Powering LNA over coax

[toomas]

LNA can be powered over coax using a bias-tee. Now when you look at the common schematics of RF LNA amplifiers and bias tees, then it seems that it's possible to just add an inductor in parallel with amp output capacitor to power it. I have little experience with RF circuits so I don't know if people actually do things like this.

Another thing is that now we have an LC tank that has a resonant frequency and Q at the amp output. If I tune this to the frequency I try to receive, it will be even better. I'm talking about ~150MHz range.

Is this idea valid? I'm thinking about modifying a cheap eBay LNA, but maybe this is a bad idea from the start.

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

[rhb]

Here are a couple of examples from the 9th ed RSGB Handbook.  As you can see the power supply over the coax is handled the same way in both circuits.

[GeoffreyF]

Inductors pass lower frequencies better than higher and that includes DC. Capacitors are the opposite. 

[StuartA]

I'm working on a very similar project. There are some wideband LNA's on ebay, 30db gain from 1-2000MHz and I'm just about to try using one of these as the amplifier for a Magnetic Loop Antenna (MLA). The idea is not entirely mine, but is described on this site https://pa0fri.home.xs4all.nl/Ant/Active%20antenna/Active%20receiving%20%20loop%20antenna%20eng.htm  and there some notes on power the LNA on there.

The MLA is a great way of avoiding EM noise. I recently put up a dipole for 5MHz but the EM background noise is S8-9. I've had a previous MLA, which worked very well on MW and LW, but didn't go to 5MHz; hopefully this new arrangement will do that.

Stuart

[Ian.M]

If the coax isn't long enough for DC drop due to the LNA supply current to be significant, I don't see any need to over-complicate it by splitting the power off to locally regulate it.  In the O.P's circuit, simply move the 50R resistor, 1uH choke, 5V feed and 1nF coupling cap to the receiver end of the coax.   

[rhb]

That was simply the first circuit I found in my library.  I looked in EMRFD first, but it didn't have anything, so I grabbed the RSGB and ARRL handbooks.

Choice of choke and coupling cap depends on the frequency of application. I'd be a bit afraid of parasitic oscillation with a 1 uH choke.  In any case, one would need to make check that.

[cdev]

Stuart A. Check out Chris Trask's amplified varactor loop design. You likely could use a different amplifier such as a modern MMIC (or even no amplifier as a resonant magnetic loop doesn't really need one, a resonant loop captures a lot of signal voltage wise.)

You can use many of the VHF-UHF MMICs for HF, they usually work fine. The values they cite are for the specified frequency ranges so you need to change the values used appropriately. VHF-UHF MMICs may have a lot of gain at lower frequencies, so much that they oscillate.

In order for them to be unconditionally stable you may need to make some other changes.

Mini-circuits has an application note on this for the PGA-103+ its application note AN-60-064 to use their example, it depends on the nature of the input, you may be fine with a relatively well matched antenna.

[NiHaoMike]

For that particular LNA, the output capacitor is basically a short at RF so short that to allow the DC to pass. Then build an adapter with a 1nF capacitor between the input and output, then add a 50R resistor and 1uH inductor to the power supply in order to replicate what's in the original circuit. Do keep in mind proper RF construction especially at the high end of the frequency range.

[cdev]

If you power an LNA over coax you may see a performance improvement contingent on your being able to supply the LNA with power without too much insertion loss at the power injection point. Thats easy for a single frequency but gets difficult when you want it to work over a wide frequency range. You may be able to afford the loss, often it doesnt matter at all, in practical terms unless there are big holes, but you really don't want it to resonate which will cause unpredictable behavior and sometimes its intermittent and desensitizes your receiver - and problems are hard to identify.

The most difficult thing to get right is to have an inductor that doesn't resonate strongly at some frequency because of its inter-turn capacitance causes self-resonance. This is why you need multiple stages and the components properties are not simply based on the inductance or capacitance or resistance values. A lot more variability comes into play. Its actually quite an interesting problem because it can drastically improve or hurt the performance of an RF LNA, and the performance changes quite a bit with small changes.

As far as commercial products I've had the best luck with a specific kind of inductor from coilcraft.. they are a lightish purple.  Ive tried to make my own inductors too. Its hard to get this right, The first inductor out of several - they should zig zag, should be air core and ideally very thin wire that also holds its shape and solders well. You can use an inductor thats is wound around a conical object like the tip of a pencil. The tip should be the part on the signal path and block the highest portion of the desired range, That should be short and in the nanofarad range, typically, the first bypass capacitor should be relatively low value relative to the later ones. Its a bypass cap...The reactances should be chosen so that they cancel each other out, for each stage, using materials to add loss. Also some bias tees are basically transformers that utilize the common mode cancellation principle. That is likely the best approach but it isnt easy to design without more resources than I have. So ive used chains of two or three L+R-Cs. The idea is to let DC through while filtering out Ac. Otherwise the LNA may resonate or swallow desired RF or more likely both. An ugly hack in that situation is to add lossyness in the form of resistance without causing problems. One way to do that in a fairly non-invasive manner is with antistatic foam in a small insulated plastic bag and pressure. Also, it is particularly important to have the PCB of an LNA especially be totally clean of flux.  NOT covered with flux residue like some I have seen.

[TheUnnamedNewbie]

Another thing is that now we have an LC tank that has a resonant frequency and Q at the amp output. If I tune this to the frequency I try to receive, it will be even better. I'm talking about ~150MHz range.

This is what is done on many on-chip RF amplifier\$^1\$. The LC components that are your bias-T become your matching network, and are chosen to be ideal for their frequency.

As cdev has said, the important thing to consider is how 'inductor-like' your inductor actually is at your frequency. If you have issues with inductors, you can always use a transmission line to transform a capacitor into an inductor. That way you can perhaps get better Q and lower self-resonance, provided your capacitor has good performance (which really shouldn't be an issue at a few hundred MHz). In more complex bias-Ts that go up to high frequencies, this is sometimes used (they use multiple inductors)

You can of course also just make an inductor or capacitor with 'only' transmission lines, but that might become harder to use as a bias-T.

Note that I don't know how much use my advice is at 150 MHz, I am not that familiar with working at such low frequencies.


\$^1\$ For those interested - mostly used up to maybe 10-20 GHz. Past that point, transformer-based architectures tend to dominate. (transformers tend to have better quality factors and thus higher performance. Capactitors suck past a few GHz on chip, inductors become great). Transmission lines are also used, but in my experience sometimes are considered the 'easy' way out since they use a lot of chip area compared to an lumped-element approach, but the latter needs a lot more work in terms of EM simulation and design effort (coupling of inductors can be a real hassle, and unlike a transmission line has a lot more parameters to play around with - BW vs Q vs area vs robustness etc).
Distrubuted amps also do not use this, as a matching network by nature limits your bandwidth to the frequency it matches at.

[cdev]

Input side networks can also really improve the performance of an LNA at a specific frequency by filtering out other stuff outside of the band of interest. The 2 meter ham band is 144-148 MHz and the 137 MHz satellite band are both close enough to 150 MHz that the literature on band pass filtration at those frequencies definitely applies. You could build a helical resonator for example.


Also, you shouldn't discount the value of dead bug style construction techniques at 150 Mhz.

For example, Leif Asbrink's (SM5BSZ) old school 2 meter LNA using a Minicircuits PSA4-5043 MMIC is classic point to point wiring and not neat looking compared to SMT, but I learned a very great deal from looking at it. The biggest thing was 'just do it'.

The core chip it uses, (costing under $3) was able to compete against much more complicated LNAs in a yearly meetup of EME (moonbounce) enthusiasts to test LNAs in Europe a few years ago. And that was without an input side network to provide selectivity (apart from his choice of capacitor) In terms of cost-benefit MMICs bring a lot to the table.

It has a noise figure of under 0.6 db at this frequency.

Note his use of the copper shielding grounding the upside down chip's legs and the inductors in parallel with resistors for the power. At 144 Mhz the placement is non-critical, nothing like how it would be higher up. The same applies to 150 MHz.

You could move the power part of the LNA into a bias tee. See how non-critical the design for a single frequency is?

[StuartA]

Stuart A. Check out Chris Trask's amplified varactor loop design. You likely could use a different amplifier such as a modern MMIC (or even no amplifier as a resonant magnetic loop doesn't really need one, a resonant loop captures a lot of signal voltage issue

Hi; Thanks for that info. I did find Trask's paper and it is on file for future reference. A principal reason for wanting to try the LNA was simply that I bought a few of the exact same types 'to play with' some months ago, and seeing that design, couldn't resist it. It is all assembled now and will be on test tomorrow  .

Stuart

[Richard Crowley]

If you power an LNA over coax you may see a performance improvement contingent on your being able to supply the LNA with power without too much insertion loss at the power injection point. Thats easy for a single frequency but gets difficult when you want it to work over a wide frequency range.

I would like to investigate how to send camera power (~12V @ 1A) through coaxial cable carrying HD-SDI digital video. Is the uniform bit-rate sufficient to make this a simple case, or does HD-SDI make a more complex problem.

[rhb]

PoE works with arbitrary signals, so it is possible.  How difficult I don't know.

[cdev]

Is it digital?

You should just try it to see what happens.

(Analog) Video is a different case I'd imagine (never having done it though) because its a broadband signal that - am I correct starts just above DC and goes up from there to- what? My video distribution amp (that I only use for 10 MHz) is supposed to be flat to 200 MHz ! It varies right? depending on the bandwidth (resolution) of the signal.

Your hypothetical bias tee inductor has to block most of the video, so it likely has to be fairly high value, and also carry an amp of current. So, a choke, likely is what you'll want. The DC block capacitor has to let essentially the entire video signal through but- doh- block the DC. maybe a mylar cap in the uf range? Then you might want to experiment with your bypass caps. You might not need much bypassing there. It might be much easier to do than a broadband bias tee but we still should remember that video is still a broadband signal, in terms of octaves that need to get through, they are just lower! I could see some combinations of inductors and caps causing weird problems, perhaps?

[MarkF]

If you power an LNA over coax you may see a performance improvement contingent on your being able to supply the LNA with power without too much insertion loss at the power injection point. Thats easy for a single frequency but gets difficult when you want it to work over a wide frequency range.

I would like to investigate how to send camera power (~12V @ 1A) through coaxial cable carrying HD-SDI digital video. Is the uniform bit-rate sufficient to make this a simple case, or does HD-SDI make a more complex problem.

This article from TI for automotive cameras may help.
   http://www.ti.com/lit/an/snla224/snla224.pdf

[toomas]

Thank you for all your replies. The LNA arrived today so I can start going through links and suggestions you gave and start experimenting.

150MHz was just a ballpark value, what I'm building the system initially for is for NOAA weather sattelite reception at 137MHz, as user cdev guessed. Tuned dipole antenna, FM bandstop filter, LNA, (hopefully less than) 10m RG316 coax and RTL-SDR v3. Some time later on I'll probably swap in a wideband antenna (discone?) and experiment with other frequencies.

Design of this LNA is as simple as it gets. It's using a 3M9009 chip. 220pF caps on input and output. Inductor of unknown value is wired directly to 5V power rail (in series with multiple 0 ohm resistors for whatever reason). It reportedly works between 10MHz and 4GHz, but I'm a bit sceptical about the lower end. Chip datasheet says it's for 50+ MHz, but I'll try to measure that.

Chip datasheet: https://www.mouser.com/ds/2/412/TQP3M9009-1209930.pdf

[cdev]

That's a nice case they gave you there, how much did that LNA cost?

You likely could replace the parts in your LNA with others if so inclined.

The NF of the chip seems high as does the gain.

Why so much gain? The MMIC I usually use (DS below) has a bit less gain (but in fact actually has almost that much at that frequency) but a significantly lower NF.   

Gain doesn't matter so much once it gets high enough to establish a low noise figure at the front end. The first gain stage is the one that matters.

I really wish that I had a machine shop capable of turning out machined slabs of aluminum like that.  Note the large expanse of flat surface and lots of screws so that the closure of the lid is effective RF shielding. 

Some issues, though. I would perhaps have laid it out a bit differently but that would be very nitpicky and probably matter little compared to the case.

You likely could remove the MMIC device they gave you and replace it with another with exactly the same footprint and a significantly better NF (I would definitely do that if you find its not quiet enough)

You might lose - my guess, around 3-4 db of gain but from my own experience that would be totally unimportant in your context.   

Also, if your current device has some protection against ESD, I dont think the replacement I have in mind has built in ESD protection. The one I usually use which has a different footprint/case style does.

Noise figures of both are quite comparable.

These can both be found in individual quantities for very little. Just a few dollars.

Mini-Circuits makes good stuff.

You might want to consider adding a second dipole to your setup perpendicular to your current one and adding a 1/4 wavelength length of higher impedance coax to impart a 90 degree phase shift. (a 'turnstile' antenna)  For right hand circular polarization. Google for that. Easy to make, much better than a dipole for NOAA reception. Or use a tall quadrifilar helix or QFH.

Unfortunately, where I live its never been quiet enough to receive noise free APT images, at least I've never been able to get images decent enough to be happy about. I have strong pager signals nearby which bleed into the satellite band. RTLSDRs don't have enough out of band rejection for my QTH.

You may find you want to put some kind of narrow band pass filter on your antenna input to exclude other nearby frequencies. 

You need a good one for APT.

Or if only one transmitter is problematic, a notch filter.

[Wolfgang]

LNA can be powered over coax using a bias-tee. Now when you look at the common schematics of RF LNA amplifiers and bias tees, then it seems that it's possible to just add an inductor in parallel with amp output capacitor to power it. I have little experience with RF circuits so I don't know if people actually do things like this.

Another thing is that now we have an LC tank that has a resonant frequency and Q at the amp output. If I tune this to the frequency I try to receive, it will be even better. I'm talking about ~150MHz range.

Is this idea valid? I'm thinking about modifying a cheap eBay LNA, but maybe this is a bad idea from the start.

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

I would not mix LNA tuning and bias for stability reasons. Why dont you work with bias tees ? You can buy them from Mini Circuits or even make them yourself.

[cdev]

If one is on a budget, and needs a broadband bias-tee, the best approach is likely building one from a purpose-built SMT part appropriate to your application.

You could have the PCB made for you by a PCB house..

Then you can put the LNA right at the antenna.

[toomas]

I like Ian.M idea of moving some things to receiver side. For the actual hardware I have right now, I only need to remove one inductor and short the output cap in LNA. Separating AC and DC before LNA just to combine them together trivially inside LNA box seems unneccesary.

Attached a picture that explains it. Top part shows LNA schematic exactly as it is now, then there is a bias tee (which may be more complicated than that, I haven't received it yet) and then there is RTL-SDR input circuit as it is (got the LDO wrong, it's not 3.3V, but 4.5V).

Bottom part is what I'll try. Gut feeling is that there shouldn't be any issues because it's all 50ohm. Less unneccesary parts on signal path should mean better signal too, shouldn't it?

That's a nice case they gave you there, how much did that LNA cost?

LNA was from eBay, 12€ including shipping to EU. Thanks for sharing your experience with trying to receive images from a sattelite. I'll go for simplest setup first and then see what I can do if it doesn't work. I'm afraid in the end I do need a better antenna and a tight band pass filter. We'll see.

I would not mix LNA tuning and bias for stability reasons. Why dont you work with bias tees? ...

After posting this thread, I realized that the parallel LC tank on LNA out would act as a band stop, not band pass element. Nothing to tune there except low frequency cutoff point. I can use bias tees, but do I have to use 3 if I really need 1, attached picture explains.

[cdev]

Less is more.

Since your RTL dongle has a built in bias tee, use that. You could make your own LNA very easily with a piece of RF trace and an x-acto knife. For satellite reception you really want a low NF. A NF over 1db may work fine, but its not optimal. Since you can get a device thats better for very little and its easy, and your dongle already has a bias tee, why not try doing that?

Find an discarded RF PCB with some 50 ohm trace on it - preferably one with two SMAs on it but thats not really necessary, you can solder some coax to it. Get a MMIC like either of the aforementioned ones, (I would try to find a PSA4-5043+) and a decent quality ceramic capacitor of an appropriate value for your DC block.

200 pf is fine, you can use one up to 1000 pf if you want or as little as 100 pf. I would use a 200 pf one to start.

Cut the trace twice, with one of the cuts being around (like right on top of) vias on both sides. (the GROUNDING on both sides being short is the most important thing to get right here) and put the capacitor across the first one and the MMIC input to the capacitor side of the trace and the output/power in on the other side. Don't just ground the appropriate places to the ground plane well, be obsessive about getting it close and short. Be careful not to heat it up for too long. be gentle with the legs especially when its hot, you don't want the chip to melt and a leg to come off. Also do your work over a tray under bright lighting so you don't accidentally lose your chip. Its easy to do.

There are lots of MMICs that could work here, however when I decided to build a LNA I did a parametric search of all of the ones I could find and decided on that one for a number of reasons. It still seems like the best all around choice for something like this to me. Depending on how you mount it, it may make more sense to dead bug it. That means mount a chip upside down with its legs in the air. I would then use a small piece of copper to connect the two ground legs. Flat copper is better than wire because less inductance to ground.

Whatever connection method you use try to keep it as short and consistent as possible. Chances are in your junk box you have some old RF something that has an appropriate trace you can chop off. Old wireless cards, for example can often yield several useful small chunks of RF trace bearing PCB with vias and everything all ready to go.

Solder the antenna end coax to the end of the trace with the DC block cap and the trace leaving the other side of the chip to the coax and the braid going to the receiver and its bias tee to the other "output" end. You're now done. Don't forget to test it with your ohmmeter before you apply power. The datasheets will give you a good idea of what you should expect as far as current draw. If the current draw is higher, or lower, remove power and check for a short or open.

Then see how well it works right at your antenna with you sending your power to it from inside. It it works well, figure out a way to weatherproof it.

The LNA you got may be great or it may be crap, but either way, the nice machined box it came in, with voltage regulator, etc, appears to be nice enough to be worth what you spent.

Also, if you wanted to replace the MMIC in it it looks like it would likely be easy.

Others here probably could evaluate it's strengths and weaknesses better than I.

How does it work so far for you with an external supply? Make sure you supply the right voltage.

[radioactive]

Not sure about the availability or pricing of these parts (I've never used them), but might be of interest...

https://www.atceramics.com/UserFiles/506WLSN6R00KT236T.pdf
https://www.coilcraft.com/bcr.cfm   (of particular interest is the '232' part).

[toomas]

How does it work so far for you with an external supply? ...

It did work, but the power-in port is not properly filtered inside. Low frequency noise (maybe <100kHz) that is picked up by wire going into power port is very apparent in SDR#. Adding ceramic disc capacitor close to LNA reduced that noise, but it shouldn't be so. Now I know where there are LDOs in LNAs (this one doesn't).

Good news is that modifying LNA to accept power over coax works. The only thing I needed to do was to short the output capacitor. I left the inductor in and put a red led in series with 2.2k that shows it's receiving power. Looks nice. It does provide gain, seems to be over +20dB. No more horrible noise issues as before.

Bad news is that it seems to be a bit too noisy for my taste.
* RTL-SDR dongle max gain +49dB gives noise floor -50dBFS (SRD# shows this)
* LNA followed by RTL-SDR dongle with gain +32dB gives the same noise floor

I ordered PGA-103+ as it was one that was suggested and has the same footprint as 3M9009. I'll replace it in the box.

One issue that I see coming is that the coax to antenna itself picks up FM a lot. I have this chain: RTL-SDR SMA connector <-> 3m of RG174 <-> SMA to BNC adapter <-> 50 ohm terminator. With SDR gain turned down to +0dB I still get clear FM station reception (SNR 28dB). My hope is that in a complete system LNA right after FM bandstop near antenna won't amplify FM stations. Then even though the coax picks up FM, it's not overloading my sdr usb dongle.

Not sure about the availability or pricing of these parts (I've never used them), but might be of interest...

Didn't know such wideband inductors exists, it looks cool.

[cdev]

You should put some kind of ESD protection on your LNA if you use the PGA-103. It can be something very simple, there is a wide range of opinion on what is best to do.

All the noise possibilities were why I was thinking just the simplest of all possible LNAs would work best.