be soldered on. I don't have an iron with a tip that small, or any small solder either.

I guess I'll have to invest in a new iron. I haven't bought one in 30 years, and try to solder the DDS chip on myself. So, assuming I can solder the chip on when my new iron arrives, I ought to be able to make a functional DDS/controller board finally, and not have to use my signal generators on this ML project anymore. All the external parts still are needed-rotary encoder, switches, LCD, etc.

I think a complete DDS section can be built for just under $100. That seems to be the most expensive part of the ML setup.  I told Mitch you didn't have any spur problems; he could hardly believe it.  His DDS uses a 50Mc clock, which I suspect is the main problem for all the spurs he reports finding. Ours uses the 100Mc clock--I think that makes a major difference for spurs at the low freq the ML radio works at for the synthesizer. Incidentally I could only find one source in my usual parts catalogs for a 100Mc clock oscillator, from Jameco. Mouser and DigiKey don't list any that high. I ordered that from Jameco, too. I have the LCD and rotary encoder already on hand, and of course the AD9850 from the one you sent. And I figured out how to register with Analog Devices, and ordered the sample AD9850 from them, too. Hopefully they'll send me two samples.