Aeroflex IFR-2398 / calibration / repair help

[andy_silicon]

Does anyone have the calibration software ?
Alternatively a dump of the firmware ?
Or possibly knowledge of the Gpib commands involved in the calibration procedure ?

I have an EBay score IFR-2398 which is mostly okay - except it fails the Level Cal.

Attenuator looks fine , 40Mhz cal signal shows at -41dBm ( should be -30 ).
From my good signal source it does look a little all over the place for level - so it may be that the level calibration is required ( looking at the service manual )

Any guidance / theories / previous experience appreciated.

I have access to a decent signal source ( Agilent E4437B ) , and a pretty accurate VSA (E4406A) , a few scopes and general test equipment to aid alignment/repair.

Note: I asked aeroflex - they are only interested in performing the calibration themselves. ( for about 450 quid )
I ask you - where's the fun in that !

Equipment porn by request ( it has hundreds of screws and is a pain to disassemble the RF section )

[andy_silicon]

Okay - so I guess I'm on my own.

Unit disassembled - cpu board removed.
Flash was socketed ( clamshell - psop device ). Cpu is intel i386ex embedded device.
Arggh - PA28F400B5T ....
Scrabble through the older adapters , and find an old WinLV programmer.
Woo hoo - flash dumped.

Now how do I find the config/system password.
Thrown the flash image into IDA , worked out that the base of the flash is 0x3f80000.
Ram starts at 0x0000000
Seems to be written around JMS C:Executive

Lots of features in the flash related to calibration but not easy to access from the unit - I guess the work begins now.

[andy_silicon]

Okay - the service manual mentions keying in the password "27329780" to get to the factory setting which allows the calibration tables to be cleared before recalibration.

For the life of me I couldn't get this password to work. ( no enter key! )

So after much reverse engineering I now see the bloody obvious.
( once you see that they used a generic routine for entering numbers it was obvious the termination of input was one of the Hz/Mhz/GHz buttons )

It's the "Hz" key to enter the password.

Sigh....

At least on the way I discovered another password for the system menu "72237908"

Needless to say don't mess with these extra menus unless you wan't to wipe out you calibration.


There do appear to be undocumented gpib commands - the work continues.

[andy_silicon]

It does look like you can perform the various calibrations from the front panel
( once you enable the factory menu with the password.

It's going to be tedious.

I think there must be a way to upload/download the calibration tables via gpib.
Not sure yet how to do that.

[videobruce]

Do you think that password would allow unlocking any options? I had a 2399A that I wished I never sold. The main reason was that damn floppy. 

Good luck on your project, I'm surprised no one stepped up to the plate yet.

[andy_silicon]

The password for the system menu may do that - as it allows enabling / disabling of features ( gpib / tracking generator / pcmcia )

Assuming of course that the hardware is installed.

[andy_silicon]

Perhaps I'm making progress and we have a fault.

The front end RF module takes 10Mhz , and the input RF to produce a 10.7Mhz IF

With the 40Mhz internal CAL signal turned on - the output 10.7Mhz IF should be -30dBm

i have -54.8dBm.

The front end attenuator performs correctly.
i'm thus down 24dBm of gain somewhere.
The block diagram of the RF module shows several amplification stages ... One of them is 24dBm !

Sounds like the culprit - now which 4 legged microwave blob is it!
( ohhhh for a schematic )

[G0HZU]

Have you checked the match at the RF input port?

I downloaded the block diagram and there appears to be a back to back diode based limiter at the input.

Could this have been damaged by a previous owner? It may wreck the input match if it fails as a short and this would cause a lot of signal loss. You could check its operation with a sig gen and a scope or crudely check it with a DMM.

[andy_silicon]

Yes I have.
I disconnected the RF input to the RF module from the front end protection - and no change.
Also the internal 40Mhz calibration signal switches the RF input off.

I'm going to try and attach some pictures - this may fail.
The amp which I suspect to be at issue is actually a 4 legged white blob with the marking 101.

[andy_silicon]

RF comes in top left.
IF exits top right.
10Mhz in the middle.
Spare port for tracking generator.

Following from the RF port.
Attenuator chain ( with the top rx/tx switch to select the 40Mhz calibration signal
Filter
1st LO Mixer
Filter
Amp (16db)
( hmmm the board and the block diagram don't appear to quite match - corrections later )

2nd LO Mixer and amp (24db)
Following the long path along the bottom
The 10.7Mhz bandpass.
Top right is an MRF5812 (8 pin soic ) before exiting top right.


Yes thats a LOT of Screw holes !!!

I will have to make a temporary cover anyway with a hole cutout to adjust the gain if I replace the 24dB amp.
The service manual say 24db gain , P1dB 15dB.
15v 50mA.

Marking 101 - anybody have any ideas?

[andy_silicon]

Ignore my previous ramblings - the block diagram does fit :-)
I'm looking in the wrong place.

The 24db amp is a tiny 4 pin square device bottom right - after the 210Mhz bpf filter ( string of RC along the bottom ) and before the 3rd mixer ( white blob MCL device right hand side )

So the marking is hard to read
028 or 02B

And I confirm that the board matches (more or less ) with the described bias conditions for this amp.
2 parallel 470 ohm resistors feed the bias from the 15v ( via an inductor )
So that gives 235 ohms over 15v - and assuming 3.7v drop across the mmic - thats about right.
( 11.75V = 235R x 50mA )

[andy_silicon]

Could the device for the 24dB amp be an MRF5711 ?
http://pdf.datasheetcatalog.com/datasheet/motorola/MRF571.pdf

The package style is correct , the bias current is correct , and the gain is correct.
The noise figure is a bit better.

[andy_silicon]

Like this - but with an 8 or B instead of an N.

http://www.ebay.co.uk/itm/380067514769

[KJDS]

I don't think I've ever seen a 24dB gain amp fail in such a way that it became a 0dB stage.

More usually when a MMIC or transistor amp fails it ends up with about 10dB of loss, so I wouldn't rely on the current loss as the fault finding mechanism.

The good news is that there is enough signal loss that it would be easy enough to trace the signal through to find out where it's failing.

[andy_silicon]

Seems a bit weird to me as well.

The problem with fault finding is that the RF module doesn't work with the cover removed.

I will have to improvise with some tin foil/copper tape.

[andy_silicon]

Nope - with the cover off - all self tests fail - no YIG
The YIG is in the RF module - and I suspect without the cover it swamps everything.

I tried replacing the cover with TiN Foil :-) , but same result.

I could remove the part and bypass it ( to see if we lose more gain - or it remains the same ) but I may as replace the part.

KJDS - any suggestions/experience in debugging such units ?

I suppose I could use some thin wirewrap to connect the bias network to an external variable resistor ( 200-700ohms) - then I could vary the resistor to see if the device is working as it should.

[KJDS]

Clamp an aluminium sheet with a 1cm hole in it over the housing to act as a lid. Move it so that the hole is positioned over where you want to probe.

Check with a scope that the voltages are ok on each amplifier. A multimeter probe can cause an RF gain stage to oscillate.

Make up a probe with a 470R resistor on the end of some thin coax. This, when probing a 50R system gives you a 20dB passive probe that won't have much effect on the circuit being tested. With care then this can work ok up to about 3GHz. If you bare the end of the coas so it touches the lid then that will give an ok ground path.

A 24dB fault should be easy to locate, it's hunting down the missing 2dB as that sort of circuit is being designed that's the real pain.

You can check most of the circuit if you've got a fast scope, which will just leave the first IF to check, assuming that it's not an LO driver faulting causing high mixer loss.

[andy_silicon]

Well - I have enough appropriate test equipment  - but there is no way to be able to clamp an aluminium sheet in place ( the RF module plugs into a motherboard - and there is no space.

A full disassembly of the chassis would be required ( the module is between a support and another module ).
I may have to prepare an aluminium sheet of the exact size with some of the screw holes in place.
Then solder a 470R in place and feed through a hole - attach the sheet - solder the coax - put the module in place.

Thanks for the 470R tip.

[andy_silicon]

Minor update.

I managed to get the unit apart enough to do some probing of the RF section.
( and shield the Yig enough to get Yig lock )

The calibration signal is verified at -30dBm as it enters the attenuator chain.

By the bottom of the chain it's a couple of dB down.
After the low pass filter and first amp it's back to -30dBm for input to the first mixer ( a minicircuits MCL-1 )

The 1st seems to be somewhere around -10dBm to 0dBm ( difficult to measure with the probe )

After the first mixer I'm having trouble reliably monitoring the cal signal - will try again later.

[andy_silicon]

I'm beginning to suspect a bodge on the input to the 1st mixer.
There is space for a single smd ( probably an R ).
However three components have been soldered to the board in an hf boost filter.
( R in parallel with C and this combination in series with an R )
If the C has gone bad , it might explain the reduced gain and overall HF fall off.
( about 30-40dB down at 2.7G compared to 40Mhz )

[KJDS]

I'm beginning to suspect a bodge on the input to the 1st mixer.
There is space for a single smd ( probably an R ).
However three components have been soldered to the board in an hf boost filter.
( R in parallel with C and this combination in series with an R )
If the C has gone bad , it might explain the reduced gain and overall HF fall off.
( about 30-40dB down at 2.7G compared to 40Mhz )

What value is the R?

It would need to be huge to cause a big drop in gain.

[andy_silicon]

At the moment I can't easily see ( it's a vertical stack ).
I'd rather be sure before disassembling it - so will check some other things first.

What does concern me a little is that the mixers are supposed to be +7dBm mixers - and I don't see that level at the LO .
More digging.

At the moment I'm only sure it's between the 1st and 3rd mixer   

[andy_silicon]

I think there is at least a problem with the 1st amplifier ( after the pre-amp)
It's supposed to have a gain of 16dB , I'm only seeing 10dB
It's a white 4 pin blob marked 101.
Frequency 3.4ghz , 60mA 4V

I can't find the part number as yet.

Also - I can get the unit to pass the level cal by using the factory menu and setting the value for the expected cal signal level.
So - that would at least seem to indicate the reason for the level cal is purely the gain problem.

I think I've also located the undocumented gpib commands for reading and writing the receiver level correction tables.
There is a correction every 10Mhz from 10Mhz to 2750Mhz.

[andy_silicon]

Nearest part I can find is an Agilent ATF-10136
Looks the same - but the gain is potentially a couple of dB lower.

Hmmm.

[KJDS]

Nearest part I can find is an Agilent ATF-10136
Looks the same - but the gain is potentially a couple of dB lower.

Hmmm.

That gain is specified when matched for optimum noise figure. I suspect that it will be higher matched for best return loss.