开云体育

  1. HP-Agilent-Keysight-Equipment
  2. Topics

HP8566B band amplitude response discrepancy - how much indicates a component failure?

 

I'm still looking at this 8566B I just repaired, and pondering what it takes to do more adjustment. Right now it looks like the 2-5.8 GHz band shows over 12.5 dB lower than in the low 0-2.5 GHz band. I'll probably start trying to tweak things to see what's needed to bring it up to par. This is the unit that had a shot YTO - the oscillator Q had B-C shorted, evidenced by the blown collector bias resistor in one of the "RO" spots. With the fresh YTO and the A20 adjustments done, the 0-2.5 GHz band seems to work fine, so I know the frequency controls and 1LO power are sufficient.
?
I think the 13 dB or so less in the next band and whatever amounts in the higher ones should be easy to correct with proper adjustments, but I have never done this on any of my 8566s. I'm wondering whether this initial difference is no big deal in the scheme of things, or indicative of some other failure. From a quick look at the adjustment document it looks like it will get quite complicated and tedious, involving alignment of the YTX with the YTO and such, band by band. There's a lot going on in there. I think the main thing next is to assess the condition of the YTX.
?
The only comparison I've done so far is in the overlap of the bands around 2.3 GHz. The low band is checked against the 100 MHz -10 dBm cal reference (which I checked with the 436A), then the response is checked at 2.3 GHz ~0 dBm from the HP70331A and then in the 2-5.8 band. The preselector peaking process seems to work through its routine, so the YTX must be doing some tuning at least.
?
Anyway, I think it's likely OK and close enough operationally that it's mostly going to be about adjustments, and not some other component failure. I'll find out eventually, but it seems a lot more involved than I had expected. I'm hoping for a fairly low frustration and anxiety level.
?
Should I worry?
?
Ed
 

开云体育

You’ve gone through the YTO loop calibration, right?? If the pretuning controls aren’t right for your replacement YTO, it might end up off-frequency far enough to make certain bands look worse than others.?

?

You do NOT want to mess with the YTX calibration process unless you have a lot of time on your hands and a suitable sweeper.

?

-- john, KE5FX

?

From: [email protected] <[email protected]> On Behalf Of ed breya
Sent: Monday, May 26, 2025 12:57 PM
To: [email protected]
Subject: [HP-Agilent-Keysight-equipment] HP8566B band amplitude response discrepancy - how much indicates a component failure?

?

I'm still looking at this 8566B I just repaired, and pondering what it takes to do more adjustment. Right now it looks like the 2-5.8 GHz band shows over 12.5 dB lower than in the low 0-2.5 GHz band. I'll probably start trying to tweak things to see what's needed to bring it up to par. This is the unit that had a shot YTO - the oscillator Q had B-C shorted, evidenced by the blown collector bias resistor in one of the "RO" spots. With the fresh YTO and the A20 adjustments done, the 0-2.5 GHz band seems to work fine, so I know the frequency controls and 1LO power are sufficient.

?

I think the 13 dB or so less in the next band and whatever amounts in the higher ones should be easy to correct with proper adjustments, but I have never done this on any of my 8566s. I'm wondering whether this initial difference is no big deal in the scheme of things, or indicative of some other failure. From a quick look at the adjustment document it looks like it will get quite complicated and tedious, involving alignment of the YTX with the YTO and such, band by band. There's a lot going on in there. I think the main thing next is to assess the condition of the YTX.

?

The only comparison I've done so far is in the overlap of the bands around 2.3 GHz. The low band is checked against the 100 MHz -10 dBm cal reference (which I checked with the 436A), then the response is checked at 2.3 GHz ~0 dBm from the HP70331A and then in the 2-5.8 band. The preselector peaking process seems to work through its routine, so the YTX must be doing some tuning at least.

?

Anyway, I think it's likely OK and close enough operationally that it's mostly going to be about adjustments, and not some other component failure. I'll find out eventually, but it seems a lot more involved than I had expected. I'm hoping for a fairly low frustration and anxiety level.

?

Should I worry?

?

Ed
 

I have only done the adjustment procedure for A20, with the "preferred method" where you enter a bunch of shift-command stuff that gets saved, then you go between recall 1 and recall 2 registers that forces the YTO (1LO) output to either 2.3 or 6.15 GHz. You're supposed to alternate between the two settings while monitoring 1LO with a counter (which I have referenced from the 8566B's 10 MHz ref to get "exact" counts) and tweak it for the correct readings through a number of iterations, since they interact. This seems to have worked as it was supposed to, after about eight iterations. I tried this earlier while it was still broken (C18 on A19 was shorted) and it was stuck around 2.1 GHz. BTW this part was discussed in the first different thread about the problem - it is here for reference in case it wasn't seen:
?
/g/HP-Agilent-Keysight-equipment/topic/hp8566b_band_amplitude/113315244
?
Once it was all back together, the only frequency control part that really changed was the YTO - all the boards were original, except for A20's tweaks and A19's repair. I figured that the pre-tune DAC stuff on A19 should be still be good and appropriate for the YTX et al, IFF that keys off the sweep/tune signal(s), and not the YTO coil drive, AND only I have previously tweaked anything else while attempting to fix the original problem. So, with everything original except the YTO, I think adjusting only its coil drive on A20 was right to align it with the the rest. I had already messed it up while broken, and the procedure did work after the fix, so presumably it now tracks properly with what it is given.
?
The big question I think is whether my assumptions were right about the overall control scheme, and any past possible tweaking by others. I'll have to review the adjustments and theory of operation for all that, unless someone happens to know a quick and easy answer. I think it boils down to whether the YTO coil adjustments on A20 directly influence the YTX control or other things, or is it only apparent in effects on their relative alignment. I believe the adjustments on A19 affect both the YTO and the YTX system, but after that they are separate things that still track each other according to proper adjustments. If someone has tweaked things wrongly in the past, then that's another separate story to resolve. Of course there's always another story, which is whether there are other failures. Oh well, one thing at a time.
?
Ed
 

When working with the upper bands which go through the YTX be sure that the <PRESEL PEAK> values aren't way off into the weeds, if they are the measured amplitude will be a lot lower than reality.
?
Each upper band can be preselected-peaked to ensure that the YTX bandpass is optimized/positioned to achieve the greatest (most accurate) amplitude.
?
Do a <SHIFT> <PRESEL PEAK> (KS=) to set the <PRESEL PEAK> values back into their factory preset values.
?
After ensuring that the <PRESEL PEAK> values have been reset, check to see if there is any improvement in the observed amplitude.
?
If you don't see any improvement in the amplitude response then you will probably need to go through the lengthy YTX alignment process, which does require having a suitable signal source.
 

开云体育

Perhaps it's the YTX (tuned filter) tracking/tuning is off kilter?

Especially, if the frequency accuracy of what signals can be seen on screen is relatively OK, that would prove the swept local oscillators are tuning OK, irrespective of the indicated amplitude.

Just thinking back some decades to when I was servicing such things for another well known measuring instrument maker.? Yig Tuned Filters were just as important to get sweeping correctly.? There again, on those things there was a user control for "peaking" the front end YTF, and you did need to use it, as things changed noticeably as the instruments warmed up.

Going back under my rock.

Dave 'KBV.
 

Typically the preselect peak values should vary about NTE +/-10 from the nominal factory preset value (32) from 2 GHz to 22 GHz. You can use these to identify frequency tuning issues/ frequency response issues on each band. (band A = 0 Hz to 2.5 GHz, band B = 2.05 GHz - 5.8 GHz, band C = 5.8 GHz - 12.5 GHz, band D = 12.5 GHz - 18.6 GHz, band E = 18.6 GHz - 22 GHz). The high, low, mid points of each band should be checked by peaking on a signal at that point and checking the PSP value.
Case 1: all preselector values at each band’s lo, mid, hi points are nominal (see above) but the signal amplitude is OOT (out-of-tolerance) then the frequency response amplitude need to be done.
Case 2: preselector values at low end of the band B read higher than nominal, and the high end of band B read higher or lower than nominal then the band B YTO tune needs to be done. All the upper bands are based on the alignment of this band, so any adjustments on this band will require the full frequency tune and frequency response alignment.
If this alignment is done carefully and properly you can achieve frequency flatness from 10 MHz to 22 GHz within half of the published specs using just the nominal PSP value (32), and using PSP at all points of interest you can get to 1/3 of the published spec or less.
I had one unit that I adjusted that was easily in spec and did not exceed +/- 1.5 dB variation all the way up to 22 GHz. It was still inside this spec range 6 mos. - 18 mos. later. These units can be very stable and accurate for years and even decades.
I am very biased toward these units, having worked on them for 30+ yrs.
Don Bitters
 

After quite a bit of experimenting and poking around in various sections, I'm convinced that the frequency alignment of the YTO and YTX are OK, and the real problem is loss of conversion gain somehow in the upper bands. It looks like that 12-13 dB drop I had in band B, 2-5.8 GHz, shows more or less all the way up, with band C a bit worse, band D a bit better, and band E way down. I got another sweeper out to cover all frequencies but not accurately leveled, so it takes some interpretation to estimate the results.
?
I did some experiments in frequency control, by marking certain pot positions then tweaking pretty far in each direction and back - I could only make the amplitude worse, not better. Likewise, I could not get much higher displayed amplitude even with certain level setting pots maxed out - none could get it anywhere near the spec display level. All during this, the pre-selector peaking routine always works, raising and lowering the tuning around the response skirt, then arriving at the peak level, just at the wrong way-too-low amplitude answer. Adjusting the YTX bias pot for band B only by around +/- a few turns had virtually no effect. On A6A11 I tried adjusting the main gain pot which had a lot of effect but still couldn't reach spec (and put band A way over the top), and neither could the slope gain B1 (for low end) at 2.3 GHz in.
?
I've studied it a lot - it's hard to keep track of what does what, there's so %#(&* many adjustments in so many places. I haven't dug in yet to key measurements on the boards and such. I will start by pulling some and checking for obvious issues, then deciding where to tack on test points. It has been very frustrating using this poor quality pdf I found - I can't read the details on the schematics. I finally found my paper manual today, so now I can see the big and small pictures. I thought some sections were missing though, of course the parts that I suspect the most now, mainly the YTX (A6A8) itself and that ACLLI thing (A6A5). I looked for them on the pdf and they're not shown here either, so I guess they are closed RF modules without details of the insides.
?
I can picture blaming the YTX for maybe having a bad mixer diode - as I understand, there's only a single one in there, so it would have to be only partially burned out since there is some conversion going on. Or, say a damaged YIG filter input link from a past abuse, or some connection failure in a module. I don't know if that A5 ACLLI whatever coupler unit is all passive or not. I have to study it some more since I thought the manual mentioned an amplifier associated with it. These other guts of course are a PITA requiring a bunch of disassembly to work on. One fairly easy one is the second converter A6A4, since it's on the top of the plumbing heap. It works fine for the low band but contains the PIN switching to route in all the high band 321.4 MHz IF, so that's yet another possible suspect. The PIN control line to it reads about +17.5 or -7.5V in its two states, which seem a bit low, looking at the drive circuitry, but there could be some extra series R here and there, and the PINs usually take a fair amount of current. Once that IF gets to the last converter, it should be OK unless something associated with all those &$%@#%* high band tweaks has gone bad in such a way to effect only them en masse, yet leave the low band alone. It has to be something upstream enough and common to all. All the variable gain stuff in there is shared by all the bands.
?
My latest observation is that in all the high bands, with probably +6dBm (or more at the lower end of high) from the sweeper, and the reference level 0 dBm, all the peak readings tend to fall between the <0 and -10 dBm level, and after some peak accumulation, harmonics start showing at a few dB higher than the fundamental peaks. I think this tells me that the mixer is being over-driven well beyond the ref level, so there's plenty enough RF power getting through the front end system, but there's a problem in the 321.4 MHz IF behind it. This is also at full 0-24 GHz sweep, so the low stuff is applied too, but it should be rejected by the YTF first thing. This could all still be explained by a large YTO/YTX frequency mismatch due to some other failure, that can't be tweaked out, so I'm not out of the woods on that yet either. I may have to devise an experiment to lock the YTO with those test setup codes, then force override the YTX over a huge range and see what comes out.
?
I'm getting pretty frustrated and tired of this. I might have to take a break and move on to another project for a while.
?
Ed
?
?
 

开云体育

Agreed, your experiments should rule out the tracking mismatch theory.? Can you measure the bias voltage while you adjust the band-B YTX bias? ?That sounds like it might be a clue -- if the bias isn’t actually changing, or isn’t changing much, it is worth understanding why.

?

Another thing you might try is a basic two-tone IMD test, if you can put your hands on two microwave sources and a suitable combiner (or one vector-modulated source.)? Third-order products above -50 dBc or so could suggest diode damage or LO power being lost on its way to the YTX.? ?OTOH if distortion looks good at << -50 while the signal level is lower than expected, that would tend to cast suspicion downstream of the 1st mixer.

?

The free .PDF manuals for the 8566B are a complete waste of time and downright misleading in places where the OCR altered letters and numbers.? A pox on the house of whoever scanned those… they wasted not only their own time but everyone else’s as well.? Artek Media has a great set of PDF manuals that I’d suggest ponying up for.? Paper and .PDF complement each other when working on these beasts, it’s not an either-or proposition.

?

As an example, ACLII is actually ‘ACLU’ (Amplifier/Coupler/Load Unit).? Naturally the ‘CIA’ (Coupler-Isolator Amplifier) is close behind it.? Ah, the 1970s…

?

-- john, KE5FX

?

?

From: [email protected] <[email protected]> On Behalf Of ed breya
Sent: Saturday, May 31, 2025 6:42 PM
To: [email protected]
Subject: Re: [HP-Agilent-Keysight-equipment] HP8566B band amplitude response discrepancy - how much indicates a component failure?

?

After quite a bit of experimenting and poking around in various sections, I'm convinced that the frequency alignment of the YTO and YTX are OK, and the real problem is loss of conversion gain somehow in the upper bands. It looks like that 12-13 dB drop I had in band B, 2-5.8 GHz, shows more or less all the way up, with band C a bit worse, band D a bit better, and band E way down…
 

开云体育

Something else worth pointing out re: the manuals is that for many instruments, the HP Journal introductions are much better initial resources than the service manuals.? The 8566A issue from August 1979 is a perfect example: ? Figure 1 on page 8 shows that there’s a fair bit of stuff in the ACLU including, unsurprisingly, some amplifier stages.

?

As usual, the first Google link brings up a crappy scan that’s barely legible in places, which why I scanned a bunch of them myself.

?

-- john, KE5FX

?
 

Ed, the A6A5 circuit (Amplifier Coupler Load Unit) maybe at fault. How much RF power are you getting at the front panel 1st LO output - +13 - +16 dBm should be about right, this front panel signal is routed through the ACzLU, IIRC, should be there for every band you check, and should have a 50 ohm SMA load on it normally. What does your noise floor look like after IP, then set to single sweep, with a start freq. of 10 MHz, stop freq. of 22 GHz, single sweep?
What you should see is noise floor peaks of about -68dBm on band A, -65dBm on band B, -59 dBm on band C, -53 dBm on band D, -47 dBm on band E, about 6 dB increase in gain per band, a staircase. This is compensation for the harmonic conversion loss in the YTX mixer filter.
What can occur is if the band switching is not happening at the correct time or 2 bands are switched on at the same time, the band gain can be all wrong, hence the amplitude can be incorrect. The switching circuitry is on A12, or A16 brd. in the RF section. You did say that one band was near normal (B or C), I think IIRC that you said that band B was close, C was worse, D was almost normal, but E was very bad as far as amplitude. I have replaced the band switch FET’s and solved what I described above.

Don Bitters

1 - 10 of 10