|
Hi all,
?
In order to ascertain if the attenuator in my SA is faulty or not, it has been suggested - very sensibly - to attempt a calibration with a fixed 10dB attenuator subbed in place of the variable 0-70dB one in the unit. Great idea in theory. Having removed the bottom case (again) I was just about able to make out the attenuator behind a whole bunch of other RF stuff and a maze of rigid coaxial plumbing. In order to get at the necessary connections it would appear impossible without some kind of service tool which could undo the sma nuts whilst avoiding all the pipe-work. I don't have such a tool. Has anyone attempted this? Would it be any easier (assuming it's even possible) to get access from the PCB side instead?
Thanks,
?
Jinxie
|
Do you have the document
"Troubleshooting and Repair Manual Volume 1 HP 85660B Spectrum Analyzer RF Section"?
You can get it here..
https://xdevs.com/doc/HP_Agilent_Keysight/HP%2085660B%20RF%20Section%20%20Troubleshoot%20and%20Repair.pdf
?Page 77 onwards details the procedure.? I did it and it is not too difficult.
SMAs can be undone with a regular 5/16" spanner but should be re done to the correct torque.?
Ebay has the correct tool for about $30.? It may be a knockoff as other branded tools are over $500.
This is the part number for the unit I got to replace my faulty unit. ?Check the serial number of your RF unit to ensure compatibility.
HP 85660-60304
|
Yes, thanks for that. I was unable to find the details for the removal of A1 (the attenuator) as for some reason the people who compiled that manual listed under A6. Anyway, I have it now. So they say to go through the front panel! Well I'd completely overlooked that possibility. Thanks for the moral support in stating it wasn't a big deal to get the attn unit out and the info on the tools. I did check the model number on mine and it's not the same as yours. Mine's 85660-60121. I gather they can be repaired. I'm just curious as to what they make the resistive wafers out of and whether one can make one's own custom precision fixed resistors from the same stuff. That would be very useful if it's do-able.
|
I looked at attempting a repair and after searching and seeing a few videos I decided against that.
The replacement unit only cost me USD$130 on ebay.? While it was a gamble, it did pay off.
I was pretty sure it was the attenuator as when I stepped through the levels manually the display did not change at certain settings.? The clicking from inside also sounded very different at those settings.
I wish I had taken pics of the procedure...
|
If the attenuator does turn out to be faulty and beyond repair and I can't find a replacement unit off Ebay or whatever, I could always just bypass the faulty attenuator internally, replace the top and bottom cases and just use any high-quality adjustable RF attenuator in series with a 10dB fixed attenuator permanently attached to the RF input at the front panel. So it wouldn't be the end of the world I guess. <shrug>
|
... I could always just bypass the faulty attenuator internally, replace the top and bottom cases and just use any high-quality adjustable RF attenuator in series with a 10dB fixed attenuator permanently attached to the RF input at the front panel. So it wouldn't be the end of the world I guess. <shrug>
Jinxie, that is not a good idea. The input attenuator setting is coordinated with various gain stages in the IF to give an accurate amplitude reading regardless of attenuator setting. This is the reason you were so easily able to test the hypothesis that the attenuator may be bad. If you use an external attenuator, it will have no way of communicating with the processor. Each time you change the reference level, you will have to manually enter the expected attenuation.
?
On very rare occasions, the attenuator may limber up with repeated exercise if there isn't a problem with either the o-rings or the "Y-contacts". Depending on the age of the attenuator, it either has these mechanical Y-contacts, or a circuit board with solid state logic inside. You will have to open the attenuator to see which it is. Replacing o-rings requires some manual dexterity, but it's not like watch repair. Somebody just posted a link to a pdf file with instructions on this forum.
?
If a section is burned out, you can possibly make one good attenuator out of two bad ones. What do you know about the history of this 8566?
?
Vladan
|
Hi Vladan,
?
Many thanks indeed for your observations. I was certainly aware that the Analyzer would no longer 'know' what level of attenuation was being applied to the source RF and that I would have to manually account for it, but provided that doesn't put any of the downstream circuitry at risk of damage, it's an inconvenience I may have no alternative but to live with.
This SA was owned by an RF engineer and I purchase it from him about 10 years ago along with a faulty Marconi/Aeroflex RF signal generator with a disclosed fault. No faults were disclosed in regards to the SA, however. The sig gen I was able to repair and it's a valuable item of equipment to have for all sorts of reasons. Anyway the fact that the seller disclosed the sig gen fault leads me to believe he was honest and was unaware of any issue with the SA. However, I've never done anything with that SA which could burn out the 0 and 10dB settings either, having used it solely for monitoring the RF environment around me using a short telescopic antenna, so no physical connection to the RF input.
As I mentioned, I do have an HP 8495K 0-70dB attenuator which might save the day. I don't think there are any issues with that. However, it's NOT the same type as the one in the 8566B...
|
Hi Jinxie,
If 0dB attenuation was damaged because of over voltage it would have damaged a lot more. It is very likely some mechanical problem like sticky o-rings. Your phase detector repair was a lot more complex, why don't you try repairing the attenuator first? Worst case is you need a new one so nothing to lose. It may be obvious but I will mention it: Once attenuator is out can test it with a DVM by terminating to 50-ohms at DVM and applying DC voltage. Just do 20*log(Vo/Vi). This attenuator doesn't have a DC block in it so DC would work.?
Ozan
?
?
On Fri, Sep 6, 2024 at 10:14 AM, Jinxie wrote:
Hi Vladan,
?
Many thanks indeed for your observations. I was certainly aware that the Analyzer would no longer 'know' what level of attenuation was being applied to the source RF and that I would have to manually account for it, but provided that doesn't put any of the downstream circuitry at risk of damage, it's an inconvenience I may have no alternative but to live with.
This SA was owned by an RF engineer and I purchase it from him about 10 years ago along with a faulty Marconi/Aeroflex RF signal generator with a disclosed fault. No faults were disclosed in regards to the SA, however. The sig gen I was able to repair and it's a valuable item of equipment to have for all sorts of reasons. Anyway the fact that the seller disclosed the sig gen fault leads me to believe he was honest and was unaware of any issue with the SA. However, I've never done anything with that SA which could burn out the 0 and 10dB settings either, having used it solely for monitoring the RF environment around me using a short telescopic antenna, so no physical connection to the RF input.
As I mentioned, I do have an HP 8495K 0-70dB attenuator which might save the day. I don't think there are any issues with that. However, it's NOT the same type as the one in the 8566B...
|
FWIW I had a faulty section in one of these style of attenuators,
when I took it apart there was some corrosion around one of the
resistor chips where it was grounded to the aluminium body, cleaned
it up and it's been fine for (some years) since.
As Ozan says, dig in to it and see what you find, you may be lucky!
On 06/09/2024 19:03, Ozan wrote:
toggle quoted message
Show quoted text
Hi Jinxie,
If 0dB attenuation was damaged because of over voltage it
would have damaged a lot more. It is very likely some mechanical
problem like sticky o-rings. Your phase detector repair was a
lot more complex, why don't you try repairing the attenuator
first? Worst case is you need a new one so nothing to lose. It
may be obvious but I will mention it: Once attenuator is out can
test it with a DVM by terminating to 50-ohms at DVM and applying
DC voltage. Just do 20*log(Vo/Vi). This attenuator doesn't have
a DC block in it so DC would work.?
Ozan
?
?
On Fri, Sep 6, 2024 at 10:14 AM, Jinxie wrote:
Hi Vladan,
?
Many thanks indeed for your observations. I was certainly
aware that the Analyzer would no longer 'know' what level of
attenuation was being applied to the source RF and that I
would have to manually account for it, but provided that
doesn't put any of the downstream circuitry at risk of
damage, it's an inconvenience I may have no alternative but
to live with.
This SA was owned by an RF engineer and I purchase it
from him about 10 years ago along with a faulty
Marconi/Aeroflex RF signal generator with a disclosed fault.
No faults were disclosed in regards to the SA, however. The
sig gen I was able to repair and it's a valuable item of
equipment to have for all sorts of reasons. Anyway the fact
that the seller disclosed the sig gen fault leads me to
believe he was honest and was unaware of any issue with the
SA. However, I've never done anything with that SA which
could burn out the 0 and 10dB settings either, having used
it solely for monitoring the RF environment around me using
a short telescopic antenna, so no physical connection to the
RF input.
As I mentioned, I do have an HP 8495K 0-70dB attenuator
which might save the day. I don't think there are any issues
with that. However, it's NOT the same type as the one in the
8566B...
|
Jinxie, here is my guess as to what is wrong with the attenuator. Below is part of your post from a previous thread.
?
This attenuator has three sections, 10, 20, and 40dB. Each section is either in line or it's bypassed. Various combinations of these three values will give 0-70dB in 10dB steps. Ignoring your measurements at 60 and 70dB for a moment, it looks like 50, 40, 30? and 20 are switching in and out correctly. The 10dB section is not properly bypassed in the 0dB setting. It's also not in-line in the 10dB setting. Thereffore:
?
- One or both plungers in the 10dB section are not working, likely because of one or more bad/missing o-rings.
- The solenoid logic is not working. If this attenuator has the mechanical logic, I would look at the Y-contacts.
?
You can determine if the solenoid is doing anything by listening for clicks when going between 0 and 10dB. Also listen for the type of sound when comparing the non-working 10dB section to another (working) section.
?
Regarding your 60 and 70dB observations, you don't say if you left the display in the 1dB/div setting, or if you changed to 10dB/div. You also don't say where on the screen the noise is. If you were observing with a 10/ vertical scale, and the noise was way down, then this would confirm that the path through the 10dB section is interrupted while in the 70dB setting. Similarly, if the 60dB setting shows noise near the bottom of the screen, that would imply that the path around the 10dB section is also open.
?
You can repeat these experiments by recallung reg. 8 and then changing the scale to 10dB/div. It's usually better to post pictures than to give verbal descriptions.
?
Vladan
?
At 70dB, I'm seeing just massive noise.
At 60dB, occasional noise fickers
At 50dB, I see a thick, fuzzy arch
At 40dB, I see a less fuzzy arch
At 30dB, I see a nice, clean arch
At 20dB, I see a nice clean arch of almost the same amplitude as with 30dB.
At 10dB, I see nothing
At 0dB, I again see nothing.
|
I did actually post pictures of what I was seeing on each of the settings on another related thread message #146621.? Having now taken a second look at those pictures however, I have noticed they were indeed at 1dB/div on the vertical scale so I shall have to re-do them on 10dB/div for the sake of clarity.
Sigh... Sunday I have some time available so will remove the attenuator and inspect its internals. I may post some photos of it's innards once that's been done.
I have to say I'm becoming rather more intimately acquainted with the guts of this analyzer than I'd ever imagined!
|
Okay, done a bit more testing. As was suggested, I bypassed the internal variable attenuator with a fixed 10dB one and attempted calibration once again. It now calibrates fine.
I thought whilst I had the front panel off but still connected, I'd run a test of the attenuator using a spare spectrum analyzer I have. Methodology was simply to squirt RF in at the input and measure its actual level of attenuation vs its stated level of attenuation for each of the settings 0 through -70dB using the front panel to change the settings each time.
Got some quite interesting results.
?
Setting (dB) ??????????????????????????????????? Actual attenuation (dB)
0????????????????????????????????????????????????????????????? -20
-10????????????????????????????????????????????????????????? -30
-20?????????????? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -20
-30 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -30
-40 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -40
-50 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -50
-60 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -70
-70 ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?? -80! (with massive noise so presumably dirty contacts)
?
Clearly the attenuator needs to be removed and inspected, so that's what I shall now do... ??????????????????
|
I have now extricated the attenuator from the guts of the analyzer (see picture). Can some kind soul point me to the disassembly instructions for this thing? I've also included a photo of the spare attenuator I have which I may have to cannibalize to repair the original one. Also, if anyone has a rough idea of the size of the o-rings I might encounter later on, that would be helpful. I have quite a selection of o-rings 'in stock' here.
?
J.
?
|
Hi, O©\rings are 2mm OD and 1mm ?D. Commonly used to seal the stem on wristwatch winders. Just be wary of cheap imports made from inferior materials!
On 8 September 2024 17:45:38 BST, "Jinxie via groups.io" <paul666@...> wrote:
I have now extricated the attenuator from the guts of the analyzer (see picture). Can some kind soul point me to the disassembly instructions for this thing? I've also included a photo of the spare attenuator I have which I may have to cannibalize to repair the original one. Also, if anyone has a rough idea of the size of the o-rings I might encounter later on, that would be helpful. I have quite a selection of o-rings 'in stock' here.
?
J.
?
Attachments:
|
|
|
Jinxie, I made a mistake, the 8566 attenuator has four sections, not three. I may have mixed it up with the 8568 attenuator. A 40dB attenuator card with good return loss to 22GHz is difficult to make, so the 8566 uses two 20dB cards to get 40dB plus another 20dB card to get to 60dB. The remaining card is 10dB. Your numbers show that one of the 20dB stages in your attenuator never bypasses.
?
The 8566 attenuator you have (85660-60121) may possibly also be found under the following part numbers. This is from some old notes I have, verify before buying anything.
?
33321YH
5086-7814
5086-6814
33321RF
and possibly 85660-60304
?
The reason for all these numbers is that HP sold parts to the outside world, as well internally from division to division. Also, each 8566 (and 8568) attenuator was supplied with a PROM containing calibration data for attenuation for each stage. Don¡¯t worry about that PROM, the corrections are of very limited use. The standard 33321 is the same hardware without the PROM.
?
Your 8495K uses either 24 or 5V for switching. It should have a label somewhere on it showing which. Option 024 is 24V and option 011 is 5V. Default is Opt. 024, but you should check what you have.
?
The disassembly instructions were referenced in a recent post:
?
?
This writeup may also exist in the files section of this forum.
?
In a few words, there are two compartments. One consists of the solenoids, the other is a channel with the attenuator cards and leaf springs. You start out by removing the end plate held in place with four socket screws. This is the end plate that is opposite from the one with the control line connector. The screws are often hidden by a decorative cover plate which is lightly glued on. Once you remove the end plate, you can slide off the U-shaped metal cover. Don¡¯t try to pry it off by putting a screwdriver into the seem on the side, that¡¯s not how it was designed to be opened.
?
After you remove the solenoid cover but before you open the rf side, check visually that the solenoids are moving by changing attenuation via the front panel keys.
?
The RF block is opened by removing the screws you see around and under the label. This is the area where you will be doing most of the work. Keep this area free of lint. Be extremely careful to not deform the leaf springs that act as contacts. They are easily bent and ruined. You will have to stress them a little while unhooking the plungers which have the o-rings on them. Maybe the o-rings are still ok, in which case you don¡¯t have to risk damaging the springs. You can inspect the o-rings without removing the plungers. The o-rings are 0.039" ID x 0.024" hydrogenated nitrile rubber, 80 durometer, shore A.
?
Vladan
|
Okay, many thanks indeed for all the info and pointers. That lot should set me in good stead for the task ahead, which I'm not really looking forward to as it seems to require a good eye and steady hand - neither of which I'm blessed with!? I really don't want the bench fully occupied with 'stuff' as it currently is until next Sunday, so will attempt this repair by making time during the week and pray for no hitches. I'm just hoping it's nothing terminal. Just out of curiosity, what material are these pads made from? I gather they can be a bit brittle and fragile. Were they like that from new or is it just another aspect of the age of this analyzer?
Anyway, ludicrously late here in my part of the world now so time for some shut-eye, methinks..
?
|
If the problem is with the o-rings (likely scenario), you will not be doing anything to the attenuator cards. You will not be removing them unless they have been overloaded. The cards are made of thin film resistors on saphire substrates.They don't deteriorate unless overloaded.
?
For o-ring replacement, you will have to carefully unhook the plastic plungers from the leaf springs. You will need some fine tools like tweezers or similar to manipulate the plungers out of the slots in the springs. The o-rings are mounted on these plungers. Then, you will put all that back in. A low power microscope or watchmaker's loupe would help. Before you start the extraction of the plungers, study the layout and understand why the rings are there. Exercise the plungers by hand to see how the mechanism works. If you don't see a ring, it may have cracked and fallen off. Compare to the neighboring section to see where it should be. The most important thing is to not bend the springs past their elastic limit while dealing with the plungers.
?
Vladan
|
The sizes I¡¯ve seen for watches weren¡¯t 1/2mm cross section. They were thicker like 0.6mm (1.0 ID/2.2OD or something like that).
The proper size isn¡¯t that hard to come by.
toggle quoted message
Show quoted text
Hi, O©\rings are 2mm OD and 1mm ?D. Commonly used to seal the stem on wristwatch winders. Just be wary of cheap imports made from inferior materials!
On 8 September 2024 17:45:38 BST, "Jinxie via groups.io" <paul666@...> wrote:
I have now extricated the attenuator from the guts of the analyzer (see picture). Can some kind soul point me to the disassembly instructions for this thing? I've also included a photo of the spare attenuator I have which I may have to cannibalize to repair
the original one. Also, if anyone has a rough idea of the size of the o-rings I might encounter later on, that would be helpful. I have quite a selection of o-rings 'in stock' here.
?
J.
?
Attachments:
|
Well, I've ordered a range of sizes just to be on the safe side. Unfortunately they won't be arriving any time soon (yeah, China) so there won't be any updates for at least a week I'm afraid. Anyway, now I've got the attenuator out, this thread's objective has been achieved, so I'll begin a new one for the actual overhaul process when those parts are finally here.
Many thanks for your assistance! :)
|
Jinxie
Jack
