E4407B Tracking Generator Repair

[analogRF]

basically IGG5 should not need any DC bias, it has a peak detector and should generate a DC proportional to the power level
and that voltage after some amplification and conditioning by opamps or transistors will become ALC_MON

w.r.t. supplies, I think only the 5V rail seems a little low. check for any ripple/noise specially when the error happens and also for current draw.

when I say no DC/BIAS I mean there is no DC of any kind at the input to IGG5.

at its RF input you mean?

how many pins does it have? cannot be just an input and output!

[smgvbest]

basically IGG5 should not need any DC bias, it has a peak detector and should generate a DC proportional to the power level
and that voltage after some amplification and conditioning by opamps or transistors will become ALC_MON

w.r.t. supplies, I think only the 5V rail seems a little low. check for any ripple/noise specially when the error happens and also for current draw.

when I say no DC/BIAS I mean there is no DC of any kind at the input to IGG5.

at its RF input you mean?

how many pins does it have? cannot be just an input and output!

yes at RF in

12 pins total. 3 on each side.  RFin, RFout,  all others go to ground.

Could the part the arrow points to be the detector?

[suj]

...
Could the part the arrow points to be the detector?

You may be right. Input resistive divider, filter capacitors and signal to AD620. On the second input of AD620, setpoint . The error signal at the AD620 output (in the sense of the difference between the actual value and the setpoint) being the ALC signal. You need to check the DC levels at the AD620 inputs and output at different RF signal levels.

[analogRF]

basically IGG5 should not need any DC bias, it has a peak detector and should generate a DC proportional to the power level
and that voltage after some amplification and conditioning by opamps or transistors will become ALC_MON

w.r.t. supplies, I think only the 5V rail seems a little low. check for any ripple/noise specially when the error happens and also for current draw.

when I say no DC/BIAS I mean there is no DC of any kind at the input to IGG5.

at its RF input you mean?

how many pins does it have? cannot be just an input and output!

yes at RF in

12 pins total. 3 on each side.  RFin, RFout,  all others go to ground.

Could the part the arrow points to be the detector?

obviously there is no DC anywhere in the RF path. there shouldn't be.

yes, it is. If IGG5 has no other pin than the RFin/RFout and GND then it is only functioning as RPP.
And by the looks of it, that circuit  you pointed is indeed the peak detector (that must be a schottky diode)
this makes it easier to probe in and also to repair if the problem is in this part.
you need to find out why at -10dBm setting and -2dBm setting the ALC_MON signal is so different
because the TG must have the same -2dBm output amplitude in both cases.

[smgvbest]

4-since the failure moves to lower amplitude as time passes, it is probably affected by heat. Why dont you spray some freeze spray (aim very locally) at different components and see if the error goes away. starting with that IGG5!

Did this.   the only component that had effect using cold spray was the HMC346 it only shifted it down (lower db reading)  but unlevel condition continued
tried the other board on the TG (one with OPAMPS and Analog Switch) , no effect there either

[analogRF]

4-since the failure moves to lower amplitude as time passes, it is probably affected by heat. Why dont you spray some freeze spray (aim very locally) at different components and see if the error goes away. starting with that IGG5!

Did this.   the only component that had effect using cold spray was the HMC346 it only shifted it down (lower db reading)  but unlevel condition continued
tried the other board on the TG (one with OPAMPS and Analog Switch) , no effect there either

did you also try freezing the the capacitor and the schottky diode of the peak detector circuit?

based on the table you provided before, you should have got unlevel error even at -2dBm because the ALC_MON drops below -1V
and according to the schematics of A2A1 this will set the unlevel error. it is puzzling that you didn't get that.

something is happening between -3dBm and -2dBm. Can you try to measure the RF amplitude at these twi levels and see if the
RF actually drops? because the ALC_MON is reporting that it drops quite a bit instead of increasing by 1dB. So check the actual RF amplitude
(assuming IGG5 is ok, you can do it at the SMA port) if the actual RF amplitude does not drop then the problem is in the ALC_MON reporting
and the circuit between the schottky peak detector (and its ceramic cap) and the ALC_MON pin must be checked (most probably the opamp on the top right)
but if the amplitude does indeed drop, then I think the HMC346 is the main suspect.

by the way, is the frequency spot on?
 

[smgvbest]

did you also try freezing the the capacitor and the schottky diode of the peak detector circuit?

Yes,  no effect noticed.

based on the table you provided before, you should have got unlevel error even at -2dBm because the ALC_MON drops below -1V
and according to the schematics of A2A1 this will set the unlevel error. it is puzzling that you didn't get that.

something is happening between -3dBm and -2dBm. Can you try to measure the RF amplitude at these two levels and see if the
RF actually drops? because the ALC_MON is reporting that it drops quite a bit instead of increasing by 1dB. So check the actual RF amplitude
(assuming IGG5 is ok, you can do it at the SMA port) if the actual RF amplitude does not drop then the problem is in the ALC_MON reporting
and the circuit between the schottky peak detector (and its ceramic cap) and the ALC_MON pin must be checked (most probably the opamp on the top right)
but if the amplitude does indeed drop, then I think the HMC346 is the main suspect.

Will do this.
I did trace and verify that indeed the via pointed in the peak detector circuit does go to the input of the AD620 as indicated.
I also did more tracing of the RF board and its 2 10 pin connectors.  i've got most pins traced out. will post image with labels shortly for reference

by the way, is the frequency spot on?

yes it is

[analogRF]

one other thing you can test, if you set the TG ALC to External and dont apply any DC to the Ext ALC
the output should stay fixed at around 0-1dBm I think. At least this is how it worked in 8590 series I remember
I dont know what DC levels can be applied to the EXT ALC to control the TG output level. Maybe it's somewhere in the manual....

[smgvbest]

I've been trying to figure out what's happening at the AD620.
but overall its looking like the Reference voltage is fluctuating and the output of the detector is stable at about .0446vDC.   so the output of the AD620 is also fluctuating.

also the output is not very stable as read at same time as the input to the AD620.   which kind of aligns with the output of the AD620.
So is the AD620 causing the Reference to fluctuate or is the Reference to the AD620 the problem
I need to record the rest of the reading for my table here.  hopefully latter I can do this
also as the unit warms up it more seems to change behavior.   the notable items is the output from the detector seems stable. so this to me so far is pointing to the AD620 or Reference to the AD620

more info to come

Code: [Select]

Attenuation dBm Min dBm Max dBm AVG (A)in
3dB -3.020 1.200 -0.910 0.0447
2dB -1.720 1.220 -0.250 0.0446
1dB -0.940 1.220 0.140 0.0447
0dB -2.840 0.620 -1.110 0.0446
-1dB -0.910 -0.390 -0.650 0.0447
-2dB -1.990 -1.370 -1.680 0.0446
-3dB -3.420 -2.490 -2.955 0.0447
-4dB -4.370 -3.560 -3.965 0.0446
-5dB -5.110 -4.540 -4.825 0.0447
-6dB -6.140 -5.560 -5.850 0.0446
-7dB -6.670 -6.680 -6.675 0.0447
-8dB -7.780 -7.750 -7.765 0.0446
-9dB -9.660 -8.750 -9.205 0.0447
-10dB -10.110 -9.470 -9.790 0.0446


[analogRF]

what are these numbers?

[smgvbest]

what are these numbers?

What I have so far

The Attenuation is the 1st column,  next is the dB min/max values (ie it was not fixed it varied) as measured with the power meter,  the next is the average dB and last is the Peak Detector Input to the AD620
i need to complete this by getting the reference into the AD620 and the output of the AD620,   just sharing the values so far.

 

[analogRF]

what are these numbers?

What I have so far

The Attenuation is the 1st column,  next is the dB min/max values (ie it was not fixed it varied) as measured with the power meter,  the next is the average dB and last is the Peak Detector Input to the AD620
i need to complete this by getting the reference into the AD620 and the output of the AD620,   just sharing the values so far.

 

you mean the amplitude actually fluctuates when sitting there? How? keeps oscillating up and down? That clearly means the ALC loop is not working. It is like having no closed "loop" control of the amplitude which is possible but unlikely from a high quality tracking generator.  If there is indeed a control loop and the amplitude just plays around like that, then the loop is not working (not closed)

It actually starts very early on even at -5dBm or -6dBm.
From the schematics I know that the Unlevel error only happens when ALC_MON goes out side the +/-1V window. So the loop control (if exists) must be closed somewhere on the BITG circuit. Perhaps somewhere on the RF signal path they sample the amplitude and compare with a reference and then control the V1 to HMC346 to keep the amplitude steady. If this loop is broken then the behavior you see is explained.
It is important to figure out the circuit behind generating V1. I think it originates from the PWR_LVL from DAC on A2A1 but that voltage could be just providing the reference value for the control loop.

But another (unlikely) option could be that there is no control loop and they just set the V1 based on some previously calibrated values and let it go  and then only check whether ALC MON is within range or not. again we need to know how V1 is produced

a little bit of reverse engineering is needed  I think

[smgvbest]

you mean the amplitude actually fluctuates when sitting there? How? keeps oscillating up and down? That clearly means the ALC loop is not working. It is like having no closed "loop" control of the amplitude which is possible but unlikely from a high quality tracking generator.  If there is indeed a control loop and the amplitude just plays around like that, then the loop is not working (not closed)

So yes, reading from the power meter (U2000H) i see the levels jumping around and the numbers I show are the Min/Max for each Attenuation level.

It actually starts very early on even at -5dBm or -6dBm.
From the schematics I know that the Unlevel error only happens when ALC_MON goes out side the +/-1V window. So the loop control (if exists) must be closed somewhere on the BITG circuit. Perhaps somewhere on the RF signal path they sample the amplitude and compare with a reference and then control the V1 to HMC346 to keep the amplitude steady. If this loop is broken then the behavior you see is explained.
It is important to figure out the circuit behind generating V1. I think it originates from the PWR_LVL from DAC on A2A1 but that voltage could be just providing the reference value for the control loop.

But another (unlikely) option could be that there is no control loop and they just set the V1 based on some previously calibrated values and let it go  and then only check whether ALC MON is within range or not. again we need to know how V1 is produced

I was planning on having to look at both the V1 which feeds to the HMC346 from the interface board on the TG so may have additional circuitry there  and the how the peak detector circuits are fullly implemented to see if there is more on the interface board.   PWR_LVL appears to go direct to the RF board with no circuit on the interface board.

a little bit of reverse engineering is needed  I think

Been figuring this was coming
Lots of probing ahead I see.

[analogRF]

So yes, reading from the power meter (U2000H) i see the levels jumping around and the numbers I show are the Min/Max for each Attenuation level.

I was planning on having to look at both the V1 which feeds to the HMC346 from the interface board on the TG so may have additional circuitry there  and the how the peak detector circuits are fullly implemented to see if there is more on the interface board.   PWR_LVL appears to go direct to the RF board with no circuit on the interface board.

First thing I would check is the control voltage PWR_LVL coming from A2A1 to see if it has any instability/fluctuation in it. Also that 5V rail which was a bit low and must be checked for noise/ripple too.
If those are good, then there is a ALC loop on BITG which is open thus the amplitude fluctuates. there should be a feedback mechanism from peak detector circuit to generating V1. I think that circuit is on the other TG board. At this point I think lots of things are pointing to that other board. The problem is more likely in there I think. The only components on the RF board that might be of concern now are the HMC346 itself which "most probably" is ok and the peak detector diode and cap which you can easily check. if those check out, I think something is happening on the other board

[smgvbest]

@analogRF

I'm working to trace the circuit out for the PWR_LVL and ALC_MON and others
in the mean time I fired up benchview and took a look at the TG Output at a few levels,   I added a strip chart which lets you see whats going on better than just looking at the MIN/MAX values
I'm posting the -10dBm and 3dBm screen caps

What's interesting to me is the dips in the chart are very regular.

[analogRF]

@analogRF

I'm working to trace the circuit out for the PWR_LVL and ALC_MON and others
in the mean time I fired up benchview and took a look at the TG Output at a few levels,   I added a strip chart which lets you see whats going on better than just looking at the MIN/MAX values
I'm posting the -10dBm and 3dBm screen caps

What's interesting to me is the dips in the chart are very regular.

yeah, these are very interesting and weird. the dips occur almost a little more than 3sec apart regularly but some are very prominent and some
are very tiny which makes it more puzzling.
-10dBm is the same as -2dBm. you should check something like -5dBm or -6dBm and see how flat it turns out.

it would be useful and illuminating if we could have a recording of PWR_LVL, ALC_MON and V1 time correlated with these power graphs.

is it possible that the unit tries to do self alignment of TG every 3 seconds? 3 seconds is normally a lot longer than a normal sweep time of spectrum analyzer when you are using the TG to measure a filter for example, so maybe in an actual measurement these dips occur every let's say 100 sweeps in the dead time at the end of the sweep and if you actually use the TG for measurement you wont see them?! In other words your TG is fine    just a thought..

[smgvbest]

yeah, these are very interesting and weird. the dips occur almost a little more than 3sec apart regularly but some are very prominent and some
are very tiny which makes it more puzzling.
-10dBm is the same as -2dBm. you should check something like -5dBm or -6dBm and see how flat it turns out.

it would be useful and illuminating if we could have a recording of PWR_LVL, ALC_MON and V1 time correlated with these power graphs.

is it possible that the unit tries to do self alignment of TG every 3 seconds? 3 seconds is normally a lot longer than a normal sweep time of spectrum analyzer when you are using the TG to measure a filter for example, so maybe these dips occur at the dead time end of each sweep and if you actually use the TG for measurement you wont see them?! just a thought...I don't think what I just described is correct but you never know...

I had trouble with Benchview crashing on me when saving the screen caps but here's  -8dBm, -7dBm and -6dBm caps

Autoalign is on by default for SA,  I will try turning it off and do these again tomorrow and see if it is related.

far as trying to time correlate them.   would doing that with a oscilloscope work?   I have the Rigol MSO1024 4 channel?  but not sure how I would sync them with the power meter in Benchview

[analogRF]

yeah, these are very interesting and weird. the dips occur almost a little more than 3sec apart regularly but some are very prominent and some
are very tiny which makes it more puzzling.
-10dBm is the same as -2dBm. you should check something like -5dBm or -6dBm and see how flat it turns out.

it would be useful and illuminating if we could have a recording of PWR_LVL, ALC_MON and V1 time correlated with these power graphs.

is it possible that the unit tries to do self alignment of TG every 3 seconds? 3 seconds is normally a lot longer than a normal sweep time of spectrum analyzer when you are using the TG to measure a filter for example, so maybe these dips occur at the dead time end of each sweep and if you actually use the TG for measurement you wont see them?! just a thought...I don't think what I just described is correct but you never know...

I had trouble with Benchview crashing on me when saving the screen caps but here's  -8dBm, -7dBm and -6dBm caps

Autoalign is on by default for SA,  I will try turning it off and do these again tomorrow and see if it is related.

far as trying to time correlate them.   would doing that with a oscilloscope work?   I have the Rigol MSO1024 4 channel?  but not sure how I would sync them with the power meter in Benchview

oh, so they do exist at all levels and always at regular intervals 
Now I am beginning to think that your TG might be ok afterall.... why dont you try to do a practical measurement/sweep with it? like a filter or an amplifier or even a piece of coax cable. let's say in a 3GHs span or whatever else...

still having time correlated measurements of those 4-5 signals will give a better insight. but I admit it's a bit of a challenge to implement that
at the very least you can record those PWR_LVL, ALC_MON, V1 and I'd also add TUNE with your 4 ch scope in roll mode
and when we put them next to these power graphs, if there is any anomaly happening at 3 sec intervals we know they are correlated with power
dips

[analogRF]

just out of curiosity, i cannot find Rigol MSO1024 datasheet anywhere... 

[smgvbest]

just out of curiosity, i cannot find Rigol MSO1024 datasheet anywhere... 

probably because it doesn't exist,  what happens when users take their sleeping meds and try to type

Try MSO1104ZS
https://www.tequipment.net/Rigol/MSO1104Z/Mixed-Signal-Oscilloscopes/

[smgvbest]

Just an update,   have had some things come up in personal life so had to step away from this for a bit.   will be back with soon I hope

[smgvbest]

Looking for help in identifying some components
in PICT0007.jpg   looking to identify all marked ones.  the one I put a D on,  the D is the only marking on the component.   I think this is a dual diode package for the peak detector.

I loose allot of power at the point of this component.   like -25dB.   this is suppling the 786.5hz RF to the Mixer (5x that but I can't measure on my 1.5Ghz SA)
in  IMG_2867-1.jpg

[smgvbest]

these 2 Sot-23's marked 8F I think are 5.1v zeners but 8F can be a NPN transistor as well
both read strange but reading is in-circuit.  one reads 24.7K forward and reverse and other is like 30meg forward and reverse.   I'll get them out of circuit to test but they are connected to V1/V2

[smgvbest]

Here's another area.   all but the one with a ? is identified but  I marked one component that looks damaged?   any thoughts on that?

[analogRF]

Looking for help in identifying some components
in PICT0007.jpg   looking to identify all marked ones.  the one I put a D on,  the D is the only marking on the component.   I think this is a dual diode package for the peak detector.

I loose allot of power at the point of this component.   like -25dB.   this is suppling the 786.5hz RF to the Mixer (5x that but I can't measure on my 1.5Ghz SA)
in  IMG_2867-1.jpg

the white elements are most likely resistors. you can check with multimeter
the "D" element is most likely HMPP-3890 single schottky diode by Agilent/Avago but are you 100% sure about the marking? sometimes it is tricky to read the labels
seems that all 4 pins of it have been used  also for a good temperature compensated peak detector it makes sense to have two diodes in one package but if the marking is only a "D" then I couldn't find anything other than HMPP-3890 which is still applicable...

the element with red dot is what is used to generate 5th harmonic if i remember correctly from previous discussions, so it is probably a diode of some kind.