RF front end design

[Space Varmint]

Hi, I'm building an HF receiver and I have almost completed the second IF strip. I will run it by cable RG-174U out of a shielded compartment to the a buffer amp and one more voltage amp and then the audio & AGC loop circuits.

Well, I have no pre-selector as of yet and most certainly plan to put one in. Here is my concern. I put the rf amp in a metal container along with the first mixer and 1st IF strip. The ist LO is isolated and comes from a PLL into the grounded metal compartment through coax RG-174U. I'm using a crystal filter that is very sharp and I see the selected signal come through very strongly but when sweeping the receiver with a signal generator I also see the IF frequency pass which is what is is supposed to do but also I see other frequencies pass through the crystal filter. I was scratching my head wondering where they came from. I'm sure some are harmonics and phase sidebands from the PLL but there are some odd balls getting through. Granted a good selectable pre-selector will get rid of most of them but I am concerned with ones that are close enough to be included in a band pass filter.

To make a long story short. Are there better ways to reduce these odd frequencies? I do see evidence that the LO is still getting into the rf amp and doing some mixing itself. Should I take out the rf amp and enclose it into it's own shieded compartment or will I still get some pick up through the power bus? I'm looking for optimum rf front end design both physical and electrical. I figure doing the rf amp right is a small price to pay for superior quality.

Any links or suggestions, lectures, whatever you got will be appreciated. Thanks

 

[RadioRon]

Hi SV

I'm concerned when you write that RF might come back over the power supply bus. This indicates that you are not confident that your power supply line filtering is adequate. Did you use coaxial feedthrough capacitors on all power supply and control wires going into the LNA box? You should. These capacitors are the optimum way to guarantee good shielding. If you use these, with a suitable value of capacitance (like maybe 1000 pF or 10,000 pF), then you will have insignificant RF getting through on those lines.

Years ago a friend of mine built his "ultimate HF receiver". He was an expert RF engineer. He built in the modular fashion, with major sections each in their own box, as you might be doing. He built each section using double sided pcb stock, and all components were mounted up in the air on insulated low capacitance standoffs that in turn were soldered to the ground plane of the main pcb. The boxes were sealed shut by soldering more blank pcb stock into a box around the main board, using the main board as the bottom of the box. He ran everything into and out of the boxes through either coax connectors or feedthrough capacitors mounted in a wall of each box. This method of construction gave the best isolation performance that I have ever seen. Admittedly, it would be a pain to repair inside a box because you have to unsolder the lid, but otherwise they were great.

If you are building the ultimate, you should aim for excellent shielding performance in the same way.

As for mystery mixer outputs, I presume that you are familiar with using nomographs to predict mixer products. I've seen what you describe and it can be quite mysterious wondering where all those mixer products are coming from. On the other hand, I see the fewest from double balanced diode mixers and the worst from home made single transistor active mixers of various kinds. In any case, reduction of products comes from your choice of IF and LO frequencies as well as the performance of the mixer itself. Also be aware of 2nd order products coming from your LNA or from your LO.

DO you have a preselector on your LO?

 

[Space Varmint]

Hi SV

I'm concerned when you write that RF might come back over the power supply bus. This indicates that you are not confident that your power supply line filtering is adequate. Did you use coaxial feedthrough capacitors on all power supply and control wires going into the LNA box? You should. These capacitors are the optimum way to guarantee good shielding. If you use these, with a suitable value of capacitance (like maybe 1000 pF or 10,000 pF), then you will have insignificant RF getting through on those lines.

Years ago a friend of mine built his "ultimate HF receiver". He was an expert RF engineer. He built in the modular fashion, with major sections each in their own box, as you might be doing. He built each section using double sided pcb stock, and all components were mounted up in the air on insulated low capacitance standoffs that in turn were soldered to the ground plane of the main pcb. The boxes were sealed shut by soldering more blank pcb stock into a box around the main board, using the main board as the bottom of the box. He ran everything into and out of the boxes through either coax connectors or feedthrough capacitors mounted in a wall of each box. This method of construction gave the best isolation performance that I have ever seen. Admittedly, it would be a pain to repair inside a box because you have to unsolder the lid, but otherwise they were great.

If you are building the ultimate, you should aim for excellent shielding performance in the same way.

As for mystery mixer outputs, I presume that you are familiar with using nomographs to predict mixer products. I've seen what you describe and it can be quite mysterious wondering where all those mixer products are coming from. On the other hand, I see the fewest from double balanced diode mixers and the worst from home made single transistor active mixers of various kinds. In any case, reduction of products comes from your choice of IF and LO frequencies as well as the performance of the mixer itself. Also be aware of 2nd order products coming from your LNA or from your LO.

DO you have a preselector on your LO?

Hey RadioRon! Thanks for reply. No no pre-selector on LO. It comes straight from the VCO through a couple of broad band amplifiers.

Yes, I'm shielding the stages. Trying to keep stuff that is comon to a particular frequency in separate grounded containers. Of course that is almost imposible completely unless I isolate the mixers too. I'm not doing that because I have done both before and got pretty good results putting the mixers in with the IF filters and amps.

I'm not sure what you mean by coaxial feed-through caps. Is this a particular type? You did remind me that by using different values of capacitor I can obtain decoupling of several different freqs.

I know what you mean about diode ring mixers but I think their best trait is they are resistant to blocking, partly due to their insertion loss.

I am not getting strange mixer products. The mixer is giving me what is handed to it by the LO and rf amp. I can actually see strange doubling and weird mixer products being produced in the rf amp itself. This is my concern. I need to clean up the rf amp. I am completely dissatisfied with it. What I am looking for here is the very best construction techniques for the rf front end. And it is not an LNA. I would be afraid to use an LNA in an HF rig. Too easy to pick up harmonics and other parasitics. I'm not concerned with gain so much as I will be putting a pre-selector on it. I just think I can do better with the rf amp.

Actually I am trying to get best results on a perforated PC board before I build gerbers for it. I can get away with it to some degree at HF frequencies. I just want to get rif of the strange mixer products coming from my rf amp.

 

[Space Varmint]

OK, I got rid of all of the odd balls and the rf amp is still in same metal box as the 1st mixer and Xtal filter and some 1st IF amps.

Here's the deal. It's something to do with the metal container. I can flex it and get horrible feedback and I can keep flexing it until I get perfect results. I think it's a ground loop problem. I need to isolate it but at least I know the design is OK.

*edit*

It's in that all important post mixer amplifier. I have a bypass capacitor on it and by bending it back and forth I can cause it to make the entire IF strip burst into oscillation or have a very clean output or anything variation in between. This really only showed up after I put it in the metal enclosure. Rf can be a weird animal...lol.

*edit* fixed

I cut the bypass capacitor off. Now I sweep the thing with about 15 MHz continuous bandwidth and the only thing I see getting through is the selected frequency and the frequency of the Xtal filter. That's it! Nothing else. It's working now!

 

[RadioRon]

Good news! I've had problems in the past with cracked capacitor lead connections, both in SMT parts and leaded parts. These can sometimes be hard to find because they are hairline cracks, very fine, but they make and break when the board flexes a tiny bit.

 

[Space Varmint]

Good news! I've had problems in the past with cracked capacitor lead connections, both in SMT parts and leaded parts. These can sometimes be hard to find because they are hairline cracks, very fine, but they make and break when the board flexes a tiny bit.

Thanks RadioRon Actually you were righter than me on this because you call out the mixer first. Close. It was the post mixer amp. I think what was happening is the rf amp (front end) was picking up some of the output from the post mixer amp. That's why it would burst into oscillation from time to time. So by cutting off the bypass cap I reduced some of the AC gain of the post mixer amp and it no longer was felt by the rf amp.

That's what I like about rf design. You don't just consider the wiring. A little more to it than that...lol.

 

[Space Varmint]

Good news! I've had problems in the past with cracked capacitor lead connections, both in SMT parts and leaded parts. These can sometimes be hard to find because they are hairline cracks, very fine, but they make and break when the board flexes a tiny bit.

Now that I think about it. I have had those exact same problems particularly with surface mount caps. You have to be careful flexing the boards too. I think that can crack the caps or resistors.

I just love a good HF receiver. You can get nice ones for about $150 that work surprisingly well. Then you get into the Kenwood's and Tentec's and Icom's and watch that price tag shoot up!

You know in that "Wireless Act" put out when Clinton was pres. they removed some of the frequencies you can listen to. Radio Shack even had to pull some of their scanners. People were hearing some interesting things from on board "Airforce 1" amoung other things.

HF don't need all that support equipment like satellites and repeaters etc. You could pretty much flip a switch and turn all that stuff off including telephones these days. Have you seen some of the data communications they are doing now days with AMTOR and various forms of Packet? It is amazing! It's getting like the internet with sending pictures all. Some of the modulation techniques and data compression have become quite sophisticated.

 

[RadioRon]

No, I've read some articles in QST but otherwise haven't been exposed to the new HF digital modes.

 

[Space Varmint]

No, I've read some articles in QST but otherwise haven't been exposed to the new HF digital modes.

That's right your a ham aren't you? You still take QST? Then you would be an ARRL member. IF so, good man. I need to be again. It's just that their technical articles really weren't up to snuff, not like the old days. Anyway, I'm WA4BJO. Hey, we need to exchange QSL cards...lol.