EMRFD Message Archive 14419

Message Date From Subject
14419 2017-10-29 16:42:20 rcbuckiii Superhet design questions
I have the following configuration for the front end of a 40 meter receiver:

40 meter BPF -> ADE-1 mixer -> W7ZOI/WA7MLH Hycas amp -> product detector -> audio stage

The ADE-1 mixer has a 10 dB amp on board. It is the same amp as used in the W7ZOI/K5IRK progressive receiver.

For testing, I am driving the ADE-1 mixer RF port with a AFG3021B signal generator set to 7.120 MHz. The generator is set for 10 mV RMS (28.3 mVpp). The output from the generator shows -40.45 dBm on my DSA-815. After taking this measurement, I connect the generator to the RF input on the ADE-1 board.

I am driving the ADE-1 mixer LO port with a Si5351a followed by a low pass filter and feedback amplifier. The output for the amp is +8.5 dBm at 16.120 MHz. The output from the mixer at 9.0 MHz is -33.5 dBm with the generator and LO connected.

When I attach the Hycas amp to the mixer output, the Hycas output is -8.3 dBm at 9 MHz. I have the AGC turned off on the Hycas. The Hycas is only showing 25.3 dB of gain with the AGC off. The QST article says the gain is between 55 and 60 dB. The article also says the output should be between –35 and –40 dBm. However, I am driving it with a much stronger signal than would be present from an antenna. I should mention that the Hycas is my own board design using all SMT components. I have one of the KA7EXM boards that needs a few more components installed. I could add the components and bring it up if anyone thinks my Hycas has a problem.

If I connect a +10 dB feedback amplifier in place of the Hycas, I see the mixer 9 MHz signal go from -33.5 dBm to -23.1 dBm which tends to indicate the BPF/mixer combination is working correctly.

The product detector I am using is the one that VE7CA uses in his HBR2000 receiver. It uses a diode mixer followed by a preamp with gain of about 4. I use the output of the preamp to drive a TDA2003 amplifier.

When all of this circuitry is powered up, and an antenna connected to the mixer RF port, I hear what I would describe as typical 40 meter background noise, but no signals. I only have a 35 foot piece of wire strung indoors but it should pick up something. Do I need more gain between the Hycas and the product detector?

One more question - what is the minimum level signal that a diode RF probe would be able to detect? My guess is that it is in the 0 to 3 dBm range.

Sorry for such a long post, but I hope to get this receiver working soon with everyone's help.


14420 2017-10-29 17:53:51 Ashhar Farhan Re: Superhet design questions
The line up doesnt seem like a superhet design at all. Is there a crystal filter in there?
- f

14421 2017-10-29 18:08:55 rcbuckiii Re: Superhet design questions

I haven't added the crystal filters yet. I have a CW, SSB, and AM filter designed for 9 MHz. I was going to put them in after I heard some signals out of the speaker. In fact, with no selectivity I would think I should hear lots of signals at the same time. I was going to put the crystal filters between the mixer and the Hycas.
14422 2017-10-29 18:42:33 Markus Hansen Re: Superhet design questions
Hi Ray. I have built 3 Hycas IF amps, one on the PC that was offered by Wes’s son Roger and two just ugly construction. All three amps had measured gain in the >55dB range. I would suspect your amp is compromised some how. Good luck tracing the problem. Persistence will provide the answer and eventually success m. 

73. Markus, VE7CA. 
Sent from my iPhone

14423 2017-10-29 19:24:22 rcbuckiii Re: Superhet design questions
Markus, thanks for the reply. I liked the design of you product detector board since it included provision for AM reception. I think it is a very good design. I believe the low gain from the Hycas was because I was driving it so hard. I had a 60 dB pad inline with the generator yesterday and I believe I was getting around 50 dB gain with that setup. I will test again tomorrow and confirm.

Ashhar, I just installed the AM crystal filter and adjusted the product detector LO to the crystal filter offset frequency. I thought I would start with the widest one first. I still don't hear any signals. I do hear miscellaneous hetrodynes and squeals as I tune around the band. Those could be coming from the PIC that controls the Si5351 or from digital equipment in the shack.

Tomorrow I am going to run a 60 foot #22 gauge wire under the roof overhang outside of the house. That will be a better antenna than the one inside. HOA prevents me from putting up a real antenna. The other thing I will do is finish the KA7EXM board and use it in place of mine. I don't think that is the issue but I have to keep trying anything that could possibly be a problem.

I also should mention that I do not hear any noise out of the speaker until I connect the antenna to the BPF.

14424 2017-10-30 05:16:52 Chris Howard w0ep Re: Superhet design questions
Use your sig-gen instead of random signals, will remove one level of complexity.

14425 2017-10-30 08:22:52 kb1gmx Re: Superhet design questions

Overload/over drive is likely the issue.  For example a 0 dbm signal cannot make it though
a Hycas (as published) as its impossible for the devices to increase a 1 milliwatt signal by 
nearly 60 + DB ( 60dbm is 1000W!) you can but likely will not get a very large signal!  What 
happens in that case is they max out and you see a smaller signal and likely lower 
represented gain.  In short you hit the compression point of the amplifiers which is 
also where distortion sets in.

For you analysis and testing you need weaker signals.  FIrst the ADE-1 is a level 7 mixer
and any output greater than -13dbm is going to be distorted. Note I use DBM to keep
measurements and signal values in the 1milliwatt 50 ohm system.

So while testing as you add gain its wise to lower the input signal by that amount, 
attenuators are your friend.

Hycas is a good IF systems with significant gain, maybe more than needed but its flexible 
and doe not need to be run wide open.

Hope that's a helpful starting point.

14426 2017-10-30 09:34:16 rcbuckiii Re: Superhet design questions

Thanks for your explanation. I didn't look at the output of the Hycas on a scope but assumed the low output was due to distortion and overload.

I am driving the mixer with a -27 dB signal from the generator so I am OK there. But the -8.3 dB signal into the product detector does exceed the 50 mW rating of the ADE-1.

In addition to finishing Roger's Hycas board and putting an outside antenna up, I am going to assemble a couple of 30 dB attenuators today. That way I can more realistically simulate antenna level signals.

14427 2017-10-31 09:09:07 kb1gmx Re: Superhet design questions
Yes very likely even the ADE-1 is in overload.  The rule of thumb for DBMs is keep the inputs
at least 10db under the LO.

The Hycas at full gain is nearly 80db so even a -80DBM signal can run it to near overload.
With AGC applied the gain can and does drop  (up to -100db of agc range if memory serves)

With superhet gain after the mixers largely does not effect the sensitivity as that's a factor of the 
input filters,  RF amps, and mixer noise as well as detection bandwidth.  The goal is to get the 
signal up to the level needed to detect(or otherwise process).  An example of that would be 
a simple diode detector where -40 dbm is very weak but detectable and more helps till you 
hit the overload region.  SSB detectors as used in DC receivers can easily hear -110DBM 
but overload is near or below -20dbm, the same is true using simple DBMs as the mixer.

The reminder is that once you can hear man made and galactic nose your at the limit.
So from the system perspective enough gain to inside the signal is above the mixer 
noise and input filter noise is about all you need.  After the mixer amp the filter is you next
block, any gain after that does not improve matters save for giving AGC capability 
(gain reduction) and insuring enough gain for detection process.

The above show why the BITX40 and many others work well (hears weak signals) but 
also suffers overload on strong signals.  The total RF gain of BITX radio is only 50db 
before losses (about 18-20db) so adding gain in superhet systems has its limits 
without resorting to AGC or later stages being able to generate truly huge outputs 
without overload.                                                 

14428 2017-11-01 09:33:46 rcbuckiii Re: Superhet design questions
Allsion, your comments are very informative.

I put up the outside antenna, built the two 30 dB attenuators, and finished Roger's Hycas board.
With the 30 dB attenuators in series and fed by the generator, Roger's Hycas and my board both perform roughly the same. Feeding a -66.3 dB signal into the BPF which connects to the 1st mixer, I get -57.3 dBm out. That is 9 dB gain which is close enough to the 10 dB design figure. The mixer/amp drives the Hycas and I get -7.8 dBm out of it with the AGC turned off. That translates to a gain of 49.5 dB so the Hycas may still be going into slight overload with the AGC off. I also see the 7.120 MHz RF signal at -45.6 dBm out of the Hycas.

Looking at the output of the Hycas with the SA set to span 7-30 MHz, I see these signals:
16.08 MHz at -16.08 dB
18.00 MHz at -31.3 dB
19.76 MHz at -64.7 dB
25.12 MHz at -43.2 dB
27.00 MHz at -46.6 dB (27 MHz is the oscillator frequency for the Si5351 module)

When I disconnect the signal generator from the BPF and connect the antenna I still do not hear any 40  meter signals. The Hycas should not be overloaded with signals from the antenna. At this point, I believe the signals are being masked by noise pickup and interaction between all the different boards. Even though they are all connected together with RG-174, there still must be interaction.

I am going to build PCB enclosures for the BPF, 1st mixer, crystal filter, Hycas, product detector, and PIC controller board. Then maybe I will be able to hear 40 meter signals. I will give an update as to my progress in the next day or so.

Thanks for your help. I'm sure I will get this going soon.

14429 2017-11-01 10:11:11 Markus Hansen Re: Superhet design questions
Ray, if you have a signal generator that has calibrated output or an crystal oscillator that you know what the output power level is, connect it in series with your attenuators then to your receiver.  Without a RF pre-amp you should be able to hear at the speaker output,  an input signal at -125 dBm or so without difficultly.  

Good luck,

Markus, VE7CA
North Vancouver, BC  CANADA
WEB: ve7ca.net

14430 2017-11-01 16:34:48 rcbuckiii Re: Superhet design questions

Once I get everything in shielded enclosures, I will run that test. My AFG3021B will only go down to -36 dBM. With the two 30 dB pads I built, I can get down to 96 dBm. But, if I can hear that signal, I will build another 30 dB pad which will take me down to -126 dBm.

The generator is capable of uHz resolution and was recently calibrated.

14431 2017-11-02 06:54:06 John Lawson Re: Superhet design questions
Ray, since you are doing measurements, if you haven't already done so you might enjoy reading Wes Hayward's July 1975 QST Article entitled Measuring and Defining Dynamic Range. Within that article in Figure 8 there is an oscillator at 14 MHz that gives off a signal of around -114 dBm. That oscillator in conjunction with a step attenuator will give you a sensitivity (MDS) reading for your receiver. I've used this oscillator over the years to measure numerous  receivers built here at my QTH.....Just another way to accomplish your measurements......Hope this is helpful. John Lawson, K5IRK

Sent from my iPad

14432 2017-11-02 16:11:45 rcbuckiii Re: Superhet design questions
Thanks John. I read the article and it has good information in it. The signal source would be a handy piece of test equipment.

14433 2017-11-02 17:35:50 Markus Hansen Re: Superhet design questions
I likewise built the same generator when I first got starting building my own receiver.  I still have and use it often even though I have to HP8640’s.  

Markus Hansen
North Vancouver, BC  CANADA
WEB: ve7ca.net

14434 2017-11-03 14:58:51 kb1gmx Re: Superhet design questions
What selectivity is after the mixer?

So far I know of none not eve a IF band pass filter using Cs and Ls.
If that is the case I'd expect everything under the sun to be present 
and overloaded as well.

Also the output of the DBM will contain at a minimum the LO (about 40db down),
All the possible input frequencies, and harmonics of both and their sums and differences..,
I usually use a diplexer at the mixer output for two purposes.  One to rout all 
non IF frequencies to a 50 ohm load and only pass the IF to the post IF amp.

If you fail to do that all the possible undesired signal will likely overload the IF amp!

Oh and the IF crystal filter is pretty much required with that much gain.

At RF leads and grounds are important.  Shielding and coax may help.


14435 2017-11-03 15:06:45 kb1gmx Re: Superhet design questions
A lot if not all of that is restated in EMRFD.

Another easy trick is get or home build the Elecraft XG2 for both calibrated signal level and 
a known weak signal source.  I built one for myself ugly style in a die cast box, its handy.

The biggest issue is many sources leak more RF than you can attenuate at the jack.
To get down to sub -100DB you really have to have a box in a box in a box with everything 
bypassed well.  An example of that is the Measurements 803, HP60X series, 8640, and others.

A good example of that is the Hayward 14mhz source. 

14436 2017-11-03 16:29:58 rcbuckiii Re: Superhet design questions

All the signal levels I show are without the crystal filter in place. I took the measurements just to see what the levels were at each stage.

Now that I am putting all stages in their own copper PCB material enclosures, I will put the AM crystal filter after the 1st mixer and before the Hycas. I think by doing all this, I will then be able to receive 40M signals. If that works, then I will add the SSB and CW filters (and associated relays) to the crystal filter enclosure.

I should mention that I have a 10 dB gain grounded gate amplifier that follows the bandpass filter. The bandpass filter has about 2 dB loss in it. The amp is probably not needed for 40 meters but I don't think it will contribute much to the noise figure.

All enclosures will be connected together with SMA connectors and RG316 cable.

14437 2017-11-03 19:27:36 Markus Hansen Re: Superhet design questions
I likewise built the same generator when I first got starting building my own receiver.  I still 
Markus Hansen
North Vancouver, BC  CANADA
WEB: ve7ca.net

14438 2017-11-05 09:15:42 jim_amos_n8cah Re: Superhet design questions
Hi Ray,

A lot of good information on here.

I have a very similar receiver on my workbench right now.  It has a 3 pole 40M BPF, SBL-1+W7ZOI AMP, filter, HyCAS, etc.   when I first assembled the receiver I had similar problems to what your are reporting that were due to the BFO getting into the receiver front end.   My initial BFO was crystal controlled, with dead bug construction.    This would get better or worse depending on the filter and the BFO offset into the filter.

I was experimenting with various homebrew crystal filter configurations, so the filter passbands were not generally exactly centered on 9 MHz.  But, the BFO had the standard 9 MHz +/- 1500 KHz offsets.   When I changed to a  Showa filter centered at 9 MHz and things got much better.    I have since started using the 3rd channel on the Si5351 for BFO.   This let's me more easily tune the BFO specifically for the filter center freq and BW.  So, the SSB Show get's one set of BFO offsets, and the homebrew CW filters get their own and they offsets are calculated for the specific filter.   I also think that there is much lower leakage / radiation to get into the receiver front end from the Si5351 board. 

So, your idea of boxing everything will help a lot, but you might also try a narrower filter before or while you reconstruct all of your circuits.  Since you're using an AM filter, it's bound to have the BFO signal in it's passband (you didn't mention if you were specifically trying to receive AM, or SSB).   This should not be the case for SSB and CW filters and demod.

I'm also using SMA connectors and interconnecting cables from China off of EBAY.   The connector quality seems so so when compared to those from US Suppliers that I've used for work.  I think there is about a 10:1 price ratio, but they do the trick for HF.   This also lets me connect and remove modules as I develope different boards.  None of them or boxed up yet, but things do seem to be working pretty well.   This will help as you try to box things up.

Good luck and let us know how it goes!

73's Jim N8CAH
14439 2017-11-05 12:23:09 jgaffke Re: Superhet design questions
> Since you're using an AM filter, it's bound to have the BFO signal in it's passband 

Put the BFO a bit beyond the skirt, either above or below the AM filter passband.
If the skirts of this AM filter are as sharp as that of the SSB filter,
would give same performance as the SSB filter except that some higher audio frequencies
would come through.

Jerry, KE7ER

14440 2017-11-06 13:02:39 rcbuckiii Re: Superhet design questions

Thanks for the information. I will keep all of that in mind. I know the center frequency of my filters as I have swept them with the SA. The AM filter is 4340 KHz low. I put a rotary encoder on the PIC so I can tune both the LO signal for the 1st mixer and the LO for the product detector. After everything is resolved, there won't be an encoder for the product detector.

Unfortunately, I am sort of at a stand still on the project. I thought I had enough PCB material on hand to enclose everything but I don't. I have ordered more and should have it by the end of the week. Then testing will continue.


I won't be using the product detector for AM demodulation. I was just using the AM filter with a wider bandwidth for testing purposes. I was hoping that with a wider passband I might hear some signals. Your idea would work fine if the product detector was being used. I believe that is how the commercial guys do it.

14441 2017-11-06 14:03:53 rcbuckiii Re: Superhet design questions
Meant to say AM filter is 4340 Hz low, not KHz.

14454 2017-11-28 21:14:11 rcbuckiii Re: Superhet design questions
Update on progress of my receiver build. Sorry it has taken a while to get back here.

The receiver I am building will actually cover 40 and 20 meters. However, for the time being I am just testing on 4o meters to keep things simple.

Since I was putting everything in shielded enclosures, I went ahead and installed everything necessary for both bands. This included the bandpass filters and low pass filters for the local oscillator. Since the LO is a Si5351A device, low pass filtering of the signal is an absolute must. I also had to install all the relays associated with band switching and the associated circuitry.

I finally got everything put together this past weekend. When I connected the receiver to the longwire antenna I had fed out my window, I could not hear any stations, only noise. It made no sense since everything was in shielded enclosures.

So I decided to put the receiver aside and throw together a NE602/LM386 40 meter receiver. I knew this receiver would work as I had built one about 25 years ago. I built the receiver yesterday and when I connected it to the longwire antenna, I only heard noise, no 40 meter signals. At this point I decided the problem had to be all the computer equipment in my office.

To confirm this, I decided to take the receiver into the back yard and connect it to the antenna. When I did this, I was immediately able to hear multiple stations on 40 meters. I went to the local Frys and bought 50 feet of RG58 and put up a temporary 40 meter dipole 10 feet off of the ground in the back yard. I opened the window of my office and ran the other end inside. I then connected it to the NE602 receiver and was able to hear 40 meter signals with no noise.

Today I put the superhet back on the bench and connected the coax to the antenna input. I was able to hear several 40 meter stations even though band conditions are currently bad. I made a few programming adjustments to get the signals centered in the crystal filter passband and all is good. The only problem I see now is that when I enable AGC on the Hycas amp, the signals drop to a very low level. I will address that problem tomorrow.

The bottom line is the emissions from the computer equipment (3 computers, a cable modem, a router, a WiFi router, and an ATA) was simply too much to prevent front end overload from an unshielded antenna connection.