Yeah, I doubt that to be the case here. I graduated Navy Micro Miniature Repair School in the 90¡¯s, and a PLCC is, ¡°no thing but a chicken wing¡± compared to what I have dealt with in the past. I have inspected this EEPROM under the microscope there is no flux, and if there were I do not use the typical, pine pitch based flux in any of the work I do. In theory you could leave the paste I use on without cleaning, but I clean anyways.
I actually prefer UltraSolve to IPA. Expensive, but worth it!
toggle quoted message
Show quoted text
On Mar 6, 2022, at 10:55 PM, Lothar baier <Lothar@...> wrote:
?
Aaaaaah ?now I see
??
I learned this the hard way back when I hacked Options on the 8753 and 8720 units ,? you have to be VERY CAUTIOUS when you desolder and resolder the EEPROM , the flux from the soldering can create problems that can manifest themselves in
different ways including corrupted cal contants ( well not really corrupted but the CPU reads wrong values ) after you remove the prom you have to clean the board area and the part from all flux residue , I did this using 99% IPA and surgical swabs ,? once
you solder the part back in remove any flux as well
?
?
On Sunday, March 6, 2022, 10:41:30 PM EST, Lothar baier <lothar@...> wrote:
Did you desolder the EEPROM ?
?
?
I have the EEPROM Binaries copied and stored for this VNA. I am leaning towards the fact this is not a corruption of EEPROM Cal Constants.
Based on my recent findings, the switch is looking more like the culprit here.
On Sunday, March 6, 2022, 06:44:14 PM EST, Lothar baier <lothar@...>
wrote:
EEPROMs sometimes get corrupted which is why agilent provided a backup disk
?
The problem with the EEPROM is that if it goes bad you cant just order one from mouser or digi key and pop it in
, the agilent part is preprogrammed with certain information if you put in a blank the CPU wont boot
?
?
Actually the uncorrected response looks almost better than what I had been running. I do have a backup of the EEPROM
which I took from the instrument prior to my repair attempts. I am going to try and load this, see if that helps. I was looking at the test you describe, and quickly realized a power meter with GPIB interface may have to happen sooner than I had planned.?
Wonder how they all of a sudden became corrupted?? Ohh well, I think this troubleshooting is fun so I cant complain
too much.
On Sunday, March 6, 2022, 01:38:34 PM EST, Lothar baier <lothar@...>
wrote:
If this is the case than there is a chance that your correction constants are corrupted , you pretty much need to run TEST 53 which re-establishes the sampler correction constants BUT
in order to run this test you need a power meter and sensor there is no way around it unless you have the backup disk for the EEPROM which originally was supplied with the unit !
?
If you get a meter and run the test I recommend you run tests 44/45/46/48 first , don¡¯t forget to move the write protect jumper on the CPU board and let the unit warm up about 1hr before
running the tests , once those tests are done run 53 and then get a floppy and back up your corrections just in case
?
Sent from
for Windows
?
New Hint - Last night I saw a mention of turning off the Sampler Cor Factors OFF.? I just did this, and now the
problem is gone the VNA passes Port 1 Test at least 10 times now..? Wow, it could be a sampler, or a sampler alignment problem.?
On Sunday, March 6, 2022, 11:10:57 AM EST, Rich Miller via groups.io <av8torrich@...>
wrote:
That was a good warning!? Yes my A5 Sampler is a part number 5087-7001. My donor unit only has P/N 5087-7000. The
donor unit was an OPT11, which may explain the difference in Samplers. I would not have caught that one right away!
On Sunday, March 6, 2022, 11:04:59 AM EST, Lothar baier <lothar@...>
wrote:
A few comments on this , I have hardly ever seen any issues with the signal processing past the sampler , as far as changing the sampler keep in mind that the R and A,B channels use
a different part number so pay attention ,? also I want to point out that changing the sampler will require you to run the sampler adjustment tests
?
Sent from
for Windows
?
?
I checked all the cables around A5 which is the Sampler attached to Port 1. These all look good physically. I am unable to change the response by moving them or shaking them. Next,
I will remove the A5 Sampler and replace it. After looking at block diagram provided in the service manual (not great for troubleshooting), I will not attempt the removal/replacement of the A11 board. It seems to me, that if there was a problem beyond the
sampler, I would expect to see an alarming result on Port 2 if the A11 board were bad. If I am wrong on this, please advise.?
I think you can just move the cables that connect the couplers to the samplers without having to move hardware around.
?
?
I think I confused everyone, sorry!?
What I have done now is swapped both couplers. The problem stays does not follow. It is something in the Port 1 line beyond the coupler. I do have extra samplers here, I suppose no
harm in trying it.?
I should note, I am now seeing these wild results below 3GHz, but nowhere as bad as I see it above 3GHz. 4 - 5 GHz is where it's most consistently bad.
Let me take a look at the service manual, and determine which sampler I need to swap. Would be nice if that were the issue!
Agreed, but does the switch ¡®know¡¯ about the band change? That¡¯s what is curious here.
?
¸é¾±³¦³ó¡You said that the problem follows. Do you mean that the problem still occurs with the port 2 coupler? So it did not follow the coupler, but remained on the port 1 side.
?
I wonder if it is a sampler problem instead. Maybe a bad cable from the coupler to the sampler, or from the sampler to the A/D?
?
There is also a clue in that it stays OK when first turned on, until it is actually switched.
?
?
You are correct about the source switching above 3GHz BUT nothing changes at the source if you switch between S11 and S22 , if you have a source issue than its almost always symmetrical
which means it appears on S11 and S22 .
The Switch is a solid state switch that is a cascade of multiple cells to increase isolation , each branch has ESD protection diodes which are supposed to protect the switch cells from
ESD if those diodes get damaged the effects appear most significant at higher frequencies than at lower ones first S11 and/or S22 of the switch also degrades which affects the source operation asymmetrical !
?
Sent from
for Windows
?
?
¸é¾±³¦³ó¡
?
From your picture, the problem is above 3 GHz. The Source oscillator does a major band switch at about 3 GHz, which suggests that might be where the problem is. The SS switch ¡®could¡¯
be erratic, but it just doesn¡¯t seem likely, especially since you are not seeing any power level changes.
?
When you power up and just let it run for a while without changing anything, does it stay good?
?
If you put a through between ports 1 and 2, and look at S21and S12, what happens?
?
Also, does S22 behave properly? From your original post, I believe it must be OK, but I¡¯m double checking that conclusion.
?
One caution about cables. They can build up static charge between the center and outer, which dumps into the ports when? connected. Usually only a concern with long cables, but¡
?
?
I just got done measuring the power out at Port 1. To my surprise, I cannot detect any significant power output fulgurations. I have tried 3GHz, 3.707GHz (worse offender in S11 Response),
4GHz and 5GHz.? They all look very stable, when switching back and forth. I am using a power meter which is not rated for accuracy at 3 - 6 GHz, but I can set the output to 0dBm look at the measurement, and switch back and forth and look for any relative change.
It is not fluctuating whatsoever.?
I also checked 3GHz with my MDO, and that is a accurate power measurement. Now I am scratching my head again.
On Saturday, March 5, 2022, 07:16:43 PM EST, Lothar baier <lothar@...> wrote:
Try to find a cheap working 85047A ,? it uses the same inerts as the 8753D/E/ES including the solid state switch and you get 2 couplers as added bonus .
?
Take one word of caution though ,? when handling the switch use a ESD mat and wriststrap or at least put some esd caps or terminations on the SMA connectors ,? those suckers are really
sensitive to ESD
?
Sent from
for Windows
?
Yes, I am leaning this way (switch being the issue). I think it could also be that I was running a bunch of two port calibrations using a Agilent 85093 this morning (testing eCal out).
I had some averaging on, 1601 Points, and 300KHz of IF B/W Set. Needless to say there was a lot of switching going on today. Let me check the power output now, and then flip it over and pray that one of the connectors on the switch is loose.
Looks like the parts hunt is likely on!
On Saturday, March 5, 2022, 06:55:27 PM EST, Lothar baier <lothar@...> wrote:
The reversing switches are notorious for failures especially if a unit is used in a environment that is not ESD protected .? what you can also do and you have to be cautious if you
do this is to remove the switch , remove the hardline from the switch to coupler port 1 and connect the coupler directly to the source , if your trace is consistent then you definatly have a bad switch !
?
?
Sent from
for Windows
?
?
It just started failing Port 1 Test. I was doing this just as this email came in. It is failing it consistently. I do not have a power meter, but I can take a look with my Spectrum
Analyzer to see if there are power fluctuations at 3GHz. This problem seems to focus at frequencies higher than 3GHz.?
Once I have a look at the output power, I will open and have a look. This is a new problem, as I never saw this before.
On Saturday, March 5, 2022, 06:38:56 PM EST, Lothar baier <lothar@...> wrote:
If the S11response changes when you switch back and forth than this would point towards a repeatability issue with the solid state switch the fact that this changes with temperature
would also point towards this direction , so here are a few things to try :
?
-
Connect a short to port 1 ,? go to the service menu there is a op check for port 1 and 2 , run the op check several time first for port 1 then port 2 then go back to port 1 and so on , see if it
passes consistently or if there is a failure
-
If you have a power meter connect it to port 1 , change the analyzer to CW and a frequency close to the upper end (3 or 6GHz), set power to 0dBm and then switch back and forth between S11 and S22
measurement and see if the power changes
-
Open the unit and check the hardline connetions , make sure the nuts are tight
?
Sent from
for Windows
?
I have an issue with S11 (Port 1). It showed it's head now that I am running two port calibrations, thus I have not detected it just yet. My hunch is I have a problem with the switch,
but not the Bridge. Here is what I am observing:
1. Upon boot up I get a good/normal S11 RL Response into a good 50 Ohm Load. This Condition will persist until I do one of the steps below.
2. If I switch to Port 2 (i.e. S22 Measurement), and then back to Port 1 the S11 Response has gone nutty (See Attached Picture).?
3. I have noticed, that if I depress the S11 MEAS multiple times, the S11 RL Response returns to normal.?
4. If you look at the enclosed picture, I have captured what I refer to as a normal S11 response, which I saved to memory (Smooth), and what it looks like after pressing S11 again (Wild
variations in response).
I have noticed there is a stark difference between instrument, cold and warm on S11. I do not however see this behavior our of any of the Port 2 Measurements regardless of temperature.
Is there anywhere else this could be coming from other than the Solid State Switch, or am I way off base in my thinking here??
|
