|
Thankyou Vladan, Don, and Ed. I aligned the crystal 3MHz IF last night, following the drill on pages 3-76 through 3-81 as closely as possible [I don`t have access to a 21.4MHz frequency synthesizer for example] The centering , and symmetry trimmers work as they should, and I was able to get symmetrical responses of the correct amplitude at each successive stage. FWIW : I got it wrong - the correct way is to apply the bypass networks to stages 2,3,4,and 5, align stage 1, then remove stage 2`s network and align stage 2, and so on. Your`e not supposed to remove one bypass network at a time as I said previously. I got fed up and called it a day at 11-00pm last night - I intend to give the 8566B a full IF alignment at a later date, but I have a 475 to fix, and two? H/P606`s too. I could use my 8656A siggenny using a GPSDO reference as my 21.4MHz frequency source. The solder joints on the 3MHZ IF sure looked just fine - seems like your soldering must be darn near perfect to last 31 years. Thankyou for your help with the notches. Buenos notches!.....................................................................................................................................................................................................Don.
On Wed, Dec 5, 2018 at 5:58 PM Ed Breya via Groups.Io <edbreya= [email protected]> wrote: Don wrote:
"I ask this because because CF on the 8566 appears to be at 10.000050MHz (+50KHz). This is OOT for the spans you are showing."
I wouldn't worry too much about the "exact" CF readout yet, until calibration. The final processing IF is ideally 21.4 MHz exactly, and the crystal filter's CF is 3 MHz exactly, but in reality the crystals have nominal tolerance and drift a bit over time and temperature. That's partly why there's the adjustable 18.4 MHz down-up converter, which also drifts itself. The 18.4 MHz is subtracted from the ideal 21.4, to make the non-deal 3 MHz crystal CF, then the result is converted back up to the ideal 21.4 - the 18.4 cancels out, but its actual frequency may be slightly different. It is controlled by the the front panel "FREQUENCY ZERO" tweak pot on the IF/display unit, and by an internal coarse adjustment.
The main idea is to land all the crystal filter sections as close as possible to the ideal 3 MHz, at nominal environmental conditions, and the 18.4 MHz coarse adjustment centered at the frequency (while the frequency zero pot is set at its midpoint) where it plus the final adjusted value of the 3 MHz CF, makes the ideal 21.4, at maximum response level. The fine tuning of the 18.4 MHz is then tweaked as needed during operation, for temperature, drift, etc - but only really needed at high resolution settings like 300 Hz or less IFBW. Even after "proper" setting, the crystals and the 18.4 MHz can drift gradually during use, needing occasional tweaking of the pot. You can see this quite easily at 10 Hz IFBW.
Ed
|
Don wrote:
"I ask this because because CF on the 8566 appears to be at 10.000050MHz (+50KHz). This is OOT for the spans you are showing."
I wouldn't worry too much about the "exact" CF readout yet, until calibration. The final processing IF is ideally 21.4 MHz exactly, and the crystal filter's CF is 3 MHz exactly, but in reality the crystals have nominal tolerance and drift a bit over time and temperature. That's partly why there's the adjustable 18.4 MHz down-up converter, which also drifts itself. The 18.4 MHz is subtracted from the ideal 21.4, to make the non-deal 3 MHz crystal CF, then the result is converted back up to the ideal 21.4 - the 18.4 cancels out, but its actual frequency may be slightly different. It is controlled by the the front panel "FREQUENCY ZERO" tweak pot on the IF/display unit, and by an internal coarse adjustment.
The main idea is to land all the crystal filter sections as close as possible to the ideal 3 MHz, at nominal environmental conditions, and the 18.4 MHz coarse adjustment centered at the frequency (while the frequency zero pot is set at its midpoint) where it plus the final adjusted value of the 3 MHz CF, makes the ideal 21.4, at maximum response level. The fine tuning of the 18.4 MHz is then tweaked as needed during operation, for temperature, drift, etc - but only really needed at high resolution settings like 300 Hz or less IFBW. Even after "proper" setting, the crystals and the 18.4 MHz can drift gradually during use, needing occasional tweaking of the pot. You can see this quite easily at 10 Hz IFBW.
Ed
|
Don,
I have a couple of questions, with all poles bypassed but one, does the centering adjust peak/dip the filter equally from side to side? Does the symmetry adjust cause the skirts to move approximately equal amounts from center vertical?
I ask this because because CF on the 8566 appears to be at 10.000050MHz (+50KHz). This is OOT for the spans you are showing. Also bypassing 1 pole seems to be causing the amplitude to increase by about +3dB. When adjusted properly the tolerance is +-0.5dB, +-0.05dB is easily achievable.
Recheck the center freq for each pole. 10KHz RBW is all 21.4MHz RBW brds, 3KHz RBW is the natural CF (21.4MHz nominal) of the Log Amp brds, the A4A6 downconverter brd, and the A4A7 brd - CF adjusts on A4A3 and A4A6 are made so that there is no freq. shift when changing RBW¡¯s and amplitudes.
I believe that your pole using C6 is not centered. The XTAL filter adjusts for 21.4 BW¡¯s and 3MHz BW¡¯s use a peaking method and the LC filters use a dip method.
Happy troubleshooting!
Don Bitters
|
Nicely done! I have seen intermittent, temp sensitive trimmer caps, also oxidizes - sometimes Contact ReNu will revitalize the surface. Just to reiterate the obvious all the poles 21.4MHz and 3MHz are aligned to the exact same 21.4MHz center pass freq. If they are badly misaligned, it is usually easier to to do single pole (all other poles bypassed) first then go to the book procedure.
Good luck and great skill in the alignment.
Don Bitters
|
That's great, Donald! Regarding the picture of the new fans on your 8566, is there a thread about quieting those down? I must have missed it. Did you actually manage to make the thing a little less noisy?
Vladan
|
My thanks also, I have a very intermittent rotary knob that finally failed hard enough to do board switching unfortunately to no good end. I moved the wires to put the scope probes on and of course it was fixed
After reading you report I¡¯m thinking the staked pins might have gone wonky so solder is in their future Regards, ? Stephen Hanselman Datagate Systems, LLC 3107 North Deer Run Road #24 Carson City, Nevada, 89701 (775) 882-5117?office (775) 720-6020?mobile s.hanselman@...
a Service Disabled, Veteran Owned Small Business DISCLAIMER: This e-mail and any attachments are intended only for use by the addressee(s) named herein and may contain legally privileged and/or proprietary information. If you are not the intended recipient, any dissemination, distribution or copying of this e-mail and any attachments is strictly prohibited. If you have received this e-mail in error, please immediately notify me and permanently delete the original and all copies and printouts of this e-mail and any attachments.
toggle quoted message
Show quoted text
On Dec 3, 2018, at 15:50, Ed Breya via Groups.Io < edbreya@...> wrote: Congrats on the fix. I'd suspect it was a gold problem as Chuck suggested, or silver plating on variable capacitor leads. Ag plating causes quite a mess like on old TI-brand TTL ICs, which is another thing to watch out for.
Ed
|
Congrats on the fix. I'd suspect it was a gold problem as Chuck suggested, or silver plating on variable capacitor leads. Ag plating causes quite a mess like on old TI-brand TTL ICs, which is another thing to watch out for.
Ed
|
Cracked solder joints are pretty popular in equipment
of this vintage, made by HP. Particularly boards that used
gold plating as the etchant resist. Gold is pretty, shiny,
and makes everyone smile, but it also makes a nasty little
barrier alloy with tin/lead solder that is very brittle.
Also, be aware that all the SMA/3.5mm connectors that are
soldered to a piece of 141 or smaller semi-rigid coax are
prone to breaking. The solder joint is too small and relies
too heavily on a solder fillet for its strength. If you put
much force on the cable it will crack this solder fillet,
opening the shield at the connector... this can make for a
wonderfully frequency dependent intermittent open circuit.
-Chuck Harris
Dave McGuire wrote:
toggle quoted message
Show quoted text
On 12/3/18 6:21 AM, donald collie wrote:
As a precaution, I resoldered all the joints in the 1st stage - no more
notch and symmetrical responses attainable. I`ll align the thing
tomorrow Cheers! Wow...cracked solder joint somewhere? Maybe around one of those
trimmers? Nice recovery.
-Dave
|
On 12/3/18 6:21 AM, donald collie wrote: As a precaution, I resoldered all the joints in the 1st stage - no more
notch and symmetrical responses attainable. I`ll align the thing
tomorrow Cheers! Wow...cracked solder joint somewhere? Maybe around one of those trimmers? Nice recovery. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
As a precaution, I resoldered all the joints in the 1st stage - no more notch and symmetrical responses attainable. I`ll align the thing tomorrow Cheers!.............................Don C.
|
OK. Thanks Vladan, and Don. Meriam and I have been sitting Zac [1 Y.O.], and Isla [almost 2 Y.O.], our grandchildren - all full on [hence the delay in getting back to you both] , but back to the 8566B : Vladan: I went through and bypassed all stages but stage 1 [on the right hand side], then all but stage 2, then 3,4,5 with the result that looking at stage 1 on its own - the notch can be moved from the top of the response [a spike going down from the top] to some distance down the high side skirt. Stages 2,3,4,and 5 can all be adjusted for good symmetry - in each case the notch is gone. It seems to be a symptom of stage 1 only. Don : The notch position, and depth is largely independent of sweep speed - sweep it too fast and it merges with the bell shaped curve of course. I hope it`s not one of the crystals - I can repair everything else. Gentlemen, do you think I should take a closer look at stage 1? I`d be interested to hear your comments. I`m about to check all the components in stage 1 - will let you know the results. Cheers!...........................................................Don C.
On Sun, Dec 2, 2018 at 3:09 PM donald collie via Groups.Io <donaldbcollie= [email protected]> wrote: Standby Vladan...
|
|
|
Donald, can you do the tests I described? /g/HP-Agilent-Keysight-equipment/message/92976You need to bypass all but one stage at a time in order to confirm if that stage is at fault. I can't see from your post if you did that. As I mentioned, bypoassing just one stage is not a good way to do this. The problem may well be in the first stage (around C6), but what you did does not prove it. Vladan
|
The problem shows up in photo 1 [Leftmost], and in photo 5 the notch is clearly visible. By adjusting C6 the notch can be moved up and down the skirt of the response. With a bypass network connected to the two pins near C6 ONLY -? the result is photo 4. The position of the various trimmer capacitors is shown in photo 2. Photo 3 shows the quiet fans installed a month or two ago.. I`m not quite sure if the stage associated with C6, is stage 1 or stage 5, but bypassing this stage removes the notch completely [and cures the assymetry in the 300Hz Res BW response]. The fact that only C6 adjustment will move the notch position, and that when a bypass network is connected to the C6 stage, the notch disappears completely [photo 4], makes me think that the fault is in this stage. Using the 4 bypass networks to perform an alignment doesn`t help.?
?
On Fri, Nov 30, 2018 at 9:56 PM Don Bitters via Groups.Io <donbitters= [email protected]> wrote: The filter shunts are designed to bypass the 3MHz IF XTAL BW filter, one for each of the 1KHz and lower filters.? If you place a shunt on one of the filters and the filter shape looks good (bell shaped curve) then that maybe the bad filter.? You can verify each of the XTAL filters by placing the shunts on 3 of 4 XTAL¡¯s and viewing the filter center and shape for each of the 4 combinations.? If one filter pole is still notched and skewed to the right on the display, then slow the sweep down to 1 to 10sec sweep.? If the filter shape and notch improves then the filter control either on A4A9, or A4A7 is bad.? Ideally with good filters with all the poles but one shunted the filter will be centered and be about 6KHz wide at the 3dB down from peak points, as you remove the shunts the visible BW narrows until with all shunts removed the BW 3dB down with 1KHz BW selected measures 1KHz.
If a pole is actually selected for a narrower BW than indicated then the swept signal through that pole will be skewed right and notched until the sweep is slowed down.
If the poles are all centered at the slower sweep time but too narrow, then check the TA caps on the appropriate control line on A4A7.
If a XTAL is no longer nominally at 21.4MHz then as the 3 shunts are applied then one pole will shift center freq. compared to the other 3 poles.
If the XTAL for a pole is the problem, then try unsoldering that XTAL and tap it lightly on a hard surface then shake it near your ear, if it rattles then you have a solder blob in the can.? This occurred when it was initially soldered in place and lodged somewhere where it was not a problem.? Then the unit got moved and jarred and the solder blob became a problem.? I have never repaired one of these XTAL¡¯s, just replaced the entire set of 7, but logically if you rattle the blob down to the bottom of the XTAL case when in a vertical orientation and heat up one of the XTAL leads you may suck the blob into the solder on that lead.? Don¡¯t heat both leads at one time, you may disturb the crystal structure in the can.? Another option is to move the ¡°bad¡± XTAL to the A4A6 brd where it sets the 21.4MHz center freq. for the Up/Down converter.? The adjustments here are wider ranged and less critical than the individual BW poles.
Once the problem has been resolved the affected brds. will have to be realigned.
When correct,? Cal signal centered, RL @ -7dBm, Lin mode, SWP auto, Span auto, stepping the
RBW from 1MHz to 10Hz the viewed signal should not shift side to side more than +/- 1 div.? and vertically not more than +/- 0.5 div.? I would typically have no movement in any direction.? Also if you run the auto correction routine and look at the correction no¡¯s. the RBW amplitude corr. no¡¯s. will be 0.05 dB or less, and the RBW CF freq. correction no¡¯s would be typically 3% or less of the RBW selected.
Let me know what you find.
Best regards,
Don Bitters
|
Herewith photos of my 8566B, including the fan mod.?
On Fri, Nov 30, 2018 at 9:56 PM Don Bitters via Groups.Io <donbitters= [email protected]> wrote: The filter shunts are designed to bypass the 3MHz IF XTAL BW filter, one for each of the 1KHz and lower filters.? If you place a shunt on one of the filters and the filter shape looks good (bell shaped curve) then that maybe the bad filter.? You can verify each of the XTAL filters by placing the shunts on 3 of 4 XTAL¡¯s and viewing the filter center and shape for each of the 4 combinations.? If one filter pole is still notched and skewed to the right on the display, then slow the sweep down to 1 to 10sec sweep.? If the filter shape and notch improves then the filter control either on A4A9, or A4A7 is bad.? Ideally with good filters with all the poles but one shunted the filter will be centered and be about 6KHz wide at the 3dB down from peak points, as you remove the shunts the visible BW narrows until with all shunts removed the BW 3dB down with 1KHz BW selected measures 1KHz.
If a pole is actually selected for a narrower BW than indicated then the swept signal through that pole will be skewed right and notched until the sweep is slowed down.
If the poles are all centered at the slower sweep time but too narrow, then check the TA caps on the appropriate control line on A4A7.
If a XTAL is no longer nominally at 21.4MHz then as the 3 shunts are applied then one pole will shift center freq. compared to the other 3 poles.
If the XTAL for a pole is the problem, then try unsoldering that XTAL and tap it lightly on a hard surface then shake it near your ear, if it rattles then you have a solder blob in the can.? This occurred when it was initially soldered in place and lodged somewhere where it was not a problem.? Then the unit got moved and jarred and the solder blob became a problem.? I have never repaired one of these XTAL¡¯s, just replaced the entire set of 7, but logically if you rattle the blob down to the bottom of the XTAL case when in a vertical orientation and heat up one of the XTAL leads you may suck the blob into the solder on that lead.? Don¡¯t heat both leads at one time, you may disturb the crystal structure in the can.? Another option is to move the ¡°bad¡± XTAL to the A4A6 brd where it sets the 21.4MHz center freq. for the Up/Down converter.? The adjustments here are wider ranged and less critical than the individual BW poles.
Once the problem has been resolved the affected brds. will have to be realigned.
When correct,? Cal signal centered, RL @ -7dBm, Lin mode, SWP auto, Span auto, stepping the
RBW from 1MHz to 10Hz the viewed signal should not shift side to side more than +/- 1 div.? and vertically not more than +/- 0.5 div.? I would typically have no movement in any direction.? Also if you run the auto correction routine and look at the correction no¡¯s. the RBW amplitude corr. no¡¯s. will be 0.05 dB or less, and the RBW CF freq. correction no¡¯s would be typically 3% or less of the RBW selected.
Let me know what you find.
Best regards,
Don Bitters
|
The filter shunts are designed to bypass the 3MHz IF XTAL BW filter, one for each of the 1KHz and lower filters. If you place a shunt on one of the filters and the filter shape looks good (bell shaped curve) then that maybe the bad filter. You can verify each of the XTAL filters by placing the shunts on 3 of 4 XTAL¡¯s and viewing the filter center and shape for each of the 4 combinations. If one filter pole is still notched and skewed to the right on the display, then slow the sweep down to 1 to 10sec sweep. If the filter shape and notch improves then the filter control either on A4A9, or A4A7 is bad. Ideally with good filters with all the poles but one shunted the filter will be centered and be about 6KHz wide at the 3dB down from peak points, as you remove the shunts the visible BW narrows until with all shunts removed the BW 3dB down with 1KHz BW selected measures 1KHz.
If a pole is actually selected for a narrower BW than indicated then the swept signal through that pole will be skewed right and notched until the sweep is slowed down.
If the poles are all centered at the slower sweep time but too narrow, then check the TA caps on the appropriate control line on A4A7.
If a XTAL is no longer nominally at 21.4MHz then as the 3 shunts are applied then one pole will shift center freq. compared to the other 3 poles.
If the XTAL for a pole is the problem, then try unsoldering that XTAL and tap it lightly on a hard surface then shake it near your ear, if it rattles then you have a solder blob in the can. This occurred when it was initially soldered in place and lodged somewhere where it was not a problem. Then the unit got moved and jarred and the solder blob became a problem. I have never repaired one of these XTAL¡¯s, just replaced the entire set of 7, but logically if you rattle the blob down to the bottom of the XTAL case when in a vertical orientation and heat up one of the XTAL leads you may suck the blob into the solder on that lead. Don¡¯t heat both leads at one time, you may disturb the crystal structure in the can. Another option is to move the ¡°bad¡± XTAL to the A4A6 brd where it sets the 21.4MHz center freq. for the Up/Down converter. The adjustments here are wider ranged and less critical than the individual BW poles.
Once the problem has been resolved the affected brds. will have to be realigned.
When correct, Cal signal centered, RL @ -7dBm, Lin mode, SWP auto, Span auto, stepping the
RBW from 1MHz to 10Hz the viewed signal should not shift side to side more than +/- 1 div. and vertically not more than +/- 0.5 div. I would typically have no movement in any direction. Also if you run the auto correction routine and look at the correction no¡¯s. the RBW amplitude corr. no¡¯s. will be 0.05 dB or less, and the RBW CF freq. correction no¡¯s would be typically 3% or less of the RBW selected.
Let me know what you find.
Best regards,
Don Bitters
|
Don, I don¡¯t have enough information from you, so let me ask you to do this:
Attach the four bypass networks to stages 2,3,4, and 5. Set the RBW to 1 kHz. This will leave only the first stage filter in the path. You should see a filter shape with skirts that go down perhaps 25 dB on either side. (Set the span wide enough, so you can see the filter skirts that far down.) The 3dB bandwidth will be around 2.3 kHz, I think.
What happens when you try to adjust the SYM capacitor in stage 1? Does it run out of range and it can never make the skirts reasonably symmetrical? Does it work well and you can adjust for symmetrical skirts? There is some interaction between the SYM and CTR adjustments, so iterate between the two until it looks like you have optimized both.
If you can adjust the first stage so the notch is gone, move the bypass network from the second stage to the first stage and repeat what you just did, this time on the second stage. This time, stages 1,3,4, and 5 will be bypassed and you will be observing? the second stage only.
You cannot adjust this filter correctly if you have more than one stage in the signal path.
Do this until you have checked all 5 stages. This should help you narrow down the problem to one stage.
Also, can you post a picture?
Vladan
|
hi Ed,
Can you identify, in detail, the specific Tantalum caps that were the problems, ?& ?what you replaced them with, ?to resolve these issues in the HP 85662a ??
thanks,?
|
Hmmmm, I hadn¡¯t thought about the caps, but I see them fail often in other units.? My problems with these filters turned out to be on the A4A9 card in the area that drives the filter selection.? Down side they are current sources, I swapped a card and it worked.? I left that card in and was just going to shotgun the parts. ? My symptoms were close to yours, I think,? I could adjust all of the sections but not to the correct place. ? Regards, ? Stephen Hanselman Datagate Systems, LLC 3107 North Deer Run Road #24 Carson City, Nevada, 89701 (775) 882-5117 office (775) 720-6020 mobile s.hanselman@... a Service Disabled, Veteran Owned Small Business DISCLAIMER: This e-mail and any attachments are intended only for use by the addressee(s) named herein and may contain legally privileged and/or proprietary information. If you are not the intended recipient, any dissemination, distribution or copying of this e-mail and any attachments is strictly prohibited. If you have received this e-mail in error, please immediately notify me and permanently delete the original and all copies and printouts of this e-mail and any attachments. ?
|
You may want to check all of the amplifier power and PIN driver circuits that select the crystal filter sections, to be sure the right states are properly decoded. The reason I suggest this is that I found a number of failed (shorted) Ta caps in the crystal filter sections in two of several 85662A IF/display units that I fixed a few months back. These were failures that crashed the -15V supply. In each spot, there is a series choke, and some were visibly overheated, but not burned open. There could be a scenario where a choke does burn open, so the power/bias in a section is lost, even though the unit appears operational. If this happens, the IFBW selection logic may be wrong for some settings, and there may be some unanticipated RF interaction between the sections, via the improperly biased PIN circuits.
Ed
|
Dave McGuire
