Agilent 8714ES repair succeed

[shadow.dark]

I find that  People don't like reading large text. Want to skip the maintenance  process and know the result directly.
I know it saves time. ( I enjoy the thinking process.it's fun. Sometimes like a detective.)
So I go straight to the juice.

There are two really fatal failures. deadly failures.
1.A4 source board has NO source signal output
2.A5 receiver A   R    R*   trace are always below -10dB

Causes of A4 board failure is U80  Located on the back of the PCB board.
It is a general purpose amplifier. The model is MSA-0386  marked A03. Amplifier input is open circuit.

Receiver failures are much more dramatic. There is a capacitor short circuit in the + 5V circuit that powers the PCM-78P AD converter.
After I simply removed the capacitor, The miracle happened.


I will continue to revise this post to clearly describe the maintenance process.

The Following Is What I Wrote Before.


Recently I got a faulty Agilent 8714ES.
It cost me 3,500 chinese yuan. It's about 3500/6.8 = 514 dollars.
It came from a bankrupt antenna factory. (Many factories are closed now in China)
But I'm worried that it's made up of several faulty boards of different instruments.
In fact, my worries are not unreasonable.
A label on the instrument says "Receiver failure". 
But now the instrument self-test shows the bus communication fault.
Over the next period of time, I will try to repair this 8714es.
The keysightsite does not provide a detailed service manual.
So I know that such maintenance will be difficult.
But I'm not totally blind either.
Because I had an 8714et, and it works normal.
These two instruments are basically the same.So I still have a reference.

It may take a long time, because I still have a two-month-old baby to look after.

[TheSteve]

The price was right - I hope you can get it going. Do check the power supplies before doing anything else of course.

[shadow.dark]

Agilent Technologies Service Guide
8712ET/ES and 8714ET/ES
RF Network Analyzers

page4-14

Backplane Bus Errors:
Reference clock not toggling.
Unable to gain control of DSP bus.
Cannot perform backplane bus tests.
Access error: wrote <num>, read <num>,
right shift=<num>
Source board: Failed self-test.
Receiver board: Failed self-test.
Test reads version numbers from fractional-N/ref, source,
and receiver boards.
Checks for 5 MHz clock from fractional-N/reference board.
Make sure all boards are pushed in and making good contact
with backplane.
Check 10 MHz output from A3J3. If not found, see
“Troubleshooting the Fractional-N/ Reference Board (A3)” on
page 4-30. Replace CPU board or check U807, U810, and
U811.

[TK]

Backplane error is common on these devices... make sure the boards are plugged all the way in, push them hard, make sure they have screws on the back or the BNC connectors have the nuts on them attached to the back chassis.

[shadow.dark]

After a full day of disassembly, cleaning, inspection and installation，The instrument has passed the self-test.
Special thanks to TK.

Now, the real problem has come to light.
As the label mentioned earlier says, the receiver is out of order.

But I can't do the next check step.
Because many panel keys failed. There is no way to follow the service guide steps for testing. Unless the buttons are repaired first.
I ordered conductive resin. I hope they can be delivered soon.

I found some fibers in the socket of A5 Receiver Assembly, which may be the cause of bus failure.

[shadow.dark]

CPU board, it reminds me of my first computer, especially the SIMM RAM  module.
The good old days.
 It was 32 MB in size,such a large capacity.
When I was in my teens,the price of this memory module alone was going to cost me an arm and a leg.

[shadow.dark]

A5 Receiver Assembly This will be my next focus.

[shadow.dark]

That's how I deal with rubber buttons that don't work.
Coating Conductive Glue on Contactors.

[coromonadalix]

You have this for keypad repairs

https://www.ebay.com/itm/262265110349

[shadow.dark]

You have this for keypad repairs

https://www.ebay.com/itm/262265110349

If the recurrence of button problems drives me crazy.
 I will think of another way.
I can find a SMT patch button  MITSUMI SOP-114HST thin film touch switch It size is 6*6*0.5mm.
It fits the existing space on PCB board.

The quality of Japanese products should still be trustworthy.

I apologize for the watermarking on the picture.  Fortunately, it is in Chinese, and should not be regard as commercial advertising.

[shadow.dark]

After the keyboard returns to normal, follow the steps of service guide for further inspection.

A, B, B*, R, R* are all below - 10dB. OK. This sign points to the A4 source board.
 My old HP oscilloscope confirms this. When I probed at the J4 port of the A4 board(source port), there was only noise.

Note that a lower CW frequency must be specified when measuring.
When you input a 3GHz signal to a 100MHz oscilloscope, you probably won't see much either.

Meanwhile, the A4J1 receiver LO is completely normal.

[shadow.dark]

The primary suspect, of course, is the power amplifier module. 1GG3-4202
I can't find any information about it at all. Only some forums post said that the device is frequently malfunctioning.
I can find this module from some local second-hand device store.
The price is 280 RMB 280/7 = 40 US dollars (yes, 6.8 turned into 7 in just a few days.I forbore to comment on this)
But the problem is, no one can make a commitment that the functions of the devices he sells are normal.
It's kind of like entering a casino.

[shadow.dark]

Well, maybe we shouldn't rush to the casino.
After all, anything in the red box could be the culprit.
Thanks to agilent, block diagrams are much better than nothing.
Here I have a problem that bothers all the maintainers.
I didn't have the extendtion cable.
So I took the most primitive technique.Connect the points to be tested and the oscilloscope with a thin wire.
When the PCB is inserted back into the chassis,test can be done through the thin wires.At least when the frequency is low.

The blue circle marked the test point I chose. Guess what, the signal is normal.

[Zucca]

Thanks for reporting this shadow.dark.

It is a pleasure to read, please continue to update this thread.

 

[shadow.dark]

I'll update it, but slowly. At least one Chinese will keep his promise.

OK, since we've found a clue, we'll follow it. The signal passes through MSG-71008SPDT GaAs MMIC Switch
and dives into the PCB board through a VIA.(red arrow marking)and never appeared again.

To find out where it is going, I removed the shield box on the other side of the PCB board and found an amplifier I hadn't noticed before.

The circuit on the back of the PCB board should be the 2GHz LPF in the functional block diagram.

[shadow.dark]

Later things are relatively simple, simple DMM test can tell U80 is  input port open circuit.
U79 on the LO side is twin brother of U80.If you're not sure, compare them.
Maybe that's what caused  A4 board failure.
I am familiar with this device.It is a general purpose amplifier. model MSA-0386  marked A03.

So I immediately placed an order with the retailer in Shenzhen.

[shadow.dark]

Of course, friends who have seen the end of the story all know that the A4 board has been repaired in this way.
And the receiver failures remain.（That's the downside of telling the story ahead of time.）

But at that time, When the VNA was not completely restored, which still made me a little panicky.
Because it confirms one of my conjectures. This instrument is made up of faulty circuit boards from many different devices.

This is a very common phenomenon in China.Factory purchased expensive instruments but refused to pay for repair them.
This shows the poor management of the factory.（Note that I don't use the word “company”.）
This situation forces operators to repair things themselves.

If an instrument breaks down, its components will be used to repair others.
As a result, all the damaged circuit boards are concentrated on one .

[shadow.dark]

After a few days of depression.（This is mainly because I sleep too little. Baby often cries like a buzzer at night.）

I plucked up my courage to go on.
After the A4 board source signal is restored. B, B* trace is back near - 10dBm. A, R, R* trace are still far less than - 10dB (- 60 - 180).

I am sorry No photos of the trace taken.

I look at the block diagram of the A5 receiver. I don't know where to start.

[shadow.dark]

But wait a minute. Why are there only two ADCs?
Shouldn't there be three ADCs for A, B and R channels?
Frankly speaking, for VNA, I just could use it, but I don't understand the mathematical principles.
So I still don't understand why A and R share one ADC Up to now.
Is it just because the 8714 series is economic instrument?I would be very grateful if anyone could tell me why.

But since A and R channels have similar phenomena，I have to wonder if there is a problem with the common part of the A and R channels（ the ADC ）

There is another reason prompted me to look up ADC.
The A and R tracks are too straight.If you turn off normalize，They are all lines without any turbulence.
Even when I adjusted scale to 0.1 dB per grid.
I can't believe there is no any noise in the receiving channel that can affect a 16-bit ADC.
At least one to two bits of the lowest bit will change. But in fact, none.

[TK]

If both ADCs are the same, swap them and try if the traces change

[shadow.dark]

Swap them is a good idea, but I have to consider the risk of desoldering 28-foot DIP through-hole components.
You know, maybe you can buy components from other sources, but no one can provide PCB except Agilent.

This part of the circuit has some interesting features, it mainly uses common commercial components.
So, the protection of PCB is the first consideration here.

Of course, if you have a whole set of HAKKO Rework Stations, you don't have to think too much about it.

[shadow.dark]

Because ADC PCM78 has a twin brother (sister? I don't know if you think the components are male or female).
The simplest way to check is to measure the resistance of each pin to the ground with DMM and compare it with his brother.
As I said, I didn't have the extention cable.Otherwise, the waveform when the chip working is the most intuitive.

I'm the lucky dog.
Soon I found that pcm78 PIN 23 short circuit to ground.
I cut off the PIN immediately to determine if the chip was short-circuited inside.Maybe I'm too excited.the chip is OK.

After a reverse analysis of this part, I found all the devices that might be in trouble. I marked it with a red arrow.

[shadow.dark]

OK, then the story comes to it's end.
 I checked almost every device(lift the pin up) until I found a short circuit capacitor at the top of the picture.
(I've removed it before take the photo.)
Compared with integrated circuits, the probability of capacitor damage is very low. But it happened.

You know, repairs often depend on luck.
 If a fault occurs in a complex PLL circuit, or a device down and you can't find where to buy.
But I was lucky this time.
 Thank you.

            This is the last post.

[wingel]

I just bought an Agilent 8714ET which was delivered yesterday.  It turned out to have the exact same problem.  The symptom was that measurements made on the R or A narrowbands inputs showed up as a flat line at -180dBm.  I had narrowed down the issue to some problem with the R/A ADC myself, but did not look forward to troubleshooting that.

And then I found your post.  I checked the capacitor feeding pin 23 on the R/A ADC and it was also shorted (~3 Ohms compared to 1kOhm on the other ADC).  Replacing the capacitor made the measurements work properly.

It's quite interesting that two 8714ET/ES units have failed in exactly the same way.  Something must be putting a lot of stress on that capacitor.

Anyway, a big thank you for doing all that hard work.

[shadow.dark]

I'm very glad to hear that your instrument is back to normal. I recently read an article that ceramic capacitors close to the mounting screw holes are prone to crack and short circuit failure due to mechanical stress. Maybe this is the reason why the two instruments have the same fault.

I'm sorry I didn't see your information for the first time, because my family and I live in Wuhan. Where all this chaos started. So we had a chaotic year. I forget if I mentioned that.

Since we have the same instruments, maybe we have the same interests. May I venture to ask if you have any interesting projects? My hp8714es has been idle for too long. I've thought about making some crystal filters, but I haven't done it yet.