2m band transverter build

[rheb1026]

Quick summary to get started:

The goal of my project is to create a 28 MHz to 144 MHz transverter for use on the amateur 2m band. The transverter should include a modest power amplifier with 5 – 10W output. The transverter should have a good, stable local oscillator, with an option for an external reference. It should also include one input for a PTT signal from the driving radio.

Ideally keep component and PCB cost under $200. All components should be available for individuals to purchase from distributors and manufacturers (ie: no junk box transistors that can’t be found anymore or crystals pulled from a television)

-------------------

I'd like to start by getting some input on the mixer section, as this is the area that I'm the most unsure of. Attached is a schematic of the mixer section. I must give a quick thanks to G4DDK for the bandpass filter design, which I pulled from his website.

I also simulated in LTSpice the bandpass filter and the diplexer. I'll attach photos of those as well. The FFT plots were done by simulating a mixer with an RF port input of 28 MHz, an LO port input of 116 MHz, and plotting the output of the IF port. The other FFT plot shows the effect that the BPF had on the same output. Appears to be pretty decent to me, but I'm certainly not an expert

Any comments and suggestions are welcome. Thanks

[rheb1026]

Got PCBs made and started to populate them this weekend. Went whole hog and put everything on one board and hope for the best instead of making up boards for each section.

I got a chance to populate all the power supply section. LM1084-5 for the +5V, ADM7150 for the +3V3. All working well. Unfortunately I goofed on the exposed pad under the AD regulator and it's not connected to the ground plane. Still seems to work okay, luckily, and doesn't get noticeably hot to touch.

I also populated the MCU (STM32F4) and PLL (AD9553) section. To my surprise everything worked great and I can see a 116 MHz signal coming out of the PLL. Unfortunately in my rush to get the board out and made I forgot to provide a trace between the PLL Lock indicator pin and the MCU. I also managed to swap the SPI Data and CS line... so had to bit bang the SPI instead of using the MCU hardware. Thankfully this only needs to be done once upon power up.

And yes, an STM32F4 is way overkill, but I had a couple on hand from a past project. An 8-bitter would be plenty capable and is what I'll likely use in the end.

The XO for the PLL is a 10 MHz Connor Winfield M100F, 100 ppb temp co, so there shouldn't be any problem with drifting that I've read so often about with other transverters.

Next is to populate the mixer section, which I'll hopefully have completed this week.

Any questions or comments welcome

[phenol]

no front-end amplifier?

[rheb1026]

no front-end amplifier?

There is a PGA-103 in both the transmit and receive path. I'll post up the schematics when I get a chance

[rheb1026]

I've attached schematics of the board. Sorry for the slightly unorganized nature of them.

I did a little bit of testing today with the receive portion of the board. It does receive, but is quite deaf. Unfortunately I don't have a spectrum analyzer to see exact signal levels. I'll be purchasing one soon so I can better see what's going on.

I also haven't properly tuned the inductors... just tweaked them to get the highest reading on the S-meter with a few minutes of fiddling.

To test, I injected signals from an HP 8656B signal generator and read the reading off of the Elecraft KX2 S-Meter:

-40 dBm  ---->    S8
-50 dBm  ---->    S6
-60 dBm  ---->    S4
-70 dBm  ---->    S2
-80 dBm  ---->    S0/S1

I also tried listening in to a local repeater, which gives about an S9 on a Kenwood HT. The transverter board and KX2 combination read about an S1.

It looks like I'll need at least another gain stage after the IF diplexer.

Any other suggestions would be very welcome!

[PA0PBZ]

Are you sure the AD9553 can drive the mixer with +7dBm? If you don't feed these kind of mixers with enough LO the conversion loss will be very high.

[rheb1026]

I'm fairly certain, but my scope bandwidth isn't quite high enough to be 100% sure.

+7 dBm is ~1.4V P-P into 50 ohms, correct?

My 100 MHz scope shows the AD9553 delivering 2.25V into 50 ohms, so I think the drive level should be sufficient.

I should also mention that I had purchased the wrong package size 2.2uH inductor for the diplexer and had to bodge an 0402 into a 1008 footprint. This may or may not be part of the problem, but the right part has already been ordered and should be here in a few days

[PA0PBZ]

+7 dBm is ~1.4V P-P into 50 ohms, correct?

Yes, correct. Did you use a 50 ohm resistor as termination?

[rheb1026]

I did, I used one of those BNC T's with the terminator attached. The scope has a 50 ohm internal termination as well, but I'm always worried I'll have a brain fart and blow it up, so I don't usually use it

[phenol]

what does R603 do?

[rheb1026]

I had put it in to limit current just in case. Currently the board just has a 0 ohm jumper in its place

[phenol]

where does the output of the rx amp go? to the RF port maybe?
I think you have to inspect the T/R switch first for correct signal propagation, then the PGA amp and retune the bandpass filter. You could get close enough by peaking it using the s meter, but i’d use a vna or tracking generator to visualize the filter response. I’d start with the frontend prior to inserting more gain stages after the mixer. Unless there’s something utterly wrong with your mixer diplexer , i wouldn’t expect dramatic improvement in rx sensitivity even with a 20db post mixer gain block if your frontend before the mixer is mistuned

when i don’t have a sinad meter, i’d put the radio in FM mode and listen to the amount in quieting i get as i tune the frontend at progressively lower input level. You could compare the result against another FM vhf radio.
It wasn’t uncommon to hit 0.1-0.12uV for 12db SINAD Using this method. However, this does not automatically guarantee that the bandpass filter is correctly tuned for symmetrical response, etc. it mostly means that signal loss BEFORE the RF amp is low enough and the gain of the amp is high enough to get the S/N that you get at a given test frequency.

[rheb1026]

where does the output of the rx amp go? to the RF port maybe?
I think you have to inspect the T/R switch first for correct signal propagation, then the PGA amp and retune the bandpass filter. You could get close enough by peaking it using the s meter, but i’d use a vna or tracking generator to visualize the filter response. I’d start with the frontend prior to inserting more gain stages after the mixer. Unless there’s something utterly wrong with your mixer diplexer , i wouldn’t expect dramatic improvement in rx sensitivity even with a 20db post mixer gain block if your frontend before the mixer is mistuned

when i don’t have a sinad meter, i’d put the radio in FM mode and listen to the amount in quieting i get as i tune the frontend at progressively lower input level. You could compare the result against another FM vhf radio.
It wasn’t uncommon to hit 0.1-0.12uV for 12db SINAD Using this method. However, this does not automatically guarantee that the bandpass filter is correctly tuned for symmetrical response, etc. it mostly means that signal loss BEFORE the RF amp is low enough and the gain of the amp is high enough to get the S/N that you get at a given test frequency.

The output of the RX amp goes to the PE4251 RF switch which then goes through the bandpass filter and into the mixer RF port. Sorry for the confusion. I was originally intending to make up boards of the different sections and testing them on their own, so they were all imported into their own sheet instead of being put onto one sheet.

Thanks for the detailed list, I'll give those a try tonight. I forgot I had one of those RTL-SDR dongles, so maybe I'll be able to use that as a better way to see the filter attenuation than using the S meter.

Actually before I read your reply late last night I attached a 2N5109 ~20dB gain amplifier after the mixer and before the radio. It certainly seemed to get the signal levels much more in line with where they should be. A -73 dBm signal from the signal generator now read an S8 on the meter. The noise floor on the S meter was also now a consistent S1

[phenol]

is that a norton feedback amplifier?

you can amplify the IF signal all you want and the s meter will deflect further and further, but at the end of the day it is the signal to noise ratio that matters.

edit: upon closer inspection of your pic, that isn’t a norton feedback amp. I’d connect a 100n ceramic cap from the ‘cold’ end of the transformer (the red wire pad) to ground.

[rheb1026]

You're correct it's not a Norton feedback amp. Just a common emitter with a transformer output.

Right, I do understand that. I guess part of my problem is that I don't understand just how much gain is necessary for this type of application. I looked at a few different transverter designs and saw that most had about 20-25 dB of gain on the front end.

I'll see what I can do for tuning the bandpass filter, since I think this is probably where my issues lie. I may have to make up a noise source of some kind and see what I can do with the SDR dongle.

I'll give that a try with the amp, thanks

[rheb1026]

Small update:

I increased the LO drive level (~3dB) using the above amplifier. The S-meter would now read an S4 with my test signal (144 MHz @ -73 dBm). My baseline reading is an S2.

I replaced my bodged 2.2uH 0402 inductor in the diplexer with the correct 1008 footprint one. No difference in performance.

I removed the PIN diodes in the antenna T/R switch. No difference in performance.

I removed the 1/4 wave lumped element TL and jumpered the inductor footprint with a 0 ohm resistor. No difference in performance.

I can safely rule out the T/R switching at the antenna end as being the problem.

I attempted to use the RTL-SDR dongle to tune the bandpass filter last night, but I couldn't tell any noticeable difference as long as the inductors were somewhat close to being centered in the range. The inductors have a pretty small tuning range to begin with (133-157 nH). But, a co-worker is going to loan me a spectrum analyzer next week. So hopefully this will give me some further insight and allow me to see where the problem lies

[rheb1026]

Big update: Got it working! Turns out that the mixer was the issue. The RF port was not connected to the transformer inside. Soldered on another mixer and I finally got the S9 signal!

And yes, my signal gen is off frequency

[phenol]

numbers still don’t add up... Let’s assume your HF radio has a reasonably accurate representation of a S9 signal (-73dbm for HF and -93dbm for VHF+) on the S meter. Let’s also assume that the mechanical attenuator and leveling systems of your sig gen are not knackered. Then you have some 22db of frontend gain, ~5db of filter and T/R switch loss (could be better or worse), another 7db of mixer+diplexer loss. This leaves 10db of conversion gain from 145 to 28MHz. Maybe the granularity of the digital s meter doesn’t let it register 10db steps above s9, so even if the signal is there, you can’t really tell.
Do you still keep the post-mixer amp? if it’s there, you should definitely be seeing more than S9 at -73dbm.
Again, put the radio in FM mode, lower the gen level to -125dbm and listen if you can hear the hiss go down. Then compare to a good vhf radio.
VHF and above is all about weak signal reception. Unless you live in a very RF noisy neighborhood, you need to have a receiver that hears well.

btw, are those 50-ohm traces really? They seem rather too narrow...

[rheb1026]

Thanks for the information. I didn’t realize that S meters were calibrated differently above 30 MHz.

I think the S meter is capped at an S9 in CW mode for some reason. I did try switching to USB briefly after and it was then able to display up to S9+30 if I’m remembering correctly. I don’t remember what kind of signal strength readings I got in that mode. There was no post mixer amp while taking this reading.

I’ll try doing it in FM mode and seeing what I get. I may try retuning the band pass filter again now that I can receive the test signal all the way through. The only VHF radios I have are Kenwood TM-V71A and TH-F6A. Would these be considered good enough?

They should be 50 ohms, I cant remember their width off the top of my head but I looked up the pcb manufacturers stack up and plugged it into a calculator. It is a 4 layer board.

[phenol]

The kenwoods are good enough as benchmarks.

Start at 1uV or so. If it’s quiet in FM mode, go down until substantial hiss is audible and tune the filter slugs for max quieting. Decrease sig gen level and repeat. Your ears are your sinad meter...

[rheb1026]

Thanks for all the help. It seems the receive is probably as good as I will get it.

I could audibly hear FM down to -122 dBm. Below that it was just noise.
CW I could hear very clearly down to -125 dBm, but the signal generator doesn’t go any lower.

Next is to test out the transmit side and all the T/R switching

[Co6aka]

It seems the receive is probably as good as I will get it.

mmm... no...

I could audibly hear FM down to -122 dBm. Below that it was just noise.
CW I could hear very clearly down to -125 dBm.

Data sheet says NF at 144MHz should be about 0.5dB so I would expect a little better. Data sheet also says input return loss at 50MHz is 6.7dB, and at 400MHz is 11.3dB... MMICs usually aren't noise-matched, and that's not so good for a power match... Typically as I've seen them, the input impedance rises as frequency goes down, so you could try redesigning your input L/C network to transform to a higher impedance. Research noise matching and you'll see what I mean. Maybe others could post some good links from their bookmarks. If you have some RF transformers spec'd for VHF you could try 2:1, 3:1, 4:1, and so on and see what your MDS becomes with those ratios. Don't go by maximum output power versus input power, go by MDS. I'm assuming you want to cover the entire 144-148 band... Otherwise you could use a double-tapped L/C tank.

Alternatively, ditch the MMIC and roll a transistor amp.

[phenol]

agreed, you have to be able to squeeze more from it. Make sure that the MMIC doesn't oscillate for some reason. It could do so at GHz frequencies and still provide gain down at 144.

There are quite a few preamp projects with PGA103 out there and they all seem to use the amp in a ''50-ohm'' environment w/o any particular reactive matching.
On the other hand, most commercial radios would use discrete transistors-- dual gate mosfets and on rare occasions phempt on 70cm and above. Granted, PGA103 has superior OIP3 to probably any dual gate mosfet but it sucks a lot more juice from the power source to achieve that.

[rheb1026]

It seems the receive is probably as good as I will get it.

mmm... no...

I could audibly hear FM down to -122 dBm. Below that it was just noise.
CW I could hear very clearly down to -125 dBm.

Data sheet says NF at 144MHz should be about 0.5dB so I would expect a little better. Data sheet also says input return loss at 50MHz is 6.7dB, and at 400MHz is 11.3dB... MMICs usually aren't noise-matched, and that's not so good for a power match... Typically as I've seen them, the input impedance rises as frequency goes down, so you could try redesigning your input L/C network to transform to a higher impedance. Research noise matching and you'll see what I mean. Maybe others could post some good links from their bookmarks. If you have some RF transformers spec'd for VHF you could try 2:1, 3:1, 4:1, and so on and see what your MDS becomes with those ratios. Don't go by maximum output power versus input power, go by MDS. I'm assuming you want to cover the entire 144-148 band... Otherwise you could use a double-tapped L/C tank.

Alternatively, ditch the MMIC and roll a transistor amp.

You're right, it's not as good as it can get, but I think it is as far as tuning what's on the board now.

I'll take a look into noise matching. Looks like I have plenty more reading to do. I'll probably make up a few boards to do some experimenting with the input amp and filtering by themselves.

My main goal was to just use 144-146 for SSB/CW, but if I could get the upper part of 2m as well, even at reduced sensitivity, then that would be a bonus.

I'd like to stick with the MMIC amp just because they're so simple to use and power consumption isn't really a huge concern

agreed, you have to be able to squeeze more from it. Make sure that the MMIC doesn't oscillate for some reason. It could do so at GHz frequencies and still provide gain down at 144.

There are quite a few preamp projects with PGA103 out there and they all seem to use the amp in a ''50-ohm'' environment w/o any particular reactive matching.
On the other hand, most commercial radios would use discrete transistors-- dual gate mosfets and on rare occasions phempt on 70cm and above. Granted, PGA103 has superior OIP3 to probably any dual gate mosfet but it sucks a lot more juice from the power source to achieve that.

Is there an easy way to check for the MMIC oscillating? Can I just look on a spectrum analyzer for any spurs up high in frequency?

I may need to get a metal enclosure  for the board as well to do some better measurements. I live pretty close to some FM broadcast towers and can see them really well when using the home brew noise source just sitting out on my desk

[phenol]

You can use a spectrum analyzer and a near field probe.