Aeroflex 3920 Autotest II results inconsistant

[rmarks61]

My shop has several Aeroflex 3920's with 3.3.1 alignment verison and 1.7.8.5 system verison. When using the Autotest feature for the tuning and alignment of Motorola XTS5000 800Mhz P25 radios there is an issue with repeatability of the results between 3920's. One will test/tune and pass all tests then put it on another 3920 and just test the alignment and several tests fail. Mainly the deviation balance tests. Tune it on this 3920 and all test pass, then put it back on the first 3920 it fails on tests it passed. It will do this on the same 3920 too. Now I have a customer whose radio shop uses the 3920 and they have the same issue. I used to align all these parameters manual using a 2975 and the tuning software but the difference is how the test is done on the 3920 verses how the service manual says to do it. The 3920 tests, for example durning the deviation balance test from low to high frequency the service manual clearly indicates that this test should be done from high to low. There is interaction between the soft pot settings and if done as per Motorola it minimizes this effect. When doing the alignment manually you can see this interaction. Really don't know if this is an issue or not but it is different than the way the service manual directs you to do it. Aeroflex sent a new alignment verison 3.3.5 but this didn't help. Further there is a new version of 3920 software about to hit the streets in order to support autotesting of the APX and maybe this will have a positive effect on this issue. The only interim solution I could come up with is to tighten the alignment parameters on one of my 3920's. I edited the pass/fail criteria by tighting it by at least 50%. This improved the ability of the 3920 with modified pass/fail parameters when it aligned a radio to pass on a 3920 with the default parameters. Why is there a repeatability issue with 3920's as it relates to Autotest and the results obtained?

Anyway I'd like to hear from anyone who has seen this or has any information on this issue with the 3920.

[escomm]

When saying high to low, do you mean you need to match the high to the low? Or go from the high to the low? Proper procedure is to take the reading at the low level and match the high to it

[Wowbagger]

I'll raise the order of test issue with the folks working on it. That should be fairly simple to change, as that is done in scripting and not the core system.

[rmarks61]

Highest frequency to the lowest. For reference see Service manual 688109C28-E. This does have an effect.

Thankyou for bringing this to the folks at Aeroflex as there are many 3920's out there. The Motorola training center in Schaumburg uses them and so does the Federal Depot in Lanham as well as most of my customers who can afford one. I'm seeing inconsistant results on Deviation Limit and Reference Oscillator too. Even with these quote inconsistances from one 3920 to the other mostly occuring in the DEV/BAL, DEV/LIMIT the Modulation Fidelity tests always pass and this is a excellent indicator of the 2-tone alignment. This is further proven by manually testing the deviation balance and observing what occurs when the tones are out by lets say 5%, the 3920 is set for a 2% tolerence in which the 5% indication would fail. You see very little effect on modulation fidelity and its 3920 tolerence is 5% with most of what I see at or below 2%. Even at 5% deviation balance the overall effect is slight. The trunking simulator indicates that the radio is 100% in spec. The only other issue that can be problematic is for the user or the folks that just follow a process and have no clue as to what the test really does/means see the word "FAIL" on a screen and start raising the flag. This is what's happening here as the new P25 system is delayed caused by(see FSB10564) and this 3920 issue. Man these autotesters are supposed to be faster, better, smarter but it looks like I'll be here for a while longer.

[rmarks61]

I take it from your response's that neither of you are aware or have observed this issue? Do either of you work with these instruments? I see your avatar, an Aeroflex employee Wowbagger. The folks at Aeroflex are keeping the parascope down for the time being. They must be aware and are either banking on this new release for 3920 or there is a systemic issue with the autotest. Either way Aeroflex better stay ahead of this!

[escomm]

Where in the manual does it say go highest frequency to lowest?

[xmo]

"Radio Alignment Procedures: Transmit Deviation Balance Alignment
...
3. Select the TX Deviation Balance alignment screen. The screen indicates the transmit
frequencies to be used. ...

4. Click the desired frequency field (starting with the highest frequency shown). ..."

[escomm]

Odd, my manual says the exact same thing, minus the parentheses.

[rmarks61]

Yes it been in the manual since the B rev.

Further examination reveals that the resolution of the softpot settings for an XTS5000 is a factor also. 0-127 in adjustment great, but when it gets to were you need less than or equal to 2% difference between high and low tones you may have 1 maybe 2 increments to set this level. So this may not be enough resolution for repeatable results. Now I don't know what the DAC's are in the 3920 or their resolution, but it must be at least 16 bit or higher. We also know they are using an 800Mhz processor in the 3920's and some have a newer 1600Mhz. Based on the sample/hold code and the size of the buffer/bin the 3920's may suffer from not enough sample rate for smoothing/averaging so as to more accuratly generate consistant results at least at the test tolerences. Maybe the 3920 is using 128 bit bin for each buffer, I don't think the 3920's with the 800Mhz CPU have the horsepower to process any more than that. This is one potential reason why Aeroflex doubled the CPU speed and produce the 1600Mhz version of the 3920. Operating system is different also, interesting. My shop doesn't own one of these with the 1600Mhz CPU so I can't do a side by side and see if there is a difference. Aeroflex should remove the PASS/FAIL results from the DEV/BAL and DEV/LIMIT tests. The overall test for the PASS/FAIL as it relates to these test should be MOD FID(modulation fidelity) as this is a direct result of how well these previous settings were made.

I still haven't heard from any that have this issue. Please if you have more than one 3920 see if your autotest results are consistant between them. We need a laymans explanation for this issue so the customer who has very little if any training can understand it and not be alarmed. Customer education is needed.

[Wowbagger]

Now I don't know what the DAC's are in the 3920 or their resolution, but it must be at least 16 bit or higher.

It doesn't work like that. We digitize at a very high rate (tens of megasamples per second), and as the signal is decimated down you get what's known as "processing gain" - by the time we get the signal down to a baseband equivalent datastream we have about 32 bits of resolution.

We also know they are using an 800Mhz processor in the 3920's and some have a newer 1600Mhz.

Totally irrelevant to this discussion.

The biggest issue is the resolution of the pots in the radio - as you said, if you need a half-click, there's not much you can do.

[rmarks61]

Decimation in this sense is by definition the down sampling of a signal. Accordingly if your "digitizing at a very high rate" then your using interpolation first. There are several methods of interpolation. Then down sampling(decimation) to achieve base band results. Its been a while, but I do recall that FIR filtration is critical for both. These processes are required to make the sampling rate of the signal compatible with the bandwidth of the signal processing system. Seems a faster CPU would make life easier for this process.

Aeroflex or Motorola should then caveat this apparently known limited adjustable resolution issue with the XTS/XTL series of transcievers as it relates to the 3920 etc.

[Wowbagger]

No, we aren't. We sample the IF at a high rate - there's no interpolation. Then we decimate and downsample that high sample rate IF.

Look, I can tell by what you are writing that you know just enough of signal processing to think you understand it without actually understanding it. I'm sure you don't think so - I'm sure you think you understand it all. When you can adequately describe why a digitized signal after a sample and hold has a sinc response (and can accurately describe what a sinc response is), and can discuss the benefits of dithering vs. a one bit delta-sigma converter, then I'll assume you might actually understand the nature of what is going on.

[rmarks61]

Wow, when are you going to ask what the Nyqust zone is? One function of the DAC is sample and hold. This sample and hold is in the time domain. Then there is the DAC impluse response which I think is what your trying to have me define/describe. Impluse response is just the overall frequency response. So the output of the sample/hold is rectangular looking in the time domain that mirrors the input level to the DAC. I believe that the "sinc function" is the math used to define the filters response but in the frequency domain. The benfits of dithering vs. one bit-sigma delta? One can improve the one-bit sigma delta converters by employing forms of dithering. One bit-sigma delta converters suffer from low level "artifacts"and dithering helps limit these. Your question is confusing as dithering is a way of improving the one bit sigma delta converters. How I do so far?? I regret if this is not a sufficent understanding, but frankly this is all off the thread topic. I just thought I would indulge you.

Granted my understanding of ADC/DAC's implementation is limited as you've so "nicely" indicated. Well it clearly appears that although Aeroflex and yourself have more knowledge in this area it has been of no help resolving this issue. Or maybe you've got you mind made up on this issue too?? At least the folks at AEROFLEX are a little more shall we say open minded. Maybe because we spent over 600K purchasing their equipment. At any rate although your the resident expert, your responses have not been helpful and this last post was down right insulting.

Anyway back to the topic question which is if anyone using a Aeroflex IFR 3920 in the autotest mode is having issues with repeatablitiy in the results? I have 3 3920's and my customers have at least one in their shops and they are all having this autotest issue. Please let me know what alignment version 3.3.x?? ,system verison 1.7.8.x?? and finally what is the serial number. This data will be collated and submitted to Aeroflex for evaluation.

I would also like to state for the record that this is not a XTS5000/Motorola product issue. The folks at Plantation have already evaulated this and have determined that the Aeroflex 3920 needs the autotest revisited.

[Wowbagger]

Look - when you start telling me I don't understand the very equipment *I DESIGNED* - I get insulted. You keep saying "it can't work like that" - it most certainly DOES work like that, I know because *I HELPED DESIGN IT*.

The issue under discussion has NOTHING to do with the stream of errors coming from you. The problem is 2 fold:
1) the order in which we adjust the softpots in the radio - which I have already called to the attention of the people writing the autocal program.
2) The limited resolution of the softpots themselves - which is a Motorola issue.

You keep talking about "limited ADC resolution in the 3900", and the CPU speed of the main processor (which, if you had helped design the box LIKE ME, you would know has NOTHING to do with the signal processing, which is handled by a completely separate set of DSPs). I kept trying to tell you that you were wrong, at first nicely, but when you start asserting that I don't know how the very equipment I helped design works, I finally have to put an end to the ********.

Now, if you are willing to actually LISTEN to what somebody who KNOWS what they are talking about is saying, we can continue this. However, I am about 1 message away from deciding you are a blow-hard and putting you on ignore.

[rmarks61]

Well, its a fact that these softpot settings are more than sufficent to satisfy TIA, FCC, Motorola standards. The static test results when using the autotest function between 3920's or the same 3920 aren't 100% repeatable. Now I want to make it clear that this doesn't mean the radio won't function on a P25 system it may very well after the 3920 does it thing but still requires verification. If the 3920 is used as a alignment tool and the results verified by dynamic testing via trunking simulator/air by ETA/CET cetified technicians with the 3920 results verified and failed autotests evaluated manually then the PASS/FAIL is generated by the certified technician. Much to my regret anybody can buy a 3920 if you have the coin. They are used by unqualified folks that because of the 3920 results create delays in projects costing the taxpayer allot of money. Some are using them correctly but others are using the 3920 autotest as the final QA of their fleet. Now can you see the bigger picture? Most haven't a clue as to what the tests withing the autotest are or mean. They need these results for liability reasons too, which is a whole different chapter and is out of the scope of this thread, but is something Aeroflex should be aware of.

OK now that I've re-hashed the issue do you agree the method in which the box you helped designed is employed can be a problem? In other words is the Aeroflex 3920 intended via autotest to certify a P25 radio for use on a P25 public saftey system? This is a yes or no question. Funny thing is there is no mention of this "certified P25 operation after autotest" in any documentation provided with the 3920 or in the Aeroflex website 3920 advert. If no then the 3920 is just another tool in the arsenal of the certified technician and Aeroflex should at least caveat this. Maybe when the autotest screen appears a disclaimer box appears and the operator has to indicate via a tic box that he/she understands the intended use of the 3920 and should expect results such that they are. If the answer is yes then I'm very dubious.

Wowbagger again thankyou for suggesting to Aeroflex to change the autotest to be more in compliance with the Motorola service manual as far as testing the XTS5000. This should be script(s) wide as all the XTS/XTL series is the same way. It is not my intent to offend or insite any board member, but if your serious about wanting to improve your product then please remain objective and give the benefit of the doubt until proven otherwise. I would never threaten to ignore you as I value the information you provide to the board not just on this but please do what you think/feel is right for you. I just want to get a handle on this so I can move on. 20K radios under maintainence so issues like this cause headaches.

[Wowbagger]

Consider this the last time I will read your responses, as you seem to not read mine.

I have raised the issue with the design team writing that code. I have now said that in THREE separate messages.

Since you seem to be little more than a troll, welcome to my ignore list. *plonk*.

[rmarks61]

Wow again! Man I read your posts with hopes they may actually be useful. It turns out your only use is to inform Aeroflex of their obvious oversight in the scripting for testing and alignment as it relates to autotest on the 3920 of Motorola radios. One can only wonder what other "oversights" Aeroflex products may have.

Now your posts are for the most part not on topic and its your comperhension that should be questioned. As a result of all this interaction you've decided that I'm a "troll" how convient .
Its is very interesting that you ignored my pointed questions about the use of the 3920. I think we both know the answer, but the forum may not so why don't you address them? Oh that's right I'm a troll. Makes it real easy for you to doesn't it? Well the folks at Aeroflex don't see it this way and are being very objective of my concerns on the 3920 autotest results.

1. Should the 3920 autotest be used as a QC device for overall system performance of an 3920 aligned radio?

2. Once a P25 radio is aligned via the Aeroflex 3920's autotest is it "certified" to then work on a P25 system?

On second thought maybe it would be better for a "real" vetted Aeroflex employee who can speak for their products answer these questions so never mind wowbagger.
I'll answer the questions as I see it. 1. NO, 2. NO

Now for everybody who may have a 3920 or are thinking of purchasing one I'm working on obtaining further information on this autotest issue. I thought I had a good chance of obtaining some valuable feedback here on batlabs on this issue because of the amount of knowledgable folks that frequent here. Never thought I would be labeled a "troll" and then put on the "pay me no mind" (ignore) list from a member that has many posts and an supposed employee of the very company whose equipment I have issue. Wowbagger's disclaimer didn't go unnoticed either. Something to think about before pulling the trigger on a $45K box.

[alex]

Guys -

I think honestly you should talk this one out either on the phone or rmarks61 you should reach out to Aeroflex about this in an official capacity. Both Wowbagger and also Rod read the forum here and can follow up or have someone follow up with you. This looks like one of those things that once you get on the phone and hash out you realize that both of you are making sense but missing each others mark with the writing....

If you can please keep it civil all of us would appreciate it.

Alex

[radioinstl]

2. Once a P25 radio is aligned via the Aeroflex 3920's autotest is it "certified" to then work on a P25 system?

Rick you should well know that the only one who can "certify" a unit on a system is the people who run that system.

No manufacture or service company like Aeroflex is goin to certify any thing on a specific system.

A Manufacture will ceritfy their SU via CAP testing and a SDoC that when aligned to specs it will work on "X" type of system that is listed in the SDoC.

I can't speak for Aeroflex, but I am pretty sure what they say is their equipment will provide results to X percent accuracy when the equipment is calibrated to spec.

Now none of this will mean that it will work on your system since neither the SU manufacture or Aeroflex (or whatever brand you use) controls or knows how your current system is performing or aligned.


So to anwser you questions.

1. No, no device can give you over all system performance QC for a radio since system performance involves radio performance and system performace which are independtant of each other.

2. No for the reasons given above

The other thing to keep in mind is when you say "certified", you need to define what the standards you are certifying to and how.

When you say the radio is certified to work on the system, you should have defined exactly what that means and how you have tested that to meet those requirements.

I understand the liability issue. If you say the radio is certified to work on the system, and it fails and a firefighter or police officer dies, the first thing they are going to ask you about in court is your " certification". If you do not have a very well defined standard of what the certification covers and how it is tested with pass/fail criteria, they the certification is useless.

If you want to provide that type of certification, there are plenty of us in the NCR area that could provide guidence in writing them, but the most important group to provide imput to that certification would be the system users.

Scott

[rmarks61]

Scott this is part of the point I've been trying to convey. The idea of using an alignment tool's "static" test data to determine the radios abiltiy to operate on a "dynamic" system is foolhardy. As you mentioned there is or should be a group that determines this via manual/real world if possible and simulated system operation. Funny thing the 3920 has this system simulation capability, but unlessed used is not relevent and this whole thread is related to the autotest function only. As I said before anyone can purchase a 3920 and use it as they see fit. What I've been trying to get from Aeroflex is a statement that basically says that the results obtained from the 3920's autotest and its evaulation is only as good as the folks running the 3920. Maybe some consider this a given, but how would it be if 200 radios that were PM'ed by the contracted Motorola service provider were returned because the customers 3920 indicated that certain autotest parameters fail, but haven't a clue to what the so called failed parameters mean or its effect. It can cause big problems believe it or not.

Again to re-cap the issues I'm having with the 3920 are two fold. First the repeatability of the 3920's autotest results either between boxes or the same box. Now when and if Aeroflex corrects the script for the order of alignment, will this have some effect, it may. Second, although there is the question of who's responsibility is it for insuring that the product is used for its intended purpose we can all agree that there is a certain level of expertise required for the effective use of this instrument. This as clear as can be made of these issues, but as this thread has de-generated it may be best to have the moderator close it and I'll handle this offline.

Thanks for all who responded.

[radioinstl]

I agree with the machine is only as good as those using it. It is not a replacement for the knowledge, skills and experience needed to do the job with out the autotest.

A good correlation to public safety is opening a roof with a saw. Were I am from we dont let the new guys on the truck touch the saw for 6 months. This gives them 6 months of time doing the old fashion way and the learning that goes with it. Then after 6 months, when they are using the saw they have the knowledge to be able to function even if the saw fails and to know why and how the saw makes it easier.

Same with the autotest. If you cant do it with out the suto test you should not be using it. You need to have the skills, knowledge and experience to know what it is testing, how it is doing it and to be able to read the results. As you have pointed out with your inconsistant results, this prior knowledge is key.

Bottom line, the autotest does not replaced the skilled tech, but gives them another tool.