HF transciever - components design

[MrWifiHifi]

Hello. 
I would like to put together a receiver for the HF band (0-30MHz) from the parts I have available. I have a lot of parts from old TVs and radios, but so far I'm only missing crystals for USB/LSB detection and a IF filter for them.  But I might be able to find that in time. I am mainly interested in how to calculate, for example, such transformers for individual stages and mixers. As far as I have found any information, mostly only the input/output impedance, their ratio and the ratio of the number of turns were solved. But I couldn't find anywhere how to get to the number of windings of the coil and its impedance for the given frequency at which it should work, or how to design the given transformer for a wider working range. Would someone be willing to explain it to me, or would someone be able to recommend literature where this issue is addressed?

It would probably be good to mention that I would like to partially copy the wiring of the Kenwood R-1000 with a dual gate mosfet mixer. Here is a link for schematics:
https://www.qsl.net/wb4kdi/Kenwood/R-1000/R-1000-RX_UNIT-schematic(hi-res).gif
And link for the page:
https://www.qsl.net/wb4kdi/Kenwood/index.html

Thanks in advance for any useful info.  If there is interest, I can also provide information about how I would like my receiver to be designed (block diagram and practical design of circuits). Because I decided to use the parts available to me, I will have to make a lot of changes and compromises. For now I would like to build only the receiver part, later I would like to add the transmitter part as well

[Solder_Junkie]

Dual gate mosfet mixers are an idea from the 1970s, these days designing almost everything for 50 Ohms in/out makes it much easier to make a modular receiver as each part can be tested using ordinary test equipment.

A good starting point is the Hybrid-Cascode (HyCas) IF circuit by Wes Hayward and Jeff Damm. This is a very easy to make circuit that includes AGC and works really well. Many home made amateur radio receivers/transceivers use this circuit.

You are welcome to copy my receiver ideas... See: https://www.qsl.net/g4aon/g4aon_rx/

SJ

[LM21]

I  once had a an FM tuner  made with RCA 40822 dual gate  mosfets. The tuner was a nice to use. But I think the  fets were too good mixers, I could hear everything from police  and  taxis to  AM  from airport. This will be much worse on HF.

I also made a converter for police radio with dgfets, but  that was then. They were nice  parts. Buy a NanoVNA and tinySA  and  use  modern components.

Today, I would use a Tayloe detector/mixer  for example  from  https://qrp-labs.com/

[mike449]

Wes Hayward's (and co-authors) book is a very good starting point. In fact, it is good for the whole journey.
https://books.google.com/books/about/Experimental_Methods_in_RF_Design.html?id=GGUJAAAACAAJ

As for crystals, you can replace them with Si5351a boards from Aliexpress, Adafruit or QRP-labs (the oscillator application).

[MrWifiHifi]

Dual gate mosfet mixers are an idea from the 1970s, these days designing almost everything for 50 Ohms in/out makes it much easier to make a modular receiver as each part can be tested using ordinary test equipment.

A good starting point is the Hybrid-Cascode (HyCas) IF circuit by Wes Hayward and Jeff Damm. This is a very easy to make circuit that includes AGC and works really well. Many home made amateur radio receivers/transceivers use this circuit.

You are welcome to copy my receiver ideas... See: https://www.qsl.net/g4aon/g4aon_rx/

SJ

Hello
As I wrote, I would like to use the parts that I have available. I thought of using that mixer with a dual gate mosfet when I saw it in several older connections, for example Kenwood R1000, TS120 or in TCVR schemes that were developed here. Would it be better to use a diode mixer instead of a dual gate mosfet?

I  once had a an FM tuner  made with RCA 40822 dual gate  mosfets. The tuner was a nice to use. But I think the  fets were too good mixers, I could hear everything from police  and  taxis to  AM  from airport. This will be much worse on HF.

I also made a converter for police radio with dgfets, but  that was then. They were nice  parts. Buy a NanoVNA and tinySA  and  use  modern components.

Today, I would use a Tayloe detector/mixer  for example  from  https://qrp-labs.com/

I'm trying to save some money for NanoVNA and TinySA. I've been looking at buying them for a long time. For now, I only have a GDO and an oscilloscope at my disposal.

Wes Hayward's (and co-authors) book is a very good starting point. In fact, it is good for the whole journey.
https://books.google.com/books/about/Experimental_Methods_in_RF_Design.html?id=GGUJAAAACAAJ

As for crystals, you can replace them with Si5351a boards from Aliexpress, Adafruit or QRP-labs (the oscillator application).

Thanks for the tip on the book. I have to check it out.

[iMo]

The dual gate mosfet mixers in your example are not the best mixers, unless you would use some special parts with high Ids current in a ring.
For example the simple diode mixers are better (do not have "gain", however), and for example the FTDX-10 is even using the $0.20 switches in the first mixer (the cheapo chips used to use in the plethora of the cheapest SDR designs) as those are pretty good.
The stuff around the mixers is a kind of science, so you have to spend some time to study the stuff first.

[ftg]

Hello. 
I would like to put together a receiver for the HF band (0-30MHz) from the parts I have available. I have a lot of parts from old TVs and radios, but so far I'm only missing crystals for USB/LSB detection and a IF filter for them.
...

If using salvaged parts is desired, you will get to measure a lot of crystals to try and find enough suitable one's for an SSB filter.
4.4something PAL colorbusrst crystals might be plentiful enough for that.

An alternative method could be making the SSB via the phasing method. Then the hardpart would be finding matching parts for the two passive all-pass networks to do the required phase shifts.

Two examples of the passive all-pass network for SSB generation would be Juma TRX2:
http://www.jumaradio.com/juma-trx2/rev-b/index.html

And the QRP-Labs Polyphase kit:
https://qrp-labs.com/polyphase.html

The QRP-Labs store page also links to this excellent paper on understanding how those polyphase networks work and are designed, it's a bit math heavy though.
http://antennoloog.nl/data/documents/Understanding_and_designing_Polyphase_networks_V4.0.pdf

As for the mixers, if you really want to go with dual-gate mosfets, you will need tight filters and many TV tuner cans of the same type to salvage similar one's.

I might personally go with some mux chip like fst3253.

If you want to make it a superhet, then it could be 6.5/6/5.5MHz for the last IF, as ~300kHz wide ceramic filters for that are used for the TV sound subcarrier and thus are common in scrap TV's. A 26MHz TCXO salvaged from some old GSM phone will yield convenient 6.5MHz when divided by 4, so it can be used to generate the IQ clocks for the RX and TX mixers.
Then the final selectivity can be done at baseband using op-amps.

The filter for the 1st IF would then have to be narrower than 6MHz.

The hardest part, if using salvaged parts would be generating the local oscillator for the 1st mixer.
TV tuner LO's usually can only do 25kHz step the narrowest, likely won't go down low enough and worse, the LO is usually hard to extract from the can.

Some 3eur aliexpress arduino, some salvaged 1602 LCD and aliexpress si5351 module for the actual LO would likely be superior to anything built from salvaged TV parts.

But it is an interesting challenge!

[MrWifiHifi]

Hello. 
I would like to put together a receiver for the HF band (0-30MHz) from the parts I have available. I have a lot of parts from old TVs and radios, but so far I'm only missing crystals for USB/LSB detection and a IF filter for them.
...

If using salvaged parts is desired, you will get to measure a lot of crystals to try and find enough suitable one's for an SSB filter.
4.4something PAL colorbusrst crystals might be plentiful enough for that.

An alternative method could be making the SSB via the phasing method. Then the hardpart would be finding matching parts for the two passive all-pass networks to do the required phase shifts.

Two examples of the passive all-pass network for SSB generation would be Juma TRX2:
http://www.jumaradio.com/juma-trx2/rev-b/index.html

And the QRP-Labs Polyphase kit:
https://qrp-labs.com/polyphase.html

The QRP-Labs store page also links to this excellent paper on understanding how those polyphase networks work and are designed, it's a bit math heavy though.
http://antennoloog.nl/data/documents/Understanding_and_designing_Polyphase_networks_V4.0.pdf

As for the mixers, if you really want to go with dual-gate mosfets, you will need tight filters and many TV tuner cans of the same type to salvage similar one's.

But for now, I would like a radio with two intermediate frequencies, the first higher, the second lower. For the first one, I would like to use wider filters from the image intermediate frequency, I think the label is OFW K 1950 for 39MHz, but I think there is a rather wide band (about 6MHz). For the second intermediate frequency, I originally wanted to get a 9MHz filter with a width of 2400Hz, but also the necessary crystals for LSB/USB detection. However, if the 4.433MHz crystals from PAL could be used for the filter and then detuned enough to be used for LSB/USB detection, I'd rather go that route.

Unfortunately, the phasing method doesn't tell me anything, but I'll look it up. So far, I only know detection using BFO.
For those mixers with dual gate mosfets, what specific filters do you have in mind? and why do they have to be tight?

I might personally go with some mux chip like fst3253.

If you want to make it a superhet, then it could be 6.5/6/5.5MHz for the last IF, as ~300kHz wide ceramic filters for that are used for the TV sound subcarrier and thus are common in scrap TV's. A 26MHz TCXO salvaged from some old GSM phone will yield convenient 6.5MHz when divided by 4, so it can be used to generate the IQ clocks for the RX and TX mixers.
Then the final selectivity can be done at baseband using op-amps.

The filter for the 1st IF would then have to be narrower than 6MHz.

The hardest part, if using salvaged parts would be generating the local oscillator for the 1st mixer.
TV tuner LO's usually can only do 25kHz step the narrowest, likely won't go down low enough and worse, the LO is usually hard to extract from the can.

Some 3eur aliexpress arduino, some salvaged 1602 LCD and aliexpress si5351 module for the actual LO would likely be superior to anything built from salvaged TV parts.

But it is an interesting challenge!

I would prefer to use a wider filter for the first IF and a narrower one for the second, so that I don't have to deal with selectivity with OP amps. As I wrote above, if 4.433MHz crystals could be used for the filter and BFO, I would be in favor. As for the LO, I'll probably have to rebuild it, one way or another. I suspect that the tuner I have has LO running at a frequency somewhere around 300MHz, but I'm not sure. But the oscillator is fitted there with a BF506, which could theoretically be used.


Also regarding those transformers for IF and mixers. So I've found this design process from older book I have.

First, it is necessary to determine the inductance for the lowest transmitted frequency, namely L1=Z1/(2*pi*fmin). This formula can only be used for an untuned "broadband" transformer. I assume that the inductance for L2 is then determined using the ratio for Z1 and Z2. The question is how to determine the impedance and inductance for an autotransformer and for a center-tapped winding?

What I noticed with some diode mixers, for example, is that there is a relatively small number of turns, both on the primary and on the secondary coil. I don't know how to do with coupling transformers. As for the thickness of the wire, it should be used with a diameter of 0.1-0.2 mm.

If anyone has any more detailed information on this, I would be happy if you could add or correct it.

[ftg]

...
But for now, I would like a radio with two intermediate frequencies, the first higher, the second lower. For the first one, I would like to use wider filters from the image intermediate frequency, I think the label is OFW K 1950 for 39MHz, but I think there is a rather wide band (about 6MHz). For the second intermediate frequency, I originally wanted to get a 9MHz filter with a width of 2400Hz, but also the necessary crystals for LSB/USB detection. However, if the 4.433MHz crystals from PAL could be used for the filter and then detuned enough to be used for LSB/USB detection, I'd rather go that route.

It is viable to make SSB filters from crystals for that frequency.
Some references about it:
https://w0qe.com/Projects/crystal_bandpass_filters.html
https://www.giangrandi.org/electronics/xtalfilters/xtalfilters.shtml
One essentially measures the oscillating frequency of all the crystals with the same nominal value and takes four with the frequency closest to each other.
The LSB/USB crystals can them be the one's the most above and below that frequency from that same pile of crystals.
Althought the oscillator might have to be built in a way that allows pulling the crystals to the desired frequency.
Super-VXO is one which people have had good sucess with.
https://www.qsl.net/7n3wvm/supervxo.html
https://www.ke7hr.com/caveradio/KE7HR-5CrystalVXO.pdf

For those mixers with dual gate mosfets, what specific filters do you have in mind? and why do they have to be tight?

Dual gate mosfets don't deal well with large amounts of strong signals on the band. And todays world has absolute tons of garbage from all the electronics around us polluting the shortwave bands.
So while dual gate mosfets have low noise and conversion gain, they are not as robust as, say a diode ring mixer.
Thus the receiver needs to have tight band pass filters per band in front of it, to limit the strong signals that make it to the mixer.

I would prefer to use a wider filter for the first IF and a narrower one for the second, so that I don't have to deal with selectivity with OP amps. As I wrote above, if 4.433MHz crystals could be used for the filter and BFO, I would be in favor. As for the LO, I'll probably have to rebuild it, one way or another. I suspect that the tuner I have has LO running at a frequency somewhere around 300MHz, but I'm not sure. But the oscillator is fitted there with a BF506, which could theoretically be used.

It is easier to make a 2700Hz wide filter on 300 - 3000Hz than 4.43361875 - 4.43631875 MHz, thus why using phasing to create the SSB signal could be desirable. LSB/USB switch is also just swapping the I and Q signals to/from the mixers, so no BFO crystals required either.

As for the LO, I don't think a free running VFO will be viable in anyway here. to cover all of the HF bands the LO would need to tune 30 - 69MHz with the 39MHz 1st IF.

I don't have any particularly good references at hand about phased SSB generation, this one I found the best myself, but it is rather math heavy:
http://antennoloog.nl/data/documents/Understanding_and_designing_Polyphase_networks_V4.0.pdf

This has less math and is aimed at students:
https://www.eng.auburn.edu/~troppel/courses/TIMS-manuals-r5/TIMS%20Experiment%20Manuals/Student_Text/Vol-A1/a1-07.pdf

[iMo]

Here is a basic reading on mixers for beginners..

[MrWifiHifi]

I also found something like this while browsing the internet. You need to translate the page.

https://www.circuitsonline.net/forum/view/99845

Although it is a design with transistors, it uses 8866kHz crystals, which can also be found in TV sets.
And then another page with all possible schematics.

https://www.qsl.net/va3iul/Homebrew_RF_Circuit_Design_Ideas/Homebrew_RF_Circuit_Design_Ideas.htm

[ftg]

Iulian Rosu's page is great for ideas and inspiration.
The "RF BASICS and THEORY" section on his main page https://www.qsl.net/va3iul/ is also decent for learnign the keywords to search to get more indepth information on some concept.

G4USP has made a 80m SSB trasnceiver that only uses BC547 NPN transistors in it. the SSB filter in that design is made with 6MHz crystals.
https://www.qsl.net/g4usp/BC547%20HF%20TXCVR/All%20transistor%20BC547%20SSB%20HF%20TXCVR.htm

[MrWifiHifi]

So for now I have some such idea about transceivers. The block diagram clearly shows the receiving and transmitting part, although I would prefer the receiving and transmitting path to be bidirectional as with some radios. I would like to add the option to broadcast CW and audio filters for CW and SSB.

[ftg]

You could use the uBITX bidirectional Termination Insensitive Amplifiers if you want.
Depending on your 1st IF, common jellybean BJT's (BC547, 2N3904, 2SC1815, etc.) could be usable for all fixed gain IF amps.

µBITX v6 schematics:
https://www.hfsignals.com/wp-content/uploads/2019/12/ubitxv6.pdf

µBITX circuit description:
https://www.hfsignals.com/index.php/ubitx-v6/

[MrWifiHifi]

I forgot to ask. I have seen the A244D (TCA440) circuit used as a detector for SSB/CW. There was an unused internal oscillator, the signal was fed from an external BFO ​​oscillator with crystals, LSB/USB was switched by diodes. But what I noticed was that the entire IF part was not used. My question is, is it possible to use this circuit in such a receiver as a whole, or is it better to use a transistor circuits?

I am attaching a photo of the connection. It is from the ODRA receiver, which was once produced in the former Czechoslovakia.

[MrWifiHifi]

You could use the uBITX bidirectional Termination Insensitive Amplifiers if you want.
Depending on your 1st IF, common jellybean BJT's (BC547, 2N3904, 2SC1815, etc.) could be usable for all fixed gain IF amps.

µBITX v6 schematics:
https://www.hfsignals.com/wp-content/uploads/2019/12/ubitxv6.pdf

µBITX circuit description:
https://www.hfsignals.com/index.php/ubitx-v6/

A very interesting solution. I like it. I saw a similar solution, that is, that the same circuits are used for reception and transmission, also in the TCVR Atlas 210. In our country, radio amateurs once copied it entirely and built it from the parts that were available here at that time.

Another question about the two-way use of individual blocks. Is there any difference in the functionality and quality of a given TCVR if separate circuits are used for RX/TX or if they are combined, i.e. bidirectional?

[CaptDon]

Are you doing this just for fun? There are hundreds of thousands of good used SWL receivers and Ham transceivers that receive 100KHz to 60MHz and with the removal of the secret resistor or clipping a certain wire or the secret 'power on button press' can transmit almost every frequency it can receive. I have built some special purpose single frequency receivers that were optomized for a specific function and got marginally better results than what I could have simply bought off of a store shelf. Being an old dinosaur ham members of the club home built a lot of items and we used to joke about 'appliance operator hams' but off the shelf equipment got so incredibly good at reasonable prices everyone has become an appliance operator for their main rig. It is still fun to build a tuned crystal rig with high impedance headphones so kids can hear AM radio without any form of power supply. One of the most sensitive receivers I ever had was a Pontiac car radio from the late 60's and a shortwave pre-converter from Gonset. Wow did that combination make the best SWL rig!! So much fun as a kid listening to BBC. Radio Norway, Swiss Radio International and so many others and the audio quality was so good, even better than what you would hear on the Swans, Kenwoods, Yaesu and others. If you had ten years to waste you still probably won't home build anything to compare with a Kenwood TS-440 and even those are dinosaurs by current standards.

[A.Z.]

what's wrong in willing to build a TRX ? (or an antenna)

[ftg]

To me the fun is in the building, less in the DX.
So I can understand building one's own equipment.

That said, I do have my Yaesu FT-897 and FT-817 for when I want to use a commercial rig.

[LM21]

There is nothing wrong about building your own, but it is hard work.  And sometimes too much work, if it does not work, no matter what you do. "To me the fun is in the building, less in the DX. " I understand  that well.

But that BC5** transceiver looks nice. I only wonder  how easy it is to make or get all the coils. And coils are often a problem.

Try to get the "Experimental Methods in RF Design" book. It has receivers from  very easy to very  hard to build.

[markus_jlrb]

>I only wonder  how easy it is to make or get all the coils

Depends what equipment you own.
Nowadays you could purchase a cheap VNA and SA, look for LiteVNA64,  tinySA Ultra (eleshop.nl)
and build your coils or other staff and be able to check the parameters not only guessing them.
As well as DSO's becoming very cheap to inspect your LO, PA,  or Audio output.

So things get much more easier as 20 or 40 years ago for hobbyists or students DIY projects.

Good luck in building your own TRX and enjoy your hf experience you made in between.

Markus

[Solder_Junkie]

"I only wonder  how easy it is to make or get all the coils. And coils are often a problem"

It is not a problem provided you use toroids and wind your own. FT50-43 and T50-2, T50-6, and similar cores are easy to obtain. I have built a 10 band HF receiver, 10 band CW transmitter and 4 band SSB/CW transmitter, all use home wound coils on toroid cores.

The Hycas IF board avoids the use of commercial coils/transformers and is at the heart of many home made receivers, it really works well and is easy to make.

SJ

[iMo]

The Amidon cores and others (powder and ferrite) have been invented in order to provide people with a possibility to make coils and transformers much easier and fast and without a need to measure them (with some +/-20% max).
They all got some well defined "AL" in nH/turns^2, where the AL param depends on the color or material type and a bit on the toroid's size and frequency (there are datasheets with tables and graphs). They have also got their Q and TC, etc., well defined.

The AL is typically from 8 to 2000.

L[nH] = AL * turns^2

The most manufacturers moved to digital also because the tunable coils, transformers, xtals and xtal filters become the most expensive and hard to source parts in their BOMs. In SDR radios they usually use standard value smd coils in input band filters (precision is not critical, none tuning required) and then in the output PA filters and/or antenna tuners -  again the larger amidon toroids where the L is not critical and +/-20% is ok. The transformers in the matching circuits or in the PA (driver and finals, swr couplers) are usually ferrite binocular cores where the L is not critical.

So almost no laborious tuning of the inductors or transformers is needed today - which saved quite huge amount of the manufacturing effort/costs.

[A.Z.]

well...

https://bec.co.uk/product-category/toko/coils/

[MrWifiHifi]

So, in general, to the design of devices for HAM radio. I know that nowadays everything can be bought for relatively little money, be it transceivers or measuring instruments. I don't know about others, but I build something and learn how it works along the way. Because if I were to buy a ready-made device, I would slowly not know how and what works there. This is how I experimented, for example, with GDO, both with the circuit and also with winding the coils for the necessary ranges. I learned quite a bit from this process. I myself like to tinker with electronics, as it always fascinates me what and how it works.