20th Jun 2015, 5:01 pm | #141 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: HP8568B calibrator noise / spurious signals
Quote:
If you look at my image below the blue region is a kind of dead zone where the analyser is doing lots of housekeeping including all the display updates etc and it is also trying to relock the YTO at the start of the Lock and Roll process. The next section is the flat green line shown by the big yellow arrow. This DC level is the DC 'error' voltage delivered from the phase detector and loop filter on the A6 module (after it has achieved phase Lock) and this signal gets summed in just ahead of TP3 on A22 at the summing amplifier U19 on the A22 module. This DC voltage effectively offsets the ramp voltage (the ramp is the third region) by the correct amount such that the 'Roll' part of Lock and Roll will always ramp the YTO (via the FM coil) such that the frequency sweep is centred quite accurately in the middle of the screen. In other words, this DC voltage in the second region (yellow arrow) is the steering voltage to calibrate the YTO at the centre frequency. In my second image you can see that this means this section of the waveform can be offset or shifted across quite a range as indicated by the series of green arrows and the dashed yellow traces. Note that I've guessed the range so it might not relect the true range. But you get the idea... Now you might expect that this means that the steering voltage should be consistent with centre frequency but this might not be the case. The error voltage from A6 corrects for more than just the drift in the FM coil. It also has to correct for drift in the main coil and also for drift in the free running LO2 cavity oscillator and other things as well. So (I think) you could see this DC voltage change for a given centre frequency even on a healthy instrument.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 20th Jun 2015 at 5:10 pm. |
|
20th Jun 2015, 8:56 pm | #142 | |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
Quote:
Probe is X10, and the scope readout does NOT auto compensate for X10 settings. I hope they are of some help, I am imagining the time between pulses should change, but should there be MISSING ones? The missing ones seem to occur when the SA screen noise spikes and jitter are at their worst. I soldered a short insulated wire to pin 7, U1 for the test. Thanks Jeremy. http://www.gatesgarth.com/A12.zip shows all the scope shots, note I changed time base settings to show more or less pulses part way through.
__________________
Best Regards, CW. 2E0ILY |
|
20th Jun 2015, 11:00 pm | #143 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: HP8568B calibrator noise / spurious signals
I'm a bit confused again by the scope plots of the lock flag as I was expecting it to be pulse free if the analyser was healthy. I think I need to look at what my HP8568B does here.
By the way, I found an old email address of yours and sent you a pasted together image of the full schematics of several modules including A6, A22, A26 and A12. I created an 8000 x 2000 image containing several merged schematic pages from the manual. There is a small loss of detail but hopefully it will be useful. I've drawn a few coloured lines on it to show the phase lock signal routing. David, if you would like a copy I can email you the merged image too? It's a GIF file that is about 700kb and is an 8000x2000 image that looks a bit like a Bayeux Tapestry
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 20th Jun 2015 at 11:07 pm. |
20th Jun 2015, 11:16 pm | #144 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,901
|
Re: HP8568B calibrator noise / spurious signals
I wouldn't mind it for my records, but I'm more involved with an 8566B which has quite a different LO locking system, which is why I've not been commenting much on the 8568. The 8566B is much more like what got moven into the 8560 series of portable analysers.
I'll PM my email David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
20th Jun 2015, 11:34 pm | #145 | |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
Quote:
Received the .gif, and yes, that's very useful indeed, thanks for sharing that!
__________________
Best Regards, CW. 2E0ILY |
|
24th Jun 2015, 4:11 pm | #146 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
I have had a thought, I have a few SDR dongles and a couple of Softrock receivers. the dongles go to UHF, the Softrocks only to 30 MHz. Could they be any use for diagnostics, used as a (very) poor man's SA? I am loathe to give up on this beast, I feel if we could just narrow down the section causing the noise I could perhaps find a replacement card if isolating to the component level is beyond me. Cheers!
__________________
Best Regards, CW. 2E0ILY |
1st Jul 2015, 1:23 pm | #147 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
Moved to testing at TP2 pins on A15 Controller board.
Surely all this noise is neither right nor good? X10 probe, same timebase and Volts per division for each shot. Instrument in 20MHZ centre frequency just over 1MHz span, although in instrument preset state things look the same.
__________________
Best Regards, CW. 2E0ILY |
2nd Jul 2015, 12:25 pm | #148 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
Another realisation. A22 frequency control board. TP16 "Stop" test pin. The voltage spikes here go from 0 to minus 15V or so when in a below 1MHz span with a 20 MHz centre frequency. But as soon as I change to a just *over* a 1MHz span, the pulses go from minus 15V up to zero volts. Does this give any more of a clue?
__________________
Best Regards, CW. 2E0ILY |
2nd Jul 2015, 2:25 pm | #149 |
Nonode
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,062
|
Re: HP8568B calibrator noise / spurious signals
None of the images look like a 'deliberate' waveform
The spikes look pretty narrow in the first set of pictures (at 100nS/div) - reminds me of the sort of thing you might see when there is a race condition but at those voltage levels it could only be CMOS, if its logic at all. dc |
3rd Jul 2015, 8:24 am | #150 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,901
|
Re: HP8568B calibrator noise / spurious signals
I agree.
I think you're moving around downstream of something which is going wrong. I'm most interested in the pictures you had of the ramp generator with abrupt jumps on the slope, transients rapidly taking the voltage off significantly and then returning it to where it should have stayed. This is not normal behaviour for a ramp generator and it is very difficult to make such things do this. Normal fault finding is to find something doing something it shouldn't, and then to ask what is causing it and keep moving upstream. When a circuit block has multiple inputs (and remembering that bad loads on outputs are a form of input) then check all inputs until you find a circuit block whose odd output doesn't have any odd input to explain it. It gets harder where loops are concerned, but it's still a necessary approach. If the digital side of things is acting silly, then which connection into the analogue stuff is the silliness passing over? David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
3rd Jul 2015, 12:29 pm | #151 | |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
Quote:
These are the pulses on a faster trigger
__________________
Best Regards, CW. 2E0ILY |
|
5th Jul 2015, 1:54 pm | #152 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
OK, I think I am more confident the pulse signals on U1 pin 7 are correct in spans under 1MHz where the YTO UNLOCK LED is flashing. If I set the machine into instrument preset mode, the go SHIFT - FREE RUN, this puts the machine into operation stopping all calls to U1 pin 7 from anything unlocked, "Inhibits all phase lock flags", as far as I understand things. I then see no pulses on pin 7 whatever span I put it in.
However, the noise spikes are still present on the other points, as before. This presumably means they are not the result of phantom or real phase phase unlock calls? Someone abroad has *VERY* kindly offered me the loan of some boards from his spare parts HP8568B. He is abroad, so I was hoping to minimise his trouble and the postage fees and at least narrow things to the board level, but I am struggling to even do that, so any advice as to which boards would make a sensible call to try swapping over very welcome. As is advice as to how to probe the digital section further, as I am becoming more convinced we need to be "back there" now.
__________________
Best Regards, CW. 2E0ILY |
8th Jul 2015, 12:12 am | #153 |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: HP8568B calibrator noise / spurious signals
Hi Chris
I've been very busy for a couple of weeks and have not been able to find time to relook at my 8568B but I had a look tonight at the phase lock pin 7 on U1 and it behaves the same as yours except I don't see any missing pulses on spans above 1MHz. However, having seen your A15 images I think there is something horrible happening on this board. Even the AS (address strobe?)pin from the CPU looks to have these glitches and they are happening way faster than the normal CPU clock speed. Can you check to see if the 14.xx MHz CPU clock is clean on A15? Eg look at the U1 oscillator output. Also look at the output of the flip flop chip at TP2-4. Also check that the +5VF rail is OK. Also, how dusty/dirty is your A15 card? Mine was covered in really thick dust when I checked the battery condition. Normally, dust wouldn't cause issues but I was shocked how caked in dust mine was.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 8th Jul 2015 at 12:19 am. |
8th Jul 2015, 10:42 am | #154 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
No problem at all, everyone is more than generous with their time and knowledge here, and it's much appreciated!
I have a terrible confession to make, and it embarrasses me deeply. I got fed up with the thing and put the covers back on and slid it into its rack cabinet. It had sudden;y started behaving itself, for no obvious reason, and I had not seen display noise for a couple of days, with it continually running, and me looking at it whenever I was around. I had hopes of a bolt from heaven having come down and sorted it. Then, it happens again, I go into my shack and it's showing a load of noise again. I'd had enough and buttoned it up. In the calm of another day I thought I would do something else and check the cal on my old Tektronix 7854 scope I'd been using. I'd bought a time / mark generator for next to nothing at a rally and thought I'd check the scope out with it. I cannot describe my abject horror when I used the acquire function and saw noise spikes were stored along with the trace... Surely the damned scope wasn't creating its own noise spikes Seems it is.... I posted some screen shots on the Tektronix forum and the experts there have seen this before. The higher the sweep speed the more likely identical looking spikes to those in many of my screen shots here show up. This tallies with what a known expert on this thing says on the Tek forum: "More important is to use a higher frequency signal which needs a faster sweep speed. Then the 7854 will spend more time during retrace when the horizontal position is invalid and false samples will be more common and easier to see." There's an adjustment that's simple to make and may fix this, so I'll give it a go and see what happens. Despite the fact the sig gen and time / mark gen both show noise spikes on the trace when in acquisition mode, they don't seem as chronic as when testing the SA. The fact nearly every point I scoped showed noise suggests this may well be a scope issue, in hindsight, *BUT* A15 TP2-* points show bad noise even in free running mode. In free running mode I never saw the noise spikes on every other test point in the SA, only in acquisition mode. So maybe this A15 board is the source of my trouble? I am not sure of 2 things: Should it be necessary yo use a scope with storage to see noise in the SA circuits at all? Can I use a conventional analogue or digital scope and expect to see it? On A15 TP-3 what sort of signal should I be seeing, and please, what does "AS" on that pin mean in simple terms? I can only humbly apologise that I have posted what are probably an interminable number of bum scope shots, and crawl under the nearest stone. It never dawned on me that the scope was the source of much of the noise. This is becoming a bit like the nursery song, "There's a hole in my bucket". I am very sorry!!! Jeremy: A15 and all cards were very clean, no dust. The machine was very clean inside when i got it, and I changed the filter element on the fans for dual density foam as used in F1 race engine air filters. It's very dear, very good and very un-restrictive to air, yet very restrictive to dust and dirt particles. It seems to be doing a good job. The two clocks show clean waveforms, no problems I can see. I think TP2-4 showed noise, like TP2.-3, even in free running scope mode. When the scope suddenly seemed to behave itself I had a think as to what the last thing I did was. I believe it was soldering a test lead to the +5V side of the 100uF cap on the 5V rail for A15. I had pulled one leg and it tested as good with my Peak ESR tester, but I decided to look at it on the scope as well. The scope showed no abnormalities I could recognise. Probably coincidence? If you haven't ostracised me forever, is my Philips analogue / digital 100 MHz "Combiscope" adequate for testing the SA? Do I *need* a scope with an acquisition mode?
__________________
Best Regards, CW. 2E0ILY |
8th Jul 2015, 2:15 pm | #155 |
Moderator
Join Date: Mar 2012
Location: Fife, Scotland, UK.
Posts: 22,901
|
Re: HP8568B calibrator noise / spurious signals
What makes you think we don't all do this sort of thing from time to time?
David
__________________
Can't afford the volcanic island yet, but the plans for my monorail and the goons' uniforms are done |
8th Jul 2015, 2:38 pm | #156 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
That's very kind of you, I'd like to know what you *really* muttered when you read this
__________________
Best Regards, CW. 2E0ILY |
8th Jul 2015, 4:05 pm | #157 |
Nonode
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,062
|
Re: HP8568B calibrator noise / spurious signals
Been there, done that !!
Its all too easy to make assumptions, occasionally they will come back to bite you but you still have to make them or you would never get anywhere at more than a snails pace. When you seem to arrive at a dead end / paradox THEN its a good time to start thinking what you assumed to be true that might not be ... dc |
8th Jul 2015, 4:22 pm | #158 |
Heptode
Join Date: Nov 2011
Location: Whitchurch, Shropshire, UK.
Posts: 612
|
Re: HP8568B calibrator noise / spurious signals
I have to say I thought it odd the noise seemed to be everywhere where there was a signal. Yet apart from A15 it seemed to be of equal sort of magnitude. Maybe A15 is the culprit. Hopefully I can either use my Philips scope of fix the Tek such that it can be trusted! Still unsure as to whether tracing noise like the SA shows needs a scope that has acquisition, or if it should show on a normal free running analogue scope, though?
__________________
Best Regards, CW. 2E0ILY |
9th Jul 2015, 9:46 am | #159 |
Nonode
Join Date: Jan 2009
Location: Bristol, UK.
Posts: 2,062
|
Re: HP8568B calibrator noise / spurious signals
I'm sure the Philips will do the job, but you may have to sketch the wave forms for posting unless you have a good digital camera!
With a non store scope I sometimes find I 'know' something untoward is happening because an otherwise stable triggered display 'flickers'. This can be due to occasional superimposed glitch (in a dark room you may even catch sight) or you see an occasional trigger jitter as the time-base fires on the edge of a glitch (especially when twiddling the trigger mode and level). Sometimes a scope with a very fine trace can identify pretty tiny amounts of noise as the trace will slightly widen. Occasionally I have actually found my ancient Tek analogue storage scope useful but I usually already know there is something bad going on Digital systems are a different ball game, if the glitch occurs 1 in a million on a 10MHz clock you are never going to see it with an analogue scope. If I was looking at such systems I would probably buy myself one on those nice new rigol scopes that look the absolute business for the price of a good scope probe ! dc |
11th Jul 2015, 3:08 pm | #160 | |
Dekatron
Join Date: Sep 2010
Location: Cheltenham, Gloucestershire, UK.
Posts: 3,077
|
Re: HP8568B calibrator noise / spurious signals
Quote:
The AS pin is the address strobe pin and I think that this is a signal that indicates that the address lines are now holding a valid address. AS is involved in both memory reads and also memory writes. So for the commonest synchronous and asynchronous modes of operation this signal should look like a clean pulse. I've not looked at my AS line but I expect it to pulse low every few clock cycles for each time the CPU tries to read a fresh address. If the CPU was getting a lot of rogue pulses here and at DTACK I would expect the system to crash a lot. So I suspect that your scope is the problem here. Maybe best to take a step back and then 'prove' a few things to rule more things out. For example, I'd still like to know how stable your LO2 signal is. This is a free running cavity oscillator that lives inside the main PLL. So the LO2 signal has to be able to achieve a certain degree of short term stability or the PLL won't be able to track and cancel out the LO2 drift and noise. It cancels out the LO2 drift and noise by 'modulating' the rest of the PLL with a cancelling signal so the overall phase noise of the analyser becomes incredibly low despite the noise in LO2. So the cleaner and more stable the LO2 signal is the better because the PLL has to cancel this LO2 drift and noise elsewhere in the PLL system. Do you still have that EIP counter? Can it measure LO2 at the little SMB test connector? Is it fast enough to display any sudden step changes in the LO2 frequency? eg if LO2 suddenly changed by 100kHz during a sweep then the PLL would have a tough time correcting this cleanly when in sub 1MHz spans where the PLL is locked for a whole sweep. The ideal tool to use here would be another spectrum analyser. Or maybe a low frequency analyser fed from a mixer and sig gen that downconverts the 1748MHz LO2 signal to something the other analyser can display.
__________________
Regards, Jeremy G0HZU Last edited by G0HZU_JMR; 11th Jul 2015 at 3:16 pm. |
|