At my second house, I have no way of listening to CW or SSB, unless I hook up my homebrew 20 meter SSB transceiver, or my FT-817 transceiver, so I’m thinking I might try to build a simple communications receiver to solve this perceived problem.
My first receiver build was a direct conversion (DC) receiver, that did not work very well, but motivated me to learn how to build better superheterodynes. I’ve attached a copy of the magazine article for that first receiver. So, I’m thinking I might have a try at another DC receiver. These seem to be the new state of the art as SDR receivers.
As with the general coverage broadcast receiver, commercial off the shelf parts will be used if there is an advantage to do so, and the cost is reasonable. My goal will be realized with a simple kitchen table receiver that works well for its intended purpose and can be used on a daily basis. A secondary goal is to make it a true kitchen table receiver, by doing all construction at my second house on the kitchen table, with no resort to any of the tools / equipment in my ham shack. A multi band receiver would be nice, but probably is not essential, at least at first.
I have an extra working QRP Labs VFO left over from the development of my 20 meter SSB transceiver, so it will probably be used for frequency synthesis, at least in the early stages of development, if not permanently.
This will be an experiment / work in progress sort of thing, sharing time with a couple of other projects I want to do. My simple SSB transceiver took three years to satisfactorily complete, and the simple general coverage broadcast receiver took another year after that, so those are the sort of time frames I operate in when tinkering / experimenting with stuff.
73,
Win W5JAG
I think this is the keeper VFO configuration.
Installed here is a "0-55 MHz DDS VFO" as widely advertised on eBay and AliExpress for very reasonable prices.
They are built around an Analog Devices DDS chip, so the output is considerably cleaner than the Si5351 devices. The clean output, and programmable offsets, will make construction of a companion transmitter, if I decide to go there, considerably easier.
Output is slightly less than the Si5351 devices, being about 0 dBm; I have not adjusted the attenuator on the mainboard. I think the AD831 mixer is okay with the lower LO level.
These are quite nice devices, and really give a homebrew radio much of the feel and features of a store bought radio. They are quite feature rich, including band switching, memories, RIT, dual VFO's, offsets for proper SSB and CW, multiple tuning steps including variable rate tuning, and quite a bit more.
The display is very legible - when RIT is activated, the frequency offset is displayed in the upper right corner of the display.
Here, as in my 14 MHz SSB rig, the VFO is paired with a nice Bourns detentless rotary encoder to give a store bought radio tuning feel. Unlike my SSB transceiver, here I have used the onboard switches, which required the use of a smaller tuning knob. On my SSB rig, I used outboard switches which allowed me to use a much larger tuning knob on the same vertical height panel. This is easy to do, as all of the switches are brought out to the XH connectors on the back of the DDS, for easy connection of outboard devices. These VFO's are also capable of direct frequency entry with a key pad. On the SSB rig, you can see the four most important switches brought out to the right of the display, and a blue 1602 LCD is swapped for the stock yellow.
Out of the package, there are a few mods that need to be made to make these devices work optimally. They are delivered with absolutely zero documentation. I believe the origin of these devices is an Indian amateur, I have attached a PDF that gives enough information to set up and operate these devices. It shows everything but the hookup of a key pad.
This is really turning out to be a nice little kitchen table communications receiver. It exceeds my expectations for the project.
73,
Win W5JAG
.
I changed the VFO sketch / display. I’m not totally pleased with the change, I did it for a questionable reason: to make band changing easier (on a receiver that presently only has one band filter). The other sketch I was using also had a limited number of tuning steps, the maximum being 10 KHz. If you want to go from 80 meters to 10 meters, that’s a lot of tuning steps at 10 KHz and possibly why the cheap encoder wore out so fast.
At some point I may be able to significantly rewrite sketches to better suit my needs, however, that day is not this day, so I am limited to making changes that are pretty self evident and explanatory in the code.
Anyway, this is the VFO / Sketch / Display I am presently using:
10kHz to 225MHz VFO/RF Generator with Si5351 - Version 2 - Hackster.io
This also required 3D printing a new temporary chassis.
This sketch has some good features - it offers multiple tuning steps, and I was actually able to figure them out a bit and make some changes to better suit my use. It offers push button band switching - in fact, it has too many bands, so I need to go through and figure out how to remove everything above 6 meters, and probably some of the shortwave broadcast bands.
The display has a bit more information, including an analog bar graph that translates volts ( in a software adjustable range ) to a fourteen (14) segment bar graph. The display is surprisingly readable, the frequency part anyway, despite its small size. I used a native white font OLED which I think is more legible than the blue font devices.
I believe it uses a different Si5351 library than the other sketch, as this one allows software setting of the “drive” to the Si5351, meaning there is some software control over the power output of the device. The default in this sketch is 8 ma, which translates to +10 dBm from the clock output. I left it at this level, as my onboard attenuator is already set for this level of drive to the AD831.
It has significant drawbacks. There is an audible, multi note, warble as the VFO changes frequency steps. This is not really audible in a band with normal activity, but it sure is on a quiet band like 10 meters. I find this highly annoying. I don’t know if this is an Si5351 artifact, or an OLED artifact as it updates. I do not recall the other OLED sketch doing this, but it might have and I just don’t recall. The 1602 LCD sketch I was using definitely did not have this issue. There is a BIG pop as it changes bands, again, highly annoying.
I’m having trouble getting the analog bar graph to properly work as an S meter. It sort of acts like it is oscillating, so for now I just have the input grounded to keep it visually quiet. And audibly quiet - I also noticed it injecting the same warble into the radio. Maybe I just need better bypassing / isolation on the display.
So, for now, it is what it is. Not sure if this firmware / display is a keeper or not. Still a work in progress.
Here's some clips of the device in action: a quiet SSB QSO on 10 meters about a half hour after local sunset, and the cacaphony of 20 CW a bit later. Both of these were direct to the input port on the MMIC - not fornt end filter at all.
I was a bunch of software updates behind on the Red Pitaya, so I upgraded it to the latest firmware that can be run without upgrading the Linux version, and tested it out by tuning up the bandpass filter on the front end.
Ultimately, it seemed like the best compromise for the top coupling cap was 22 pF. I started with 47 pF. The -3 dB points look to be about 3600 Khz and 4050 KHz and 3500 Khz is about - 7 dB. The receiver has plenty of gain to deal with that. Pics are before and after tuning it.
The before and after difference is pretty stark. I'm not sure how one of these top coupled double tuned bandpass filters could ever be tuned up without at least some type of crude spectrum analysis.
A couple of minor improvements, and one rather dramatic improvement made today.
The cheap rotary tuning encoder was just worn out, quite prematurely, I think. It was suffering from such severe contact bounce and erratic output that it was just unusable. I had a spare in my RF building tool kit, so I replaced it.
I replaced the 1N4148 silicon diodes in the meter rectifier with leaded 1N60P germanium diodes. This has the desired effect of making the meter more sensitive to weaker audio, but some fine tuning may still be required. It might be desirable to make the gain of the LM386 meter driver variable to complement the trim pot in series with the meter positive lead. It could be that trying to make a simple audio strength meter as sensitive and linear as a proper AGC driven meter in a superheterodyne is a hopeless undertaking.
The dramatic improvement was replacing the LM386-1 with a LM380N-8. I’d been considering replacing the LM386-1 with an LM386-3 to get more power output on the existing 10 volt DC bus. That would, however, only net me about 3 dB more power output and it would still be an LM386. I had some LM380N-8 in stock, so I went with that. 10 volts is the minimum Vs for an LM380, so I connected it directly to the incoming DC bus, typically 12-14 volts depending on how the power supply is set. The receiver plays a LOT louder, and sounds better with less distortion. Gain is increased 8 dB over the LM386N. This receiver has high gain, and is completely stable.
The LM380 circuit is pretty minimal - I used a 100uF electrolytic output cap and a series RC zobel network of 10R and 10 nF on the output to ground; a 10 uF non polarized ceramic on the bypass pin, I also brought out an unterminated lead from the bypass for possible future use as a mute line; Vs is taken through another 10R resistor direct to the main DC bus, bypassed with a 100 uF non polarized ceramic capacitor. The inverting input is grounded and the chip is driven through the non inverting input.
Considering that I initially wrote:
I seem to be close to done here The radio works well, it’s completely suitable for daily use. It is so much better than my first direct conversion receiver which was just a poorly working toy. This radio does not drift, does not overload, has digital readout, is sensitive down to atmospheric background noise on all bands, has AGC, and plays loud and sounds good. It's a real radio, not a toy. It's only real flaw is that it is DSB. I'll concede that is a big flaw. It’s 100% a kitchen table build.
I still need to make some decisions on the multi band capability, specifically whether I want to go full general coverage with the previously built bandpass filter board, or just add one or two more narrow band filters. I also need to sort out what type of display I want to use, and sort out the metering situation, so still some tweaking to do, but the basic radio is finished.
73,
Win W5JAG
I added metering capability, but not without some unexpected difficulty.
After surveying the usual simple ( and complex ) designs in the published amateur literature, a near universal theme arose for audio derived metering - some type of JFET source follower, followed by a single stage amplifier, usually a BJT, then a half wave rectifier and filter, and a microamp meter, all virtually identical. The connection point in the receiver is more ambiguous, but it obviously has to be on the low level side of the volume control, or different volume control settings will alter the meter response.
So with safety in numbers in mind, that is what I built, using a ubiquitous leaded MPF102 JFET as the source follower, a mmbt3904 amp, and 1N4148 diodes in the half wave rectifier. The meter is a cheap edgewise CB radio type, probably 50 uA movement. I connected it to the output of the product detector, which is at the 32 dB gain point in the radio.
With disappointing results. As in, no results whatsoever. Nothing budged the sensitive meter.
So I connected it to the top of the volume control, which is at the 100 dB gain point in the radio, and simple background noise pinned the meter. I changed the meter sensitivity resister to 500K and was then able to adjust the meter sensitivity to less than full scale.
Now the obvious problem is that, unlike almost all, if not all, other DC and regenerative receivers, this receiver uses AGC. Very effective AGC. When the meter is adjusted for background noise, only the very strongest signals would provide any positive meter deflection, and then only two or three needle widths, because of the strong AGC action. The better method was to adjust the meter for half scale on background noise, and then watch how far the meter deflects BELOW that point when a station stops transmitting, as that minimum point represents the gain reduction before the AGC recovers and brings the receiver back to maximum gain. Obviously, the stronger the signl it was hearing, the more gain reduction it has to apply to hold the output level constant. But this is not really what I wanted, so about the only things I learned are: 1) the LM370 AGC subsytem is highly effective, and 2) while the 20 dB or so gain of a single stage amplifier is insufficient, the full gain of my high gain receiver is way too much.
I thought about this a bit, and decided I would reconfigure the first LM741 op amp (that I was using as a 3 KHz filter), into a gain stage, and use that to drive the meter circuit. I didn’t care much for that filter - after using it for a week or so, I felt it didn’t add much to the AGC action or selectivity, and whatever it did add was outweighed by a propensity to ring at the high volume levels I sometimes need if I am in another room.
That worked, but after some more thought, I had an epiphany that this was taking a lot more effort, and using more parts than I needed, to actually accomplish what should be a simple task.
So, the epiphany here was that what I really needed was a separate audio strip dedicated just to driving the meter, and - drum roll here - since an LM386 is really just a power op amp, and since we discussed earlier in the thread that a single LM386 can have its gain set anywhere between 26 dB and 73 dB ( if you don’t care how it sounds at the high gain end ), and since this would only be driving a meter, who cares how it sounds, one LM386 can do the job.
And it can.
I built a source follower from an MPF102, connected that to the output of the product detector, and then connected my high gain AGC audio strip to it so it would be isolated from any shenanigans going on in the meter strip.
Then, I whipped up a minimum circuit using an LM386N-1, I used the usual decoupling, a 10 uf bypass cap at pin 7, but omitted the output zobel network and used a small 100 uF cap at the output, set the gain to 46 dB, connected it direct to the output of the product detector, and - it worked. I reduced the meter sensitivity pot to 2K to get a better range of adjustment in my particular setup. I didn’t even use an “expensive” LM386 - the part used here is one of the 10 / $1.00 USD types from AliExpress - undoubtedly some type of reject, counterfeit, or some other dubious origin.
It worked well enough that I stripped off all the other now unneeded stuff to clean up the board.
There is still some room for tinkering here - I think diodes with a lower junction voltage would give a better response on weak signals, so I will probably tinker with that. That could lead to perhaps making the gain of the LM386 variable to adjust meter sensitivity in place of the pot at the output of the rectifier block. A smaller output cap might be okay, or even better. It might be that putting the source follower at the input to the meter circuit would be better than using it to feed the high gain strip.
The point here is that it appears to me that the usual way of doing this is over complicated when a single LM386 - a part everyone has in their junkbox - can do the job better and cheaper. Any DC or regen receiver can add meter eye candy with little additional effort or expense.
Win W5JAG
Block diagram to date.
Win W5JAG
Hooked up the RF attenuator to verify that it was in working order. I haven't yet measured the attenuation, it sounds like at least 10 dB with the present 5K control pot and 3K3 current limiting resistor. It can be configured for more attenuation, I think, but I doubt that will be necessary. I have yet to have a reason to think it even needs an RF attenuator. The HP MMIC RF amp in he front end is sufficiently quiet - even at maximum RF attenuation, hooking up my random wire antenna to the auxiliary input clearly increases the background noise in the receiver at 29 MHz. The overall end to end gain seems to be about right.
I had an XH connector for the RF attenuator pot but managed to rip it clean off the board. I didn't feel like replacing it right away.
Win W5JAG
I added another 3 KHz cutoff low pass filter ahead of the AGC subsystem, same circuit as before. The purpose here is to minimize wideband noise going into the AGC subsystem so that it is not acting on audio that I am not listening to, and, further, to provide an audio pick off point ahead of the AGC that, likewise, will not include, in part, extraneous wideband noise, in the event that I decide to add some kind of signal strength metering.
I constructed this last week, but weather considerations forced me to hightail it from the lake house, so I have not had much time to evaluate the addition. I got back last night, and, my initial impression, listening to some nets this morning, is that it has made a susbstantial improvement in the AGC action.
Both of my 3d printers have been down for the better part of a year. I finally got the huge one at my office working again, so I have printed out a quickie test chassis, better suited to the form factor of the PCB used here. This is not a final variant - it has some significant flaws - but is a huge improvement over what I was using, which was a reject from the general coverage broadcast receiver build. This variant is small ( so it would print fast ) and mounts the VFO under the main circuit board, so that the ground plane of the main circuit board shields ( sort of ) the VFO from the receiver.
73,
Win W5JAG
Increasingly, this variation of the DC receiver is looking like a keeper.
Continuing the incremental improvements, RF filtering and blocking ( a 4.7 nF cap and 1 mH choke) was added to the output of the product detector. The 2.2 uF coupling capacitor was combined into an RC high pass filter with an Fco of 270 Hz.
The significant improvement was the addition of an Automatic Gain Control (AGC) system in the form of a National Semiconductor LM370 communications subsytem IC. This is a very, very, old chip - dating all the way back to 1968. It has not appeared very often in the amateur literature ( I've only found two instances) and I am only aware of one commercial amateur product that used it, the Hallicrafters FPM-300, which used the chip as a speech compressor. The designer of the chip, Bob Hirschfeld, then W6DNS, is arguably even more interesting than the chip. An MIT grad, he designed the LM370, and probably all of the LM37X line for National, and ( I think ) the LM3909 LED flasher. Following a nasty divorce, he quit semiconductor design, went to law school, pretty much single handedly began the fathers rights movement that was just beginning when I got out of school, and then somehow managed to wind up disbarred.
Anyway, the data sheet for the chip is attached. Notable is that it fetaures 40 dB gain reduction, and 35 dB forward gain, for a 75 dB spread at room temperature. It features additional control inputs that can incorporate exernal amplifiers into the AGC control loop, and that is how my LM370 is configured - everything except the LM386 power amp is in the AGC loop. Also notable is the rather high, by today's standards, THD when the chip is in gain reduction. Because the chip provides additional significant voltage gain, I turned the LM386 down to the default voltage gain of 20 (26 dB). My personal opinion is that the default gain of 20 is the optimal way to use the chip.
The data sheet circuit looks simple, but oscillated vigorousl and persistently, and was thus unusable. After the better part of a day of tedious trial and error, I eventually have the chip completely stable and operating properly. It does what it says it does - it is an effective, self contained, wide range AGC system, making the DC receiver as pleasant to listen to ( excluding the extra sideband ) as a decent superhet with AGC. I have attached the circuit I would up with. It is very similar to the circuit in National Semiconductor AN51 which is required reading for this chip. Construction wound up to be pretty tight, as I did not anticipate the degree of extra components that would be required.
Overall end to end gain is now a completely stable 123 dB - 32 at RF, the rest at AF. I'm listening to the Friday night Collins west coast net on 3895 KHz as I write this; the receiver is on the kitchen table, I'm in the living room watching the end of the Cotton Bowl.
Happy New Year,
Win W5JAG
Starting to resemble a receiver.
I added a dc voltage controlled RF attenuator in between the MMIC RF amp and the product detector input, the same as used in the general coverage broadcast receiver, the circuit for which can be found in that thread. The PC board is in really poor condition in this area but I managed to find space and pads to shoehorn the circuit in.
I added a 3 KHz cutoff low pass filter built around another 741 op amp to the beginning of the audio strip. This makes the receiver much more pleasant to listen to, and makes tuning easier as well. It sounds pretty good actually.
I cleaned up around the audio strip, eliminating as much as possible the skywire / flying leads, and added some additonal bypassing. I trimmed the header pins on the smd to DIP IC adapters so they fit tight against the sockets.
Around the LM386, I added a 20K resistor between the unused input and ground to set the offset to correspond to the volume control pot I am using. I removed the 10 uF electrolytic cap between pins 1 and 8, and replaced it with a 10 uF non polarized ceramic capacitor placed right at pin 1, and 0 ohm jumpers to pin 8.
In all, the rework was about halfway to a total rebuild of the rig and audio strip, but worth it. It looks better and is very well behaved. It's not microphonic, it doesn't ring, it doesn't oscillate, and it as quiet as it is going to get with an LM386. The sudio strip is a little bigger than I would like, but there is still room on the board to make it a superheterodyne if I choose to do so.
Next up is audio leveling circuitry to keep the strong stations from overloading the audio amp. I have not seen any evidence that the product detector can overload, but the audio strip can be overloaded on strong signals, plus non AGC receivers are just not pleasant to listen to unless you are right in front of them, riding the gain. I don;t want anything that demands that level of my attention. I don't think it will need anywhere near as much gain control as a superhet does - 20 dB might be enough to tame the worst of the beast. Reworking the LM386 around pin 1 will make it easier to use the variable gain aspect of the chip if I decide to go that route for audio leveling. and, not coincidentally, 20 dB is the amount of gain variation Nat Semi specified in the data sheet ....
For those unfamiliar with DC receivers, I've added a general block diagram of the receiver to date, showing the configuration of the receiver.
OOPS! Obviously the AD831 requires 10 volts - it should have been drawin in the block diagram.
73,
Win W5JAG
Some slight additional progress over the weekend. To increase the overall gain, I added a Mini Circuite MAR-6+ MMIC between the bandpass filter and mixer input. This was not my first choice, the former being to install an RF amp between the two transformers of the filter, that could have variable gain control applied to it. Upon inspection of the transformers, and generally poor condition of the board in that area, I decided that I could be opening myself up to a lot of extra work, so I fell back to the MMIC. The MAR-6+ data sheet is attached earlier in this thread, as this exact part was at one point in service on the SA602 product detector board. An advantage of installing the amp in this manner is that it can also add gain to the unfiltered input port, and from any additional bandpass filters yet to be installed without further work.
I did not try to measure the gain, but at this frequency, gain should be at least 22 dB with a noise figure less than 3 dB. The suspect impedance matching between the transformer and MMIC does not seem to have affected it much, as best I can hear, bu,t as I say, I have not tried to measure it.
If my arithmetic is correct, overall gain iat present is 22 db MMIC + 10 dB AD831 + 30 dB 741 + 46 dB LM386, or 108 dB which makes for easy loudpeaker listening. The RF component of this seems about right, but I would like to spread the AF part out a bit if this particular variant moves to completion.
I added a volume control connector for convenience. There is still plenty of room in the top corner of the board for at least one additional bandpass filter and a raduino, or multiple filters if I keep the VFO separate.
The bottom half of the board purposely has enough free room to convert to a single or double conversion superheterodyne should I choose to go that route,
Win W5JAG
"Second, 56 dB audio gain is inadequate for use with a loudpeaker."
"To get additional gain quick, I went ahead and put a 10uF cap across pins 1 and 8 of the LM386 to boost gain to maximum. It appears to remain completely stable."
It is possible to significantly increase the gain of the LM386 by placing the 10 uF bypass cap in series with a resistor and connecting this combination between Pin 1 and Ground. A 3 Ohm resistor will provide 74 dB gain.
Some preliminary questions have been answered already.
FIrst, the AD831 wil perform as a product detector.
Second, 56 dB audio gain is inadequate for use with a loudpeaker.
To get additional gain quick, I went ahead and put a 10uF cap across pins 1 and 8 of the LM386 to boost gain to maximum. It appears to remain completely stable.
Using the filter bypass connector, it has heard SSB and CW as high as 20 meters.
RIght now, it is hooked up to the end fed antenna through the band pass filter, and monitoring 3987.5 KHz for the start of the Razorback SSB net. I've been hearing tune ups, and one person saying "hello, hello" with good clear audio, no trace of hum.
Here it is on the kitchen table, with it's older brother, the 20 meter SSB/CW transceiver. Not shown is the general coverage broadcast receiver.
edit: it heard every station that checked into the net from all four corners of the state (I'm in the extreme northwest corner), and every out of state station including one guy who was on the wrong sideband. Everybody kept saying he was unintellegible, but not when you can hear both sidebands at the same time .... It's not picking up any hum from the power supply or hash from the raduino, no AM or FM broadcast station breakthrough. I'm pleasantly surprised. I think it might have potential.
Win W5JAG
This is a functional audio strip.
SMD devices are used, mounted on breakout boards with header pins, and then inserted into sockets. I did this mostly because I was curious as to how it would work. Generally speaking, I think it appears to be a poor practice, and that it would be better to just solder the breakout board via short leads to the mainboard, skipping sockets, or if sockets are to be used, just do it the easy way and use a DIP device that directly fits the socket, avoiding this Rube Goldberg setup.
That said, some parts, like SA602/612 are no longer available as DIP. Any such parts found are old stock or counterfeit, so if one wants to socket a new manufacture part, this is the only way to do it.
Anyway, because the AD831 uses an oddball supply voltage ( 10 VDC ) that has to be accomodated, I decided to just run everything on 10 volts. My currently preferred small audio amp, the TDA7052A, is not that great at 10 volts, so I fell back to a part that everyone knows, and some even love, the venerable LM386. I admit to being a fan of it, although I like the TDA7052A better. The LM386 used here is undoubtedly counterfeit, being as it came in a strip of 10 parts for only $1 USD from AliExpress, but appears to work properly. The data sheet indicates that the bypass cap at pin 7 is optional. This is complete nonsense - it should always be used. In this case I used a 10 uF non polarized ceramic cap. It may or may not be in the data sheet ( can't recall ) but the chip can be muted by putting pin 7 to ground. I also always use a zobel network on the output, here 100 nF in series with a 10 ohm resistor to ground. Power is decoupled with a 10 ohm resistor and 100 uF non polarized ceramic cap. Output is through a 220 uF electrolytic.
Here, the chip is set to the default gain of 20 (26 dB).
The preamplifier is a 741, genuine from TI via Mouser. As I have never used a 741 ( or any op amp ) before, to start with I used the circuit at page 60, item B, of W1FB's Design Notebook, reproduced here*. Op amp gain is the ratio of the feedback resistor to the inverting input resistor, 820 to 1 in the example circuit, which is an insane 58 dB of gain. I went with a 33K feedback resistor which puts the gain at a more modest, yet still very high, not to mention way better behaved, 30 dB. I used 2.2 uF electrolytic caps in the audio path, but for the decoupling and bypass caps, I used non polarized ceramic caps.
Overall gain of the strip is 56 dB, with the possibility of adding another 20 dB at the LM386 if I get desperate. Parts used are mostly 0805, but some 1206 are used. The non inverting input on the 741 has to be set to one half the supply rail, so I used a formed flying lead 1/8 watt resistor for one half of the divider. The feedback resistor is set off on its own for ease of gain setting. It might have been better to use an SMD trimmer here, and I might at some future point.
Overall, it is starting to look a little like a receiver. It's pretty ugly, but electrons don't have eyes and don't care.
Win W5JAG
* fair use exception to copyright is claimed here, as the use is for commentary or teaching,
A bit of progress on the alternate version of the receiver.
This version starts with an AD831 mixer evaluation board. These boards are widely sold on eBay and AliExpress at varying price points. Shown is the single voltage evaluation board which as near as I can tell, since the board is multilayered, is a realization of the circuit shown at figure 10 in the Analog Devices data sheet.
My first experience with this device was in prototyping my 14 MHz SSB transceiver using a 23 MHz LO, and 9 MHz IF. I did not use the part in that rig for a number of reasons, but I thought it interesting enough that I wanted to come back to it at some point in the future, which, apparently, is now. The data sheet indicates the IF can go down to zero, and it can be used as a modulator, so it should work as a product detector for a direct conversion receiver.
There are substantial drawbacks. Noise figure is pretty high - 10 dB.. Voltage is critical; it operates in a narrow window of 9 to 11 VDC, or +- 4.5 to 5.5 VDC. Current draw is substantial - 100 ma (or more ). The evaluation boards take up considerable real estate and come with SMA connectors which must be used, worked around, or removed. Removing them is not for the faint of heart, and likely impossible without a hot air tool. Like the mainboard, this eval board has seen abuse - I used a hot air tool to remove the SMA connectors.. I have removed the output coupling capacitor ( probably a 10nF ceramic ) and repaced it with a 0805 zero ohm jumper. A 2.2 uF electrolytic cap on the mainboard takes recovered audio from the eval board.
A 78M10 regulator supplies power to the board. LO from the Si 5351 Raduino was measured at +8 dBm. To get it down to the suggested - 10 to - 20dBm LO level, the Raduino output is passed gthrough an 18 dB attenuator before being applied to the LO port.
The RF input port is connected to the filter by a formed flying lead 0 ohm jumper. I left the SMA connector at the filter output in place to make it more convenient to finally tune the filter, or to bypass it entirely and connect directly to the AD831 board RF input port, if desired. Because of the poor antenna at my lake home, and high NF of the mixer, it may be desirable to finish the front end filter as an RF amplifier.
Win W5JAG
I've not been happy with this project - with the exception of the switched 0-30 MHz bandpass filter board, since the VFO debacle it's really been a total trainwreck, with little more than the filter board and the completely unexceptional Si5351 frequency synthesizer to show for it. At times it showed promise, but for now I've lost confidence in it.
I'm pretty tempted to toss the SA602 product detector board in the trash for good, but for now i've just set it out of sight. Way out of sight.
Shown here is the typical double tuned bandpass filter for a single ham band, in this instance 80 / 75 meters. The toroid transformers are 35 turns on a red T37-2 core, with 3 turn secondaries to match ( sort of ) in and out of 50 ohms. The resonating capacitors are 300 pF discs, with parallel 60 pf trimmers. For now the resonagtors are coupled together with a 47 pF capacitor.
This is built on the first SMD board I used, that was part of the prototyping for my 14 MHz SSB transceiver. You can see it's had a pretty hard life, and is missing probably 5 - 7 % of it's pads from not knowing what I was doing when I started SMD, and from the ordinary removal and reinstallation of components while tinkering.
It looks like I got pretty lucky right off the bat, because without so much as touching a trimmer cap, it shows a very distinct bandpass response from about 3.4 to 4.1 MHz. It's clearly over coupled as shown by the twin peaks, but for now I'm going to leave things alone and I'll come back and tweak it later. I may decide to couple the resonators together with a transistor making it an RF amplifier. Won't know if that is necessary until later.
Shown are the sweeps with my Red Pitaya in both log and linear, from 1 to 7 MHz. I make these graphs by setting the RP to sweep a designated frequency range, then use the max hold display on the spectrum analyzer to freeze the peaks. After a few minutes, I get a crude outline of the frequency response. I don't know how a double tuned filter could be tuned correctly without some sort of a spectrum analyzer.
Win W5JAG
So here is the 1602 LCD version of the test VFO.
The sketch can be found here:
A simple VFO for the Si5351 for either LCD, LCDI2C or OLED · GitHub
and is a modified fork of the previous sketch. This sketch works with the small OLED's, 1602 LCD's, and I2C controlled 1602 LCD's simply by uncommenting the desired display.
This 1602 LCD is being driven by an I2C converter board which appears to be satisfactory for the simple two line display provided by the sketch. These 1602 LCD's are a commodity item and can be purchased with the convertor board attached, or the pieces can be bought separately. I have a large number of 1602's in stock from development of the SSB transceiver and general coverage receiver, so I just bought some I2C converter boards. These convertor boards are also a commodity item and only about $0.60 USD / piece from AliExpress, so considering the cost of the extra wire, solder, and larger PCB locking connector or pins and headers ( if desired ), and the extra time involved to directly connect a 1602 to the Arduino, I see no economic incentive not to go the convenient I2C route.
I don't care much for the green / yellow 1602 LCD's but this color 1602 has one advantage over the others - the display can be read in all lighting conditions and even without the backlight illuminated if the application calls for saving every last ma. It is possible the white background displays may also have this characteristic, and I have some in stock, but have not got around to testing them, so can't say for certain.
The 1602 LCD also appears to draw much faster than the OLED display, The frequency updating appears to me to be a little sluggish with fast tuning on the OLED, however the 1602 LCD is fast and crisp under the same conditions. For test purposes, I am for now using a cheap detented encoder. Daily driver radios get a quality Bourns encoder without detents so that they tune smoothly like a quality built real factory radio. Whether or not this perceived display lag is an issue is a matter of personal preference and intended application.
As far as current draw goes, the bench power supply I was using only resolves current to 5 ma or so. It appears the regulator / pro mini / Si5351 board / OLED configuration appears to draw between 35 to 45 ma. The regulator / pro mini / Si5351 board / 1602 LCD and I2C translator about 55 - 65 ma with the backlight lit.
73,
Win W5JAG
This is revision 2 of the Si5351 VFO which is, in all respects that matter, completely unremarkable from every other Si5351 VFO out there.
A Pro Mini Arduino is now used in place of the too large Uno. So far as I know, if a 5 volt 16 MHz device is selected, there is no difference in capability or memory capacity between the Pro Mini and the Uno that is important to this use of the device.
The Pro Mini does not have onboard programming hardware and external interface like the Uno or Nano, so I used a CH340 dongle to program it from the Arduino IDE interface. The sketch loaded into the Arduino was obtained from here:
https://gist.github.com/NT7S/24206e12342e4631e2a7
This is a very simple program, but appears to be suitable for a single band radio, either direct or single conversion.
The VFO is assembled on a cheap phenolic paper based 5 x 7 cm perfboard. The encoder has a momentary contact switch on the shaft that triggers tuning step size selection. The power supply uses an LM317T set for 5 volts output. It is overkill for this application, and is intended to have sufficient current capability to power a complete receiver.
I damaged the OLED display with the hot air gun when I was removing it from the first perf board. I thought that it would be kaput for sure, but the part of the display I am using still seems to work fine. I don’t care much for the small display and would prefer to use a 1602 LCD, so I’ll probably also use this board to experiment some with that type of display.
I fried my first Pro Mini, and the replacement had the serial data and serial clock line pins at a different physical location from the first, so I used a couple of 0 ohm jumpers to connect the new pins to the old clock and data lines. The 10 nF debounce capacitors at the encoder lines were an afterthought. I’m not sure they are even necessary, and it would probably be easier just to put them on the encoder itself if they are. Consequently, the underside of the board is a bit messier than it should be.
This VFO is not going to win any beauty contests, but should be functionally satisfactory for a simple receiver. There may or may not be a third revision, which would just be for cosmetic improvement, or unless it ibecomes necessary to accomodate the additional lines / parts for a 1602 LCD.
Win W5JAG
I've started work on a VFO.
I built a test jig using an Arduino Uno, an Adafruit Si5351 breakout board, and some sort of ridiculously small OLED display.
I know absolutely zero about this tyoe of stuff, so I'm using the Si5351 VFO on the Arduino site as a starting point.
https://projecthub.arduino.cc/CesarSound/10khz-to-225mhz-vforf-generator-with-si5351-version-2-acdc25
The sketch provided wouldn't compile for me, it was missing an end character so maybe it was just a copy error from the site.
But, anyway, it's a starting point to experiment with and it seems to work - at least it generates RF on the start up frequency. I haven't added an encoder yet. I figured I would do that after I had it making RF on a preset frequency.
Win W5JAG