E4407B Tracking Generator Repair

[smgvbest]

I put the TG back together and hooked it back up.  Setup for zero span starting at 10Mhz @ -10dB, enable the TG and all was good.   went up to about 1.5Ghz in 100Mhz steps and at 1.5Ghz it went unlevel.
I attached the U2000H to it and the actual reading at close, not perfect but close.  the U2000H is way over due for a CAL. (see attached images)

so yes, TG goes unlevel initially only above 1.5Ghz but longer it is on the frequency keeps getting lower.  it's now unlevel at 1.29Ghz.

[Jay_Diddy_B]

Hi,

The U2000H Power meter will be very useful when tracking this problem down.



The output amplitude of the TG is proportional to the product of the signal levels at A and B.
The signal level at A should be fixed at +16dBm.
The signal level at B should vary depending on the setting of the TG output level. It is the 'fine' control. The coarse steps come from the step attenuator.

Can you use the U2000H and measure the signal levels at A and B?

Jay_Diddy_B

[analogRF]

Hi,

The U2000H Power meter will be very useful when tracking this problem down.

(Attachment Link)

The output amplitude of the TG is proportional to the product of the signal levels at A and B.
The signal level at A should be fixed at +16dBm.
The signal level at B should vary depending on the setting of the TG output level. It is the 'fine' control. The coarse steps come from the step attenuator.

Can you use the U2000H and measure the signal levels at A and B?

Jay_Diddy_B

looking at the circuit, I don't think the final amplitude is controlled by the amplitude of 600MHz. Rather I think it is controlled by the Hittite variable attenuator just before the main mixer. The 600MHz goes to a mixer directly and then through quite a bit of processing including what I think is a frequency doubler and then reaches the variable attenuator just before the mixer. Maybe  -4.5 to -11dBm only indicates the acceptable range for 600MHz? or could it be that both the 600MHz amplitude and the Hittite chip together control the RF amplitude?

What I still dont quite get is that what they are mixing with the 600MHz? It looks like a free running oscillator!? with no locking?!
They need to create a 3.92GHz out of the 600MHz but it should be a very stable and locked signal...i cannot see how it is....

[analogRF]

I just traced back the 600MHz from A8A1A1 and as I expected there is no amplitude control/variability on this signal. -4 to -11dBm probably shows the acceptable range. Actually on the schematic where the 600MHz to BITG is generated only says -7dBm

The fine tune is done by the Hittite variable attenuator as explained before.

still I dont know what kind of signal they are mixing with the 600MHz and if it is not a locked signal then it makes little sense... so I may be still missing something in that regard

[analogRF]

I put the TG back together and hooked it back up.  Setup for zero span starting at 10Mhz @ -10dB, enable the TG and all was good.   went up to about 1.5Ghz in 100Mhz steps and at 1.5Ghz it went unlevel.
I attached the U2000H to it and the actual reading at close, not perfect but close.  the U2000H is way over due for a CAL. (see attached images)

so yes, TG goes unlevel initially only above 1.5Ghz but longer it is on the frequency keeps getting lower.  it's now unlevel at 1.29Ghz.

mmm...so the unlevel is frequency dependent AND perhaps temperature dependent??
since it looks like having to do with temperature as well, see if you can locate any overheated component on the BITG board
and also check the supply voltages on that board maybe...

still my main suspicion is the output section of BITG including and specially that IGG5 hybrid. However, it would be helpful if you
could check the LO input and 600MHz input to the BITG at a frequency that level fails (zero span) just to make sure those inputs are ok although most probably they are. power meter should be enough to do this.

[smgvbest]

Hi,

The U2000H Power meter will be very useful when tracking this problem down.

(Attachment Link)

The output amplitude of the TG is proportional to the product of the signal levels at A and B.
The signal level at A should be fixed at +16dBm.
The signal level at B should vary depending on the setting of the TG output level. It is the 'fine' control. The coarse steps come from the step attenuator.

Can you use the U2000H and measure the signal levels at A and B?

Jay_Diddy_B

looking at the circuit, I don't think the final amplitude is controlled by the amplitude of 600MHz. Rather I think it is controlled by the Hittite variable attenuator just before the main mixer. The 600MHz goes to a mixer directly and then through quite a bit of processing including what I think is a frequency doubler and then reaches the variable attenuator just before the mixer. Maybe  -4.5 to -11dBm only indicates the acceptable range for 600MHz? or could it be that both the 600MHz amplitude and the Hittite chip together control the RF amplitude?

What I still dont quite get is that what they are mixing with the 600MHz? It looks like a free running oscillator!? with no locking?!
They need to create a 3.92GHz out of the 600MHz but it should be a very stable and locked signal...i cannot see how it is....

On the other side of the A2A2 board is a component marked 40 / 0410-2110 / 9251   is 0410-2110 a freq range?
see pics

would the 600Mhz mix with that, then later mix with the LO

There is a TUNE signal coming in on J1 from the control board

[suj]

On the other side of the A2A2 board is a component marked 40 / 0410-2110 / 9251   is 0410-2110 a freq range?

Maybe it's a heated quartz resonator? Or another type of oscillator. Maybe the mixer mixes the 6th harmonic of the 600 MHz signal with the signal of this oscillator? Then strong filtering. In my opinion this part of the system must provide a signal with a constant frequency of approximately 3.92 GHz. The 6th harmonic of 600 MHz is 3.6 GHz. 320 MHz is missing to 3.92 GHz. A lot for a quartz resonator ...
See the service manual for the 1.5 GHz tracking generator block diagram. Some 800 MHz oscillator is used there. Maybe it's something similar?

EDIT.
It's a bit of a mystery to me. The mixer is up to 1 GHz. I can't see the PLL closing anywhere. The 600 MHz signal is synthesized, but for this oscillator (if it's an oscillator) I don't see dividers or phase detector.

[smgvbest]

Something though is getting a tuning signal from the A2A1 board as well as osc-pwr.  So there is a oscillator there somewhere and it gets tuned

I’m going to have to get a SMB connector as well as 10 pin extension cables made up before I can do some measurements.  The boards fir the A2A2 are stacked tightly.  I do have one of the thermo cams for my iPhone but need to see the components. 

[KE5FX]

That looks like a big inductor, usually something you'd see in an HP switching supply.  0410-2110 is the HP part #.  Odd that it has polarity markings, though.

[smgvbest]

And it’s got 3 terminals

[analogRF]

That looks like a big inductor, usually something you'd see in an HP switching supply.  0410-2110 is the HP part #.  Odd that it has polarity markings, though.

that's why it is not an inductor. and it has 4 terminals by the way not three, two of them go to ground.

it should be easy to trace the pins and see where they are connected to and where the supply lines are

[smgvbest]

Oh. I missed that center one.  Thanks for pointing that out

[suj]

The element marked LF3 is probably 850 MHz LPF from Mini-Circuits (LFTC-850). One of the possibilities is then to multiplicate x6 frequency behind the LPF (maybe x2 + x3 - is the black element in front of the interdigital filter a diode?). If we assume that the MC ADE-2ASK mixer has a maximum frequency 1 GHz, the BPF should be 653.56 MHz (if the multiplication is x6). Then this signal would be created by mixing 600 MHz and a free-running (!!!) 53.56 MHz oscillator. If the multiplication behind LF3 is x7 (rather cumbersome), the BPF is 560.2 MHz and the oscillator is 39.8 MHz. Strange...

[analogRF]

Sandra, can you post close up pictures of the designated components in this photo? their markings are not readable

[analogRF]

the RLC Elec. filter P/N 9135-0389 is a 783MHz bandpass filter. 784 x 5=3.92GHz 
so there must be a multiplication by 5 (picking up 5th harmonic) before reaching the main golden mixer


however, the mystery remains how this frequency is locked if at all. In all low frequency TGs that I have repaired before (HP and Anritsu and Advantest) there is always a VCO generating a fixed frequency which is the start frequency of LO and it is always locked to a stable reference.


i was thinking that maybe they used that ADE-2 mixer as a phase detector (hard driving of RF and LO ports) but with above information, it is not possible....

[analogRF]

I think I was wrong about that black mystery component at the back of pcb. It has in fact 6 pins (see the pic)

still no idea what that is but the +/- pins are Vcc and GND.  But now it is beginning to look like a VCO. I think the control voltage is probably the
single pin on the top that connects to the big through hole resistor

Sandra, can you post a close up photo of the rectangular area that I designated on the picture

[smgvbest]

I think I was wrong about that black mystery component at the back of pcb. It has in fact 6 pins (see the pic)

still no idea what that is but the +/- pins are Vcc and GND.  But now it is beginning to look like a VCO. I think the control voltage is probably the
single pin on the top that connects to the big through hole resistor

Sandra, can you post a close up photo of the rectangular area that I designated on the picture

I'm posting all the close-ups I did but those 2 components I think will need better photo's than these
i think the close-up of those 6 pins may be good if not let me know and tomorrow (its bed time) i'll open the BITG back up and get better photos

[smgvbest]

More

[analogRF]

the markings on those components are not visible

[smgvbest]

the markings on those components are not visible

Didnt think they where.  i'll work on getting the BITG torn down again for pics and post soon as possible

[analogRF]

Something though is getting a tuning signal from the A2A1 board as well as osc-pwr.  So there is a oscillator there somewhere and it gets tuned

I’m going to have to get a SMB connector as well as 10 pin extension cables made up before I can do some measurements.  The boards fir the A2A2 are stacked tightly.  I do have one of the thermo cams for my iPhone but need to see the components.

osc_pwr just turns the TG on/off

the TUNE voltage (pin 5) comes from a SPI DAC on A2A1 and is 0 to 12V. It must be for tuning a VCO on BITG so I think that dark cylinder is indeed a VCO creating 1384MHz and it is fine tuned by this TUNE signal perhaps during the calibration process for TG.
Is there a frequency calibration process for TG in this instrument?

It does seem there is no phase/freq lock mechanism on this TG board and it is just tuned by A2A1 and then it is let go. That VCO is probably very stable... there is no feedback from the BITG to A2A1 board to give information about frequency accuracy, so that TUNE voltage is set just once during some calibration routine i guess ... It seems the 600MHz is there just to make up the math 1.384-0.600=3.92/5
why didnt they make that VCO just to be 784MHz then?

there is also a PWR_LVL (pin 8  ) coming from another SPI DAC on A2A1 (0 to -11V) and I think that signal controls the HMC346 variable attenuator

can you trace that TUNE signal from the connector to a pin on the BITG RF board?

[smgvbest]

Got it apart and took some pics
Hope they help

[smgvbest]

Jury Rigged back together got a FLIR pic as best I could in unlevel state

[smgvbest]

There is an align TG in the SYSTEM/ALIGNMENTS/ALIGN NOW menu

Yes I can trace the TUNE signal.  BRB

[smgvbest]

Tracing it so far. It does not go directly to the RF side of the A2A2 board
Goes thru a analog switch which switches between a fet op amp and J3 which is not connected,  then to Q1 which I haven’t been able to find yet.  It’s a 6 pin can.   See Attached