Intermod with broadband sources

wa6jbd

Contributing Member
Joined
Oct 28, 2019
Messages
50
I’d like to start an on-going discussion about something that I’ve seen nothing on in the various forums and mail lists. It’s got to be a problem for others besides us. I’m referring to intermod caused by broadband signals such as cellular and DTV that manifests as desense.

A few examples… A site with a full performance UHF DTV transmitter with 8VSB modulation, which to an analog receiver is a 6 MHz wide wall of noise. Virtually every UHF repeater on the site with a 5 MHz TR spacing gets desense whenever it’s own transmitter is keyed, but in a test with the DTV transmitter turned off, no such desense occurs. Duplex links with spacing wider than 6 MHz don’t get the desense at all.

Another example is 700/800 MHz P25 sites that are experiencing desense with or without their own transmitters being keyed up. Universally, where this has been seen, is sites with nearby cell sites operating 600 MHz (T-Mobil) and 700 MHz (Verizon?) LTE with carriers as wide as 20 MHz, similar to the ‘wall of noise’ DTV signals mentioned above.

The 600 and 700 MHz cellular works out as a 3rd order mix to hit the 806-815 MHz input band on the P25 sites. Additional filtering on the receivers reduces the desense only by the amount of the insertion loss of the filter, so it’s an external environment thing that can’t be filtered out.

I'm seeing other examples, but these should give you an idea what I'm talking about. Back in the good old days, an intermod mix required an almost direct hit on your receiver before you’d notice it, then you could listen and identify what it was. Now, it’s intermod manifesting as desense that’s absolutely intractable and there seems to be little that can be done.

Has anyone else observed this sort of thing? I’d be interested in hearing how widespread it is, and what, if anything, people have done to reduce it.

Mel - WA6JBD
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
I've witnessed, worked with and tried to mitigate IMD in many systems from repeater front ends to wide band preamps on wide band antennas to multi carrier satellite distribution, etc. Not sure if there is enough room here to cover everything. There are ways to measure levels into preamps or receiver front ends to estimate if IMD will be generated and to what extent. With digital TV transmitters very close by the resulting IMD would usually show up as an elevated noise floor instead of discreet IMD products from narrow band sources, although enough narrow interferers will usually produce what looks like an elevated noise floor. I think IMD is rampant everywhere in radio systems unless you have a single repeater way out in the sticks with nothing else around for 50 miles.

I've also designed and built test equipment that generated upwards of 32 carriers for testing and evaluating satellite receiver components and could see IMD being generated within the test equipment. I had to eventually upgrade the equipment by breaking up the output spectrum into three different bands handled by three separate amplifiers of higher IP1 and IP3 ratings then combine them back together to reduce IMD. That test equipment was nice to view IMD and the point its produced and to watch the non linear attributes of IMD as you increase RF levels into amplifiers and other components. Once IMD starts, for every 1dB increase in RF levels into the component generating IMD, the IMD increased by 2dB and so on.
 

RFI-EMI-GUY

Prolific Contributor
CS Forums $upporter
Joined
Mar 8, 2014
Messages
1,086
It is a real issue. Many of the popular IM calculator programs ignore the bandwidth of the transmitters and receivers. They report only on channel hits. Simple math 2a-b=c without the introduction of bandwidth of 2a, b and C.

The bandwidth of transmitters effectively doubles and triples (etc) with each IM order. So it is even a problem with LMR equipment. I experienced it once with 5th harmonic energy from an FM broadcaster affecting a UHF 466 MHz receiver.

The solution is supposed to be Low PIM antenna systems. But how much is enough?
 

MotoBill

Prolific Contributor
CS Forums $upporter
Joined
Aug 20, 2018
Messages
194
A few examples… A site with a full performance UHF DTV transmitter with 8VSB modulation, which to an analog receiver is a 6 MHz wide wall of noise. Virtually every UHF repeater on the site with a 5 MHz TR spacing gets desense whenever it’s own transmitter is keyed, but in a test with the DTV transmitter turned off, no such desense occurs. Duplex links with spacing wider than 6 MHz don’t get the desense at all.

Mel,

On the surface it would seem that isolators are needed if the case of the UHF repeaters contributing to their own desense whenever their transmitters are keyed. However, it may well be that mixing is taking place external to the PA and associated transmission plant that's often very difficult to locate, especially at high density communications sites. Have you tried isolators or have you performed any PIM measurements?

It's difficult at best to measure for IM products or for the presence of interfering signals occurring at low levels near the bottom of a receivers useful sensitivity range when the spectrum contains large signals impacting the dynamic range of any such measurements. This is of course an interesting topic that almost certainly has no quick and easy answers.
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
The noise floor of a radio site with UHF TV transmitters is going to be way above normal. There is also the problem of keeping the TV transmitter out of the repeater receiver and many duplexers band pass capability is not up to the task. Taking a simple measurement with a spectrum analyzer or even an HP power meter right at the receiver can reveal a lot. You might have 0dBm or more energy from the TV transmitter making it past the duplexer and present at the repeater receiver. Depending on the front end of the repeater receiver that energy could be running the front end preamp or first mixer of the repeater receiver into IMD land, generating copies of the TV transmitter spread all over the place and in the worst cases running the front end preamp into saturation. Although any preamp will start generating IMD well before saturation.

A spectrum analyzer can see some of these problems and an HP power meter or equivalent can tell you the absolute levels of aggregate power being presented to the repeater receiver. As a final test you can possibly measure these levels at the output of the repeater receiver preselector and if its an integrated assembly you might have to open the receiver up and measure right at the board at the input to the preamp or mixer if there is no preamp.
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
On IMD produced inside the repeater transmitter, if there are UHF TV transmitters at the same site you might want to consider extra band pass filtering on the transmitter to help keep the TV transmitter energy out in addition to an isolator or preferably a dual isolator. Also consider retuning the isolator to absorb the most energy at the TV transmitter frequency.

Isolators are typically tuned centered on the repeater transmit frequency. You might have 35dB isolation with a single or maybe 70dB with a dual. But 40 or 50MHz away on the TV transmitter frequency you will only have a fraction of the isolation. Retuning the isolator will be less useful for protecting the transmitter in case of antenna failure or other VSWR problems, but it can double or quadruple the absorption of TV transmitter energy heading for your transmitter.
 

xpr8300

Prolific Contributor
CS Forums $upporter
Joined
Mar 1, 2012
Messages
647
In a past career, we were co-located with several DTV stations in a rooftop environment (1100' agl). Our systems were both 700/800 and UHF and the TV transmitters were UHF. Same deal, TV transmitters off, happy LMR receivers. TV Transmitters on, IMD all over the place (and a -70 noise floor AFTER our window filters). We spent way too much time and money on trying to get rid of them and finally just said to hell with it and moved.

The TV people just kept saying "we are in our channel mask, not our problem" and refused to do anything. Getting them to go dark for a few minutes of testing was like pulling teeth.

2 blocks away on a 1000' rooftop, the receivers were happy and the noise floor down to a much more reasonable -110. Still not great, but hell that was 40db improvement before you start on the IMD products.
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
If the TV transmitter is say 50kW and its 100dB down 30MHz away from your UHF repeater, its still leaking out -23dBm of broad band noise on your repeater input frequency plus antenna gain and minus coupling. This is where testing the repeater receiver performance connected to the antenna without the repeater transmitter on will tell you if its hopeless or if your transmitter IMD is contributing to the desense and you have some control over the situation.

I often use a bird wattmeter with rotating directional coupler slug and inject the receive frequency into the coupler pointing towards the receiver. First find your receiver sensitivity on a test bench for 12dB SINAD and note that. With the directional coupler setup described above and the directional coupler in line with the antenna port of the duplexer and the input side of the directional couple capped off with a load, find what level it takes to reach 12dB SINAD. That should be exactly the amount more in level than the directional coupler and duplexer losses are. Now remove the load on the directional coupler, connect the antenna and measure sensitivity again.

The receiver sensitivity is going to get worse by some amount with the antenna connected and that is your site noise floor contribution. If it goes up by up to 10dB then that's expected at any busy site. In that case you might be going from a receiver sensitivity of -125dBm to -115dBm which equates to going from .126dBm to .399dBm. Not great but its what's happening at most busy sites everywhere and the repeater should be perfectly useable, just not stellar. If doing the same test reveals your sensitivity goes from -125dBm with a load to something like -85dBm connected to the antenna with your transmitter off, then you have no choice but to move to another site or pray the TV towers fall over and their building burns down very soon.

If your site noise floor contribution is (only) 10dB and that goes down the toilet when YOUR transmitter keys up, then fixes are available like more transmitter filtering or isolators to keep the TV transmitter energy from creating IMD within YOUR transmitter and the now the TV transmitter building doesn't have to be burned down and you can be friends with them again.

In a past career, we were co-located with several DTV stations in a rooftop environment (1100' agl). Our systems were both 700/800 and UHF and the TV transmitters were UHF. Same deal, TV transmitters off, happy LMR receivers. TV Transmitters on, IMD all over the place (and a -70 noise floor AFTER our window filters). We spent way too much time and money on trying to get rid of them and finally just said to hell with it and moved.

The TV people just kept saying "we are in our channel mask, not our problem" and refused to do anything. Getting them to go dark for a few minutes of testing was like pulling teeth.

2 blocks away on a 1000' rooftop, the receivers were happy and the noise floor down to a much more reasonable -110. Still not great, but hell that was 40db improvement before you start on the IMD products.
 
OP
W

wa6jbd

Contributing Member
Joined
Oct 28, 2019
Messages
50
Thanks for all the replies so far. A few directed replies below...

@RFI-EMI-GUY - I've searched for an intermod calculator that would take transmitter BW into account, but I don't think such a thing exists. What I've done instead is use PRFIntermod and enter the wideband transmitter as multiple transmitters in one MHz increments, and specify the entire receive window (i.e. 806-815). It'll shake out the possibilities, and if it doesn't find 3rd order mixes, I'm not convinced I've found the problem. Using an N9320 with very narrow bw, preamp, peak hold and long integration times, I've been able to actually see the noise floor do stair-step level changes across the window filter passband, that sometimes actually matches a portion of the cellular band, which confirms the diagnosis of LTE spectrum being mixed into the receivers.

@MotoBill - In the case of the UHF radios, they do have circulators and additional pass cavities to keep the DTV transmitter energy out, but whatever goes on there is external. I'm of the opinion that it's occurring in the DTV transmitter itself, but it's hard to prove.

@CARC383 - You mentioned a couple of things of interest. One is the use of an HP power meter to measure aggregate power coming into the receiver. I've thought about doing just that, but haven't followed through. I'll give it a try and see what it has to say. Once while investigating the short life span of crystal filters on a VHF station on a busy mountain top, I received an RF burn while handling the coax coming from the receive antenna. I put a watt meter and load on it and measured 25 watts every time a low band 1/4 kilowatt transmitter keyed up. A couple of cavities in front of the crystal filter (which should have been there in the first place) solved that problem.

The other thing you mentioned is the IMD test set you built. Was that a commercial product? I'm trying to get our management to purchase one, and this thread is partially to get information to support the utility of said device.

We're seeing variations in how the problem manifests, depending on the specific site. Some, it appears our own transmitters are contributing. Others, we can shut the transmitters off and there is no discernible improvement. These are 700/800 Motorola GTR cabinets, btw. The filtering is pretty good, but I have one site where the mix appears to actually be occurring IN the receive window filter. A 3 dB pad in the input side of the filter reduces the mix products by about 6 dB, on the output it reduces by 3 dB. I'm not sure what to think about that. We've also seen cases where a really strong mobile signal (on the order of -30 dBm) will cause a low level mix. If it falls on it's own receiver, it'll create a stuttering audio of the last couple of syllables of the mobile when it unkeys. If it falls on a different channel's input, the dispatcher will hear it as crosstalk from another talkgroup. That one has only been seen at sites that are either duplexed to a single antenna, or have extremely close spacing of the T and R antennas... like, 2 feet.

It's not all doom and gloom, though. Out of about 60+ sites in the system, most see about 1-2 dB of desense, which is just environmental noise.
 

RFI-EMI-GUY

Prolific Contributor
CS Forums $upporter
Joined
Mar 8, 2014
Messages
1,086
This software for LMR takes into account transmitter and receiver BW. I know the developer Doug Perry of Riverview Software Solutions, LLC located in Alabama. It may have to be modified to accept modulation bandwidth in MHz or maybe it will since it appears he is marketing for wireless industry. The software can be leased.



 

KA1RBI

Contributing Member
Joined
Feb 12, 2014
Messages
78
Has anyone else observed this sort of thing? I’d be interested in hearing how widespread it is, and what, if anything, people have done to reduce it.

Just to add a data point - I've got a Radio Shack PRO-106 scanner that if you go to certain geographic areas it is TOTALLY deaf throughout the entire 700 MHz band. The symptom is the S meter is pegged 100% full scale EVEN WHEN the receiver is tuned to a vacant frequency. If you set up a search to scan the band the S meter remains stuck at full scale. These tests are on a rubber-duck antenna.

I fully understand this radio is not and was never intended to be commercial grade, and these models are uniquely terrible members of the line, but still...
 

Bill_G

Prolific Contributor
Joined
Jan 31, 2015
Messages
863
I use scanners for just that purpose: to find splatter and fubar generators.

A receiver has two jobs: receive the signal of interest, and reject all others. The "reject all others" part is where many rcvrs fall down.

But, in the case of broadband IMD - splatter - fubar, there is a direct hit on the freq of interest that no radio can reject
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
Thanks for all the replies so far. A few directed replies below...

@CARC383

The other thing you mentioned is the IMD test set you built. Was that a commercial product? I'm trying to get our management to purchase one, and this thread is partially to get information to support the utility of said device.

We're seeing variations in how the problem manifests, depending on the specific site. Some, it appears our own transmitters are contributing. Others, we can shut the transmitters off and there is no discernible improvement. These are 700/800 Motorola GTR cabinets, btw. The filtering is pretty good, but I have one site where the mix appears to actually be occurring IN the receive window filter. A 3 dB pad in the input side of the filter reduces the mix products by about 6 dB, on the output it reduces by 3 dB. I'm not sure what to think about that. We've also seen cases where a really strong mobile signal (on the order of -30 dBm) will cause a low level mix. If it falls on it's own receiver, it'll create a stuttering audio of the last couple of syllables of the mobile when it unkeys. If it falls on a different channel's input, the dispatcher will hear it as crosstalk from another talkgroup. That one has only been seen at sites that are either duplexed to a single antenna, or have extremely close spacing of the T and R antennas... like, 2 feet.

It's not all doom and gloom, though. Out of about 60+ sites in the system, most see about 1-2 dB of desense, which is just environmental noise.
On the filter that shows IMD improvement with attenuation on the input but not the output tells me the filter is producing IMD. It could be from internal corrosion or dissimilar metal junctions inside. It would be nice if the filter came apart so you could see the insides and look for problems.

The test set I hand built was specifically designed for testing DirecTV receivers in house and after it was made in about 2013 all new receiver designs were subjected to it. I built a second updated version then the company had a bunch of similar units made for all the receiver repair facilities in the US and Mexico. Besides the box shown here there is about $250k of broadcast modulators, upconverters and support equipment to finish off the system. The test set was not made specifically for IMD testing but it did hit receivers with more signal than they would ever see off a dish and could create IMD within the receiver for further evaluation. The test set originally output 32 broad modulated carriers ranging from 27 to 36MHz wide across the 250 to 2150MHz spectrum and I think the later version did over 40 carriers.

1.JPG2.JPG3.JPG
 

RFI-EMI-GUY

Prolific Contributor
CS Forums $upporter
Joined
Mar 8, 2014
Messages
1,086
I’d like to start an on-going discussion about something that I’ve seen nothing on in the various forums and mail lists. It’s got to be a problem for others besides us. I’m referring to intermod caused by broadband signals such as cellular and DTV that manifests as desense.

A few examples… A site with a full performance UHF DTV transmitter with 8VSB modulation, which to an analog receiver is a 6 MHz wide wall of noise. Virtually every UHF repeater on the site with a 5 MHz TR spacing gets desense whenever it’s own transmitter is keyed, but in a test with the DTV transmitter turned off, no such desense occurs. Duplex links with spacing wider than 6 MHz don’t get the desense at all.

Another example is 700/800 MHz P25 sites that are experiencing desense with or without their own transmitters being keyed up. Universally, where this has been seen, is sites with nearby cell sites operating 600 MHz (T-Mobil) and 700 MHz (Verizon?) LTE with carriers as wide as 20 MHz, similar to the ‘wall of noise’ DTV signals mentioned above.

The 600 and 700 MHz cellular works out as a 3rd order mix to hit the 806-815 MHz input band on the P25 sites. Additional filtering on the receivers reduces the desense only by the amount of the insertion loss of the filter, so it’s an external environment thing that can’t be filtered out.

I'm seeing other examples, but these should give you an idea what I'm talking about. Back in the good old days, an intermod mix required an almost direct hit on your receiver before you’d notice it, then you could listen and identify what it was. Now, it’s intermod manifesting as desense that’s absolutely intractable and there seems to be little that can be done.

Has anyone else observed this sort of thing? I’d be interested in hearing how widespread it is, and what, if anything, people have done to reduce it.

Mel - WA6JBD
I should add. 3rd order mixes on-site are nearly impossible to rectify. You may be spinning your wheels, especially at DTV power levels.
 
OP
W

wa6jbd

Contributing Member
Joined
Oct 28, 2019
Messages
50
On the filter that shows IMD improvement with attenuation on the input but not the output tells me the filter is producing IMD. It could be from internal corrosion or dissimilar metal junctions inside. It would be nice if the filter came apart so you could see the insides and look for problems.

The test set I hand built was specifically designed for testing DirecTV receivers in house and after it was made in about 2013 all new receiver designs were subjected to it. I built a second updated version then the company had a bunch of similar units made for all the receiver repair facilities in the US and Mexico. Besides the box shown here there is about $250k of broadcast modulators, upconverters and support equipment to finish off the system. The test set was not made specifically for IMD testing but it did hit receivers with more signal than they would ever see off a dish and could create IMD within the receiver for further evaluation. The test set originally output 32 broad modulated carriers ranging from 27 to 36MHz wide across the 250 to 2150MHz spectrum and I think the later version did over 40 carriers.

In regards to the filter, we did manage to take one apart the other day. I can't say as I'm impressed. I'm bothered by the fact that the cavity resonators are half machined out of the aluminum housing, and half copper fastened to the lower aluminum part. Also, the output section is a copper stripline section screwed onto a resonator post on one end, and a crystallized looking solder joint at the output connector. Another interesting anomaly was noted with these filters, but that is to be the subject of a separate post.

Your test set is truly a work of art! I am impressed with that.
 
OP
W

wa6jbd

Contributing Member
Joined
Oct 28, 2019
Messages
50
I should add. 3rd order mixes on-site are nearly impossible to rectify. You may be spinning your wheels, especially at DTV power levels.

I've had some success in the past, but only in cases where I can apply the fix to a problem transmitter. In poking around on the web, I did find a white paper of sorts at the Anritsu website discussing wideband PIM, but the discussion was directed towards cellular operators, not LMR. Same affect, however. At least someone in the industry is recognizing the problem.
 

CARC383

Prolific Contributor
Joined
Jan 16, 2019
Messages
240
In regards to the filter, we did manage to take one apart the other day. I can't say as I'm impressed. I'm bothered by the fact that the cavity resonators are half machined out of the aluminum housing, and half copper fastened to the lower aluminum part. Also, the output section is a copper stripline section screwed onto a resonator post on one end, and a crystallized looking solder joint at the output connector. Another interesting anomaly was noted with these filters, but that is to be the subject of a separate post.

Your test set is truly a work of art! I am impressed with that.
Thanks on the test station build, I like to create eye candy for myself and if someone else appreciates it all the better. On the disassembled filter, it would be interesting to clean and treat all dissimilar metal junctions with some Caig Deoxit R5 then resolder crystallized joint and see if it still goes non linear. If it tests ok then its either the crystallized solder joint and you fixed it or just a matter of time until the metal junctions get funky and it starts all over again.
 
OP
W

wa6jbd

Contributing Member
Joined
Oct 28, 2019
Messages
50
On the disassembled filter, it would be interesting to clean and treat all dissimilar metal junctions with some Caig Deoxit R5 then resolder crystallized joint and see if it still goes non linear. If it tests ok then its either the crystallized solder joint and you fixed it or just a matter of time until the metal junctions get funky and it starts all over again.

Yeah. I've got to devise some sort of test to measure PIM before and after, with some of the suggested cleaning. I'm real curious if the odd construction of the things is just asking for trouble.