JPS Voter not always initially voting the best site

[RF_Burns]

Hello,
I have a 5 site, 2 channel simulcast system using Harris Synchrocast 3 with NetXpress and NetXpress LX Muxes with VF-40 E&M audio cards over Motorola 4.9Ghz PTP600 links. The repeaters are MTR2000.

Everything works great except 6-7 times out of 10 it will not initially vote to the best site. It then takes 2 or 3 seconds for it to vote the best site, long enough that short transmissions could be only noisy. I've gone over the settings with Benny at JPS and while he is very helpful, the problem persists.

Its just not one site, its all over the County. I can stand at the Hub site and watch the voter while a mobile drives around the County. The terrain is mostly rural and fairly flat so a mobile can usually hit all sites with 2 sites being definitely noisy while a portable usually hits one site clear and 2 others less so. Generally it will vote a noisy site, then after a second or so it will vote to the clear signal.

I made a video of it and as your can see it happened nearly every time. Sorry about the background noise, the place is full of screaming fans!
I'm standing at the Hub with a portable radio, the Hub site is the furthest left card, next a site 12 miles south, next site is 14 miles west, next site is 19 miles north, the furthest right site is 35 miles north so it doesn't hear my portable.

https://docs.google.com/file/d/0BwUTaLs ... edit?pli=1

Both channels do the same thing, so its not a voter specific fault.

Any ideas?

Thanks

[Bill_G]

Since you're using NetExpress, you're using ethernet, not the T1 provision in the 4.9 Canopy, correct?

The voter is not getting enough high freq content to work correctly. The codec in the Harris is rolling off the content above 4khz leaving little for the voter to work with. Go into each MTR and boost the line equalization all the way up on the high end. If you have any local receivers feeding the voter minus the Canopy / Harris link, you will have to cut them.

There is a similar equalization adjustment in each SVM card that you might have to apply [SW3, 7 & 8] on each rcvr through a Canopy / Harris link. You may want to consider using AM mode instead of FM mode [SW3-5] in the voter.

[FMROB]

I just watched that video and it would appear the the line levels may be a bit low. I see very few red blinks upoin high points in the voice test. It appears to only blink at the end of the transmission or with squelch noise (hard to tell with fans). I am not saying that this is the problem, but it should be corrected. The SVM level adjustment should be made so that on spoken voice the peaks of voice flash the red.

Although I am not familiar with the Harris equipment, what Bill states is correct, even with rtna circuits or local copper. That has been my experience.

[RF_Burns]

Bill,
Correct, we are using Ethernet. The Harris specs have the VF-40 audio cards rolling off steeply above 3khz and the VF25's I have used in T1 links say the same. I have seen this on those systems but maybe once or twice in 10 times. Benny says the SNV12 sees everything above ~2khz as noise (which seems low to me).
All receivers even the local receiver to the Hub go though the same Mux.

I can see where boosting the high end may help.

Thanks

[Will]

Bill,
Correct, we are using Ethernet. The Harris specs have the VF-40 audio cards rolling off steeply above 3khz and the VF25's I have used in T1 links say the same. I have seen this on those systems but maybe once or twice in 10 times. Benny says the SNV12 sees everything above ~2khz as noise (which seems low to me).
All receivers even the local receiver to the Hub go though the same Mux.

I can see where boosting the high end may help.

Thanks

Exactly. Boost the highs with a simple selected series coupling capacitor where the RX audio goes into the voting cards.

[Bill_G]

Bill,
Correct, we are using Ethernet. The Harris specs have the VF-40 audio cards rolling off steeply above 3khz and the VF25's I have used in T1 links say the same. I have seen this on those systems but maybe once or twice in 10 times. Benny says the SNV12 sees everything above ~2khz as noise (which seems low to me).
All receivers even the local receiver to the Hub go though the same Mux.

I can see where boosting the high end may help.

Thanks

Exactly. Boost the highs with a simple selected series coupling capacitor where the RX audio goes into the voting cards.

Don't need to. The MTR's have a wide range (+/- 12db) equalization cut/boost in the line driver alignment screen, and the JPS has a three level boost on the other end. But, if he has a local receiver, he will have to cut it to match the linked curves with a simple RC network across the SVM input.

[Bill_G]

Bill,
Correct, we are using Ethernet. The Harris specs have the VF-40 audio cards rolling off steeply above 3khz and the VF25's I have used in T1 links say the same. I have seen this on those systems but maybe once or twice in 10 times. Benny says the SNV12 sees everything above ~2khz as noise (which seems low to me).
All receivers even the local receiver to the Hub go though the same Mux.

I can see where boosting the high end may help.

Thanks

You're welcome. There's a thread around here someplace where I solved the exact same problem a couple years ago. You might want to add 120ms delay to each SVM to give the cards a little time to settle down before voting. The delay heard in the field will be easy to deal with.

[RF_Burns]

thanks Bill,
I believe I found the post that you are talking about from June of 2010.

In my minds eye I think the equalization is set to max attenuation above 3khz in the MTRs and I never gave this any thought. I'll work on this and post my findings.

Thanks again

[d119]

I'd still like to hear what the audio through one of these comparators sounds like when the thing is configured for status-tone based voting.

[FMROB]

All of the units that we service are status tone based systems and they work well. Less "comparator" nosie than the spectra tac, and easy to set levels. The audio is not so tragic.

[Batwings21]

I'd still like to hear what the audio through one of these comparators sounds like when the thing is configured for status-tone based voting.

Listen to a online feed of Oak Lawn IL fire aka Mabas21 this is a 12 site JPS using status tone over pots lines. Porter County IN sheriff is a 9 or so site JPS with mixed telco and mds subrate t1 links.

For the Oak Lawn one "Central" is the console directly connected to the JPS there are a few others on control base that sound like crap ( not my equipment )

[RF_Burns]

I checked the equalization settings on the MTR2000 stations, they are actually set for -6dbm (voice) and 0dbm at 2500hz, so in effect there is +6dbm of Eq in place. I should actually compensate for this in the SNV12 cards or redo the MTR and SNV settings.

I also checked the settings of the VF-40 audio cards in the Muxes. They are set to flat response 5-3000hz +/- 0.5db so they should be good. All the sites go through the same audio cards, set to the same parameters so there should not be any difference between sites.

While at the site, 3 Departments were out and both channels active. Many times I watched both voters vote the wrong site. One time a mobile was voted to the furthest site (40 miles away) over the closest site 2 blocks away, it was unintelligible noise for at least 3 seconds before it corrected.

Previously the system was setup with XPR repeaters and Rad Muxes (same voters) and it was doing the same thing, so I'm not sure fooling with the Equalization will help.

[Bill_G]

That's new information that the voter was reused in a previous system, and it did the same thing back then. I would go back through how it is set up very carefully, and I'd do a full end to end alignment, line freq response, etc. These are good voters, and they do exactly what they are set up to do. The problem is they have a lot of possible settings. I am very sure it is an equalization problem. What is causing it needs to be determined.

[FMROB]

Ok, so I will go back to my original thought, how did the line check out. There appears to be low audio ont he voted circuits, as there is very little or no red indicator lights on spoken voice. This can cause an issue.

From the manual:

As shipped from the factory, the RX and TX levels for each SVM are set to -10 dBm. If level requirements differ, both the RX (input) level and TX (output for transmit audio) may be adjusted using the RX and TX level controls on the SVM PC board.
To check a module for correct audio levels, monitor the SVM’s PEAK and NORM LEDs while a typical speech reception is coming in from the associated receiver. Press the module’s SELECT pushbutton to force-vote the site so you can listen on the SNV-12 speaker. When the RX level control is adjusted properly, the NORM LED on the SVM will be lit on normal audio levels. The PEAK led should only come on during the highest audio peaks. If the NORM LED never lights, or if the PEAK LED lights frequently, then the audio level into the SVM is not correct and the RX level potentiometer should be adjusted.
It is important the audio levels be correctly adjusted, as they will affect the operation of the DSP algorithms for both voting and line fault detection.


Also, check this out

SW5-4 COR Delayed with Receive Audio
When RX audio delay is added using an SVM’s rotary switch SW1, the COR (unsquelched) indication can either be reported immediately or delayed along with the RX audio.
When not delayed, the audio delay is used to compensate for other system delays so that the site is voted and repeaters keyed, etc., before the audio is passed through the system.
By turning on dipswitch SW5-4, the reporting of active COR following its detection is delayed by the same amount as the RX audio. This can be useful when the voter is used to select from among signals arriving from very different media. For example if most sites are linked by microwave, but one site is linked via satellite, it’s easier to ensure proper voting if the positive COR (unsquelched condition) is delayed along with the audio.
The ability to delay COR is not available with SVM-1 modules. See Section 5.6 for more information.

[Batwings21]

1) I second that unless you are talking quietly or have a radio with poor deviation the audio levels are not adjusted properly. Injecting a full quieting 1k tone at 2/3 system deviation plus pl should just light the yellow light steady and you can measure +5db at the test point on the front of the svm.

2) check this dip switch on the cpm

S4 Voting Criteria; Transition Timer
S4 determines the time that one site must exhibit a better SNR or noise level than the presently voted site before the SNV-12 will change the voted site. If site "A" is presently voted, site "B" must maintain a better signal than site "A" by the amount set by the CPM switches SW2-1, 2, 3 for the entire duration of the time set by SW4 in order to be voted over site "A". The ideal position for this setting depends on the system requirements for voting speed and smoothness of voting operation.

[FMROB]

Batwings,

I agree with your 2/3 dev and light the yellow light, however I found that the audio quality andvoting characteristics to be somewhat sluggish with the JPS unit with yellow light only 90% of the time audio. In his video, only one red light light on the far right card once, and it was upond dekey. Spoken voice, even on peaks was not lighting red - which I found to be troublesome.

So, the reason why I queston this was the first time I worked with a JPS I started setting the system up the "spectra tac" way. We were using mixed transport mediums, micro, rtna, and local copper. So I went thorugh and did all the alignments for a -13 system, equalized the lines, etc. on the far end equipment. Everything looked and sounded great on the meters. However, I was exhibiting exactly what his video showed, low audio across the board but everyhting was cheking out. So I read the manual and was laughing about the simple SVM adjustments. I tweaked up all the individual svm cards to have red on spoken voice peaks and the system came alive. Audio was great, clear, inteligible, and loud. I was so worried because because at the same time I was taking a department that was running wide band for over 15 years to installing a new voter and narrowbanding. You could imagine the beads of sweat running across my brow, but it all worked out.

Good catch on S4.

[Bill_G]

I will have to check mine to see if they have +5db at the front panel test point with full deviation. Average audio gets a blinking / almost solid yellow with only the occasional red. The boomers will paint it red, but still sound good.

RF_Burns sent me an email, and I sent him the dip switch settings I use. We'll see how they compare.

[Batwings21]

Yeah if you have people who are consistantly low deviaton you may need to add a little boost to the system by changing your alignment procedure. The main thing is that the yellow lights are on when someone is talking and the red lights frequently flash with the peaks as FMROB said.

[RF_Burns]

My little test portable is a tad shy on deviation, when users are talking the yellow is flashing constantly with red peaks.

I heard back from Harris today who said "Your voter is receiving the E&M signal before there is any audio with which to compare - so it is trying to vote by comparing noise floor level among the channels, pretty much a random draw."

So I've decided to try switching from E&M to Pilot Tones. That should ensure audio is there when the Pilot tone drops. I made it around to all sites today but one so Monday I'll setup Pilot Tones in the last MTR's then I'll switch the SNV to Pilot Tone operation and see how that goes.

[RF_Burns]

Just a followup on this issue. I programmed the MTR's at the last site today for Pilot Tones and went back to the HUB and switched the voters over.. IT WORKS!! WOO HOO!
Out of about 20 keys on each channel, it only voted incorrectly once, and to be fair the site it voted is still pretty clear and it did switch back to the Hub site within a blink of my eye. I had to check the Stats page of the voter to be sure of what I saw.

So apparently the E&M is instant, but there is a delay for the audio to start streaming from the codec. So the voter initially sees a signal with perfectly clear audio from every site who's squelch opens. At some ms later then the audio starts to flow, but by then the voter has made up its mind. Using Pilot tones means when the tone drops, the audio is right there for voting.

Thanks for your help guys.

[Bill_G]

Ok. I guess none of us were quick enough to ask if you were using status tone or cor to indicate carrier det at a rcvr site. Yes - voip has latency but E&M typically passes in near real time. I'll have to read up on the VF40 card to see if there are settings to delay them. NXU's have such a setting to help sync the voice with closures - ie: ptt followed by several tens of milleseconds of silence before the voice starts.

[RF_Burns]

I've always used E&M and never thought there would be a large delay, but up till now its be T1.

I've heard some transmissions getting cut off periodically tonight. I'll have to add 2175Hz filter on the Rx as I believe audio is faulting the pilot tone detector in the SVM modules

[Bill_G]

Ah - good point. You said in it's previous incarnation it was using T1's. The signaling would have been perfectly aligned with the audio. Don't know why it didn't work back then. Shoulda. I think there is a deeper problem and you found a work around.

Did you compare the setting I sent you to yours to see if there was a difference?

[RF_Burns]

Bill,
I should have been clearer and said, up till now, previous systems I have installed have been T1. This is the first voted system I've done with Ethernet, first with Rad Muxes and after upgrading to simulcast with Harris.

My settings were the same except for COR polarity and I had 60ms of Console delay as it uses tone keying.

I'm now using Pilot Tones on Rx and E&M for Tx key to the remote sites.

Harris had suggested changes to packet size and jitter buffers as the fix. I figured using Pilot Tones would be the best fix as the radio Rx audio should be immediate when the radio drops the status tone.

[Bill_G]

Do you have jitter issues on the copper side, or did they just suggest increasing the jitter buffer? You can check the 49600 ethernet stats for more detail then you'll get from the Harris.

[Batwings21]

When using pilot tone over IP connections we have had jitter problems resulting in pops and clicks over the repeater when the jitter buffers get full. Going to E&M on the nxu-2's was the solution for those systems, but we did not have the issue you were having.

[techmi]

To All who have contributed and or participated in this thread, I want to extend a HUGE thank you. If you came to this thread looking for information on how to make the JPS Voter system work the way it was intended then I highly recomend that you read thru all of the posts on this thread. By applying the "best practices" suggested by others on this board to a system that I was working on that was operating in a "Less than optimal" configuration, completley changed the way the JPS Voters operated.

Like one of the other posters on this thread said; "WOO HOO It WORKS"............

Batboard Rules !!!!!!!!