E7495A 1-Port RF Measurements

[ddcc]

A while back, I acquired a slightly broken E7495A unit, where the spectrum analyzer and signal generator work fine independently, but 1-port insertion loss and return loss measurements aren't right. For example, when performing a sweep from 375 - 376 MHz with a 10dB attenuator + 50 ohm load connected to the source, I get a flat 0dB return loss. Since there's no schematic for this unit, I ended up opening the RF assembly, and reverse engineering out the relevant area of the board.



The source input comes in at J7, and is connected both up through the coupler to the output of the unit, and down through C towards the input path at A. The reflected signal comes back from the coupler at D through E, and again towards the input path at A. Using a SDR, I've verified that the sweep looks correct.




I've also measured both the output signal at C, and the reflected signal at E, and they both seem correct. Unfortunately, there's some aliasing/noise because 375 MHz is the lowest output frequency of the E7495A, but my DS1054Z scope is only rated for 100 MHz, and my passive RP2200 probes are 150 MHz, even with me using the ground spring.




Also, the SPDT switches on this path are switching between the two signals, approximately 72% reflected and 28% output at 130Hz (yellow CH1 at U121, blue CH2 at U9, pin V_CTL). The signal that is not being driven through U9 is switched into a 50 ohm termination.

[ddcc]

By the time both signals are amplified, DC biased, and about to enter U9, they look a bit different.




The immediate output of A before the blocking cap looks fine (blue CH2 is pin V_CTL), but at A it becomes quite weak.




I'm a bit stumped what could be the problem, because from A, the signal is switched onto the receive path, past an attenuator, into the 1st mixer, etc, which all seem fine. However, I don't have much background with RF, so I'm probably missing obvious. For reference, here are the full annotated top and bottom photos.

[kirill_ka]

Did you measure the actual power which goes out of the RF out port? I don't how it could be related to your problem, but my E7495 seems to have some intermittent issue with the attenuator. Sometimes when I adjust the generator settings it starts to output way more power than it should.

[Wallace Gasiewicz]

As I understand the unit, the one port and two port measurements use the noise generator, not the optional CW signal generator. But maybe there is some power setting they share.
I will look at the one port loss on my unit using the 10 dB attenuator and load as you did. 375-376 MHZ. will let you know what I see.

Having said that,
I have had the same problem with power out in CW mode. It is about 20 dBm high. This happens when I first turn it on after it has been off for a day. When rebooting, the power out goes back to normal. The power setting in RF generator is unaffected. However the power out is up about 20 dBm. I have verified this with other instruments.
Page 403 of the 7495 manual indicates different power outputs available in CW mode for 7495 A units and B units. The power out also varies with the Firmware version. Version 6.2 and higher allow more power on the B unit. 18 dBm more power. No additional power out is supposed to be available on the A unit.
I have the new 6.25 firmware and an "A" unit.
The "A" unit is not supposed to go higher than -23dBm, mine sometimes does.
What unit and firmware are you using??
What is the real difference between and A and B unit??? Just firmware?? I do not know.
I suspect there is some incompatibility in using 6.25 in an A unit.
But... it would be nice to have the higher output.
Perhaps there is some string in the 6.25 code that is responsible for setting these values. Perhaps the string or command is not the same in all parts of the 6.25 program, Perhaps there is a command to put in an attenuator that is not always implemented depending on what part of the 6.25 the unit gets its instructions.
Any thoughts?

Wally

[kirill_ka]

One more thing, in the two port insertion loss mode GUI allows to set the level above -23dBm. If you do so, the output is unleveled and doesn't change with the power setting (until you bring it below -23dBm).
This seems to be the expected behaviour. In by case (as I mentioned above) the power level jumps up unexpectedly (yes, ~ +20dB) when I change generator settings from the spectrum analyser screen.

[Wallace Gasiewicz]

Getting back to ddcc

I do not think I am following you entirely
How did you analyze the board? I think you said you put 130 MHz into J7 and terminated output with 50 ohm load.
Signal comes in J7 and then through coupler and out the RF output right? Then into a 50 ohm load.
So the very small signal you are measuring as reflected power is as we would expect, right?
Am I following so far? Problem is somewhere else, I guess.

I did use the insertional loss and reflected power on my unit and found that the reflected power into a good 50 ohm load is about 35 dB. Insertional loss was about 20 dB

Can you measure the actual signal strength of the signal out in the reflected and insertional loss mode? This may tell us something.
My signal strength is about 8 dBm. Measured on an HP VNA.

Kirill:
My 7495 does not put out the higher power CW today. Sometimes it does and sometimes it does not.

[ddcc]

Did you measure the actual power which goes out of the RF out port? I don't how it could be related to your problem, but my E7495 seems to have some intermittent issue with the attenuator. Sometimes when I adjust the generator settings it starts to output way more power than it should.

I can look at the spectrum power measurements on my SDR, but I don't think those are designed to be accurate. The closest I can get is to disable hardware AGC and manually "calibrate" the pre-amplifier gain on the SDR so that the measurement matches the expected power coming out of a known source at some frequency, and hope that the response is relatively flat across the entire spectrum. I only have a 33120A (20 MHz max, which is too low for the SDR), and the reference power output on my E7495A is disconnected since I have the receiver board flipped upside-down, so I'll need to figure something out.

What unit and firmware are you using??
What is the real difference between and A and B unit??? Just firmware?? I do not know.

Also 6.25 on an E7495A. I believe there are some hardware differences between the A and B, e.g. slightly higher spectrum, but otherwise I think the changes are pretty minor. Both units do already have an 0 - 130 dB electronic attenuator connected to the output, it's the 5087-7103 module in the unit. For a simple test, you could just bypass it by directly connecting ATTN IN/OUT together. I can try it, though I'll need to figure out a setup for RF power measurement first.

How did you analyze the board? I think you said you put 130 MHz into J7 and terminated output with 50 ohm load.
Signal comes in J7 and then through coupler and out the RF output right? Then into a 50 ohm load.
So the very small signal you are measuring as reflected power is as we would expect, right?
Am I following so far? Problem is somewhere else, I guess.

I'm using the built-in source on the E7495A, so I'm not injecting anything into J7, it's just connected to the internal RF source board. Yes, the small signal I'm seeing should be the reflected power, I'm just not sure why it's not being measured correctly. I can take a photo of my setup. Measuring the actual signal power output in one-port or return loss modes will be a bit trickier, because I think it's sweeping a bit too fast for my SDR. I'll try to figure something out though.

[ddcc]

Sorry for the delay, got distracted for a bit.

I don't have an accurate power reference, and the 50 MHz reference on my E7495A is disconnected since the receiver board is flipped upside down, so for now, I set the E7495A signal generator to output 375MHz at -23dBm (max power), and adjusted the hardware gain on my LimeSDR accordingly.

Physically, the connection is going from the E7495A source output through a DC block and 10dB attenuator into the RX1 low input on my LimeSDR.

[ddcc]

When I set the unit to measure return loss, from 375 - 376 MHz, I see the following measurements. They suggest that the unit is sweeping at around -60dBm, which seems quite low.

Interestingly enough, when I connected my LimeSDR directly to the rigid coax going into J3, bypassing the attenuator (the attenuator silkscreen is flipped), the measured signal power is still the same. But, when I switch back to the regular signal generator, the signal is around +20dBm stronger at J3 compared to TX output, so the attenuator is engaged in that mode.

[kirill_ka]

When I set the unit to measure return loss, from 375 - 376 MHz, I see the following measurements. They suggest that the unit is sweeping at around -60dBm, which seems quite low.

Hi,
If you still interested, I can measure the output power in this mode.

[ddcc]

Hi,
If you still interested, I can measure the output power in this mode.

Thanks for the offer! Given Wallace's earlier measurements, I believe the problem on my unit is probably in the source board, I just haven't gotten around to opening it and looking inside. Unlike the receiver board, this side of the assembly is more complicated because all the other boards are installed on top (e.g. power meter, GPS board, complex source, etc).

[kirill_ka]

Ah, I missed that. So at 8 dBm it looks like the maximum (unleveled) power.
Removing the power meter and the GPS should be easy. However, note that E7495 will wait forever trying to initialize the power meter, if you disconnect it's ribbons.

[ddcc]

Managed to open up the source board, will probably need some time this weekend to figure out how it works. It was a real pain to disassemble, because the backend board, power meter, and GPS board are all mounted on the top lid, and as you mentioned, the unit will hang at startup if any of the first two boards are disconnected. Because there are numerous connectors that connect at both the top and bottom, I ended up needing to reassemble the unit with the top lid underneath the source/receiver assembly, so that I can access the unit while it is operating. Here's a photo of the top.

[ddcc]

Hmm, I'm wondering if the low measurements I'm seeing are just an artifact of using the SDR to look at a swept signal. In the attached photo, notice how the measurements at J7 on a swept signal from 375 to 376 MHz appear to be around -45 to -60dB. But, when I change the sweep to 375 to 375.0001, the peak is at -10dB. Since I'm using the same return loss measurement mode, doesn't it seem unlikely that there would be an issue with the output power only when sweeping?

Unfortunately, if true, this would mean that I'm back to square one, since the source would be fine and the fault would have to be somewhere on the receiver board?

Also, here's a copy of the annotated top source board.

[ddcc]

I looked around the back of the source board some more, and noticed that connector J1000 feeds into an AD8361 power detector. There does exist a signal path on the receiver board where the attenuated input signal can be fed through J2 (Detector Out or "Diag/AutoRange Cnx") to J1000 (Detector In or "Rxr Auto Range Cnx to Log Det DAC") on the source board, which appears can also be used for power measurements from the output coupler.

Since the source board has a board-to-board connector on the back, I can't easily access it when the unit is powered up, so I ended up testing the AD8361 individually by providing external 5V, connecting J1000 to a signal generator set for 50 Ohm, and probing the linearized DC output of the AD8361. Based on the results, it seems that the AD8361 is a little out-of-spec; it's configured for ground reference mode with an output intercept of 0 but I'm measuring 80mV with the input shorted, and the measured gain is ~6.26 Vdc/Vrms instead of the expected 7.5 Vdc/Vrms. The only other thing is that the rated intercept drift is ~43mV/degC at 3V up to ambient (25 degC). Since my measurements are at 23.6 degC, I'd expect a negative drift of up to -60mV, and not +80mV. But, at least to me, the AD8361 doesn't seem sufficiently out-of-spec to be the root cause?

The only other power detector on the source board is what appears to be a temperature-controlled diode detector coupled to the output signal. You can see the cutout for it to the left of U1000, and the silkscreen label for RT1 on the other side of the source board. I haven't measured its output yet, but I'm hoping it's not the root cause, because I can't identify the detector diode.

[Wallace Gasiewicz]

ddcc and Kirill

Hello again.
I am still following the discussion.
I made some measurements on my 7495
7495 set up 375 MHz to 375.001 Mhz
Measurements on one port insertional loss after calibration

With 50 ohm load on the 7495, Insertional Loss Reading is about 18 dB on 7495
Feeding into Agilent 8935-6380 service monitor Insertional loss is the same and service monitor reads 6.4 dBm power. This unit has a power meter built in.
Almost the same results feeding into HP 8713 VNA.

My conclusion is that the power output of the RF OUT of the 7495 in insertional loss mode is about 6.4 dBm (into 50 ohms)
Let me know if any other measurements from my undisassembled 7495 might be helpful.



Good Luck
Wally

[Wallace Gasiewicz]

Since i privately e mailed Kirill and DCC I thought I would summarize my latest adventure.
Recently my 7495 would not lock no green light. I e mailed Kirill and he suggested that there was too much jitter. He was correct. The PLL would not lock and was wandering a few KHz. I traced the problem down to the 60 MHZ volt controlled TCXO (silver Box about 1x2 cm) on the Backend Board. When I heated it up it locked. I supposed something like the temp control part of the VCTSXO was not working correctly.
I blew the PLL chip trying to rig up some resistors to heat up the unit.

I then replaced the PLL dip chip and purchased another backend board.
The new backend board made the unit lock up on GPS very quickly band and magically enabled all the options with the exception of the CW generator.
So I switched the VCTCXO s and everything works.
The old VCTCXO even started working on the purchased board.
The new backend board is labeled Test Board. and apparently enables ALL the options with the exception of CW. I have not done the code hack on the 6.25 upgrade to enable options. CW 500 and 510 are "Factory Option"
Probably the new Backend Board is a test board for service people the serial number is 00000022.
Perhaps this new board might help DCC since it appears to control lots of things.
Wally

[kirill_ka]

What is the backend board? Is it the one which contains SHARC DSP?
E7495 doesn't autodetect HW options. I didn't investigate where it stores the information about them. They are not present in the FLEXLM license file. It's possible that the HW configuration is stored on the backend board.

I then replaced the PLL dip chip and purchased another backend board.

Do you mean you both fixed the original board and got a replacement one?

[Wallace Gasiewicz]

 I accidentally burned the PLL DIP chip on the original board
I replaced it and then I purchased a replacement board to get a 60 MHZ volt controlled temp compensated oscillator
The replacement board worked in the condition it arrived in except for CW option
So I switched the vctcxo I thought was the problem and the original board works fine.
The thing I thought was a problem (the vctcxo) started to work on the replacement board (I tested the board)
I do not think it was just soldering since I tried to re solder the original vctcxo several times.
So both boards work now, except that I do not trust the purchased board since it had the old vctcxo on it.
The backend board is the first board on the top of the stack next to the RF power meter board
I do not know what a SHARC DSP is, there is a CPU on backend board along with lots of memory
Wally

[kirill_ka]

Ok, thanks. Now I got it.
The CPU on the board is SHARC. It's ADSP 21160M
https://www.ebay.com/itm/HP-Agilent-E7495-63002-Board-Assembly-/271152854216

[Wallace Gasiewicz]

Yea, I bought a similar board.
Now the question is how to enable the CW option on the new board, it works fine on the repaired old board..
This is not a big deal since my primary goal has been accomplished at a reasonable cost.

If you look at the board one of the DIP ICs near the silver 60 MHz osc is the PLL. The output from this was going up and down to change the osc by several KHz. When I heated up the osc with hot air, the thing locked nicely. Both the green light came on and the PLL output settled down, as I would expect. I think that the Temp Compensation of the osc was not cooperating and was making the osc go up and down and the PLL was trying to correct it.
The output from this board is lots of strong harmonics of 60 MHz, I only measured up to 1 GHz but good strong harmonics every 60 MHz.

It is interesting that ALL the 200 options were enabled by this board
The board also changes the serial number of the unit to 00000022


The 500 and 510 CW functions are listed as possible options in option manager and when I try to enable them the unit asks for a KEY.
Maybe the key is the serial #, I did not try that. There is a phone #.
Again,I have not done the option hack on the 6.25 upgrade as I did not want any more options.

All of this is Just Interesting

Wally