|
Hi all,
Some of you will be aware I was trying to find the source of the YTO unlock error I was getting displayed on the screen of my 8566B. I'd been advised elsewhere to scope the power rails for any signs of instability and at that time, aside from some small variations in the expected voltage levels, all seemed fine. However, I wasn't happy I'd got the right scope for the job and the optimum settings for this kind of check, so I've had another try and discovered something seriously amiss. There is some small level of noise on the DC rails, but no more than I would expect from looking at *any* power rail. The issue that jumped out this time (having got the right tool for the job; setting right etc.) is the presence of persistent, random, voltage spikes. The excursions from the DC value go at least 60mV plus and minus from the DC value. They're really skinny if that's any clue; so narrow they're sometimes hard to see. But they ARE there nevertheless. I have two questions here:
1. Could these spikes cause the YTO unlock condition?
2. What could cause these spikes, given that they're present on ALL the DC rails: -40V, -12V, -5.2V, +5.2, +12V, +22V and so on?
I'm pretty sure this isn't operator-induced error, but I guess I can't altogether rule out the vague possibility it may be an internal fault in the oscilloscope itself. But let's assume for now the spikes do not originate within the 'scope itself.
Thanks.
|
Jinxie,
Have you measured with the earth clip directly connected to the
probe's tip? Close to the area you are looking at. You will be
surprised what you see. As you say you measure the same at any
power rail its probably some stray field omni present. The earth
clip to the tip makes a good pick up loop....
My best, 73 de
Harke
On 27/02/2022 13:58, Jinxie wrote:
toggle quoted message
Show quoted text
Hi all,
Some of you will be aware I was trying to find the source of the
YTO unlock error I was getting displayed on the screen of my
8566B. I'd been advised elsewhere to scope the power rails for any
signs of instability and at that time, aside from some small
variations in the expected voltage levels, all seemed fine.
However, I wasn't happy I'd got the right scope for the job and
the optimum settings for this kind of check, so I've had another
try and discovered something seriously amiss. There is some small
level of noise on the DC rails, but no more than I would expect
from looking at *any* power rail. The issue that jumped out this
time (having got the right tool for the job; setting right etc.)
is the presence of persistent, random, voltage spikes. The
excursions from the DC value go at least 60mV plus and minus from
the DC value. They're really skinny if that's any clue; so narrow
they're sometimes hard to see. But they ARE there nevertheless. I
have two questions here:
1. Could these spikes cause the YTO unlock condition?
2. What could cause these spikes, given that they're present on
ALL the DC rails: -40V, -12V, -5.2V, +5.2, +12V, +22V and so on?
I'm pretty sure this isn't operator-induced error, but I guess I
can't altogether rule out the vague possibility it may be an
internal fault in the oscilloscope itself. But let's assume for
now the spikes do not originate within the 'scope itself.
Thanks.
|
On 2/27/22 07:58, Jinxie wrote: Some of you will be aware I was trying to find the source of the YTO unlock error I was getting displayed on the screen of my 8566B. I'd been advised elsewhere to scope the power rails for any signs of instability and at that time, aside from some small variations in the expected voltage levels, all seemed fine. However, I wasn't happy I'd got the right scope for the job and the optimum settings for this kind of check, so I've had another try and discovered something seriously amiss. There is some small level of noise on the DC rails, but no more than I would expect from looking at *any* power rail. The issue that jumped out this time (having got the right tool for the job; setting right etc.) is the presence of persistent, random, voltage spikes. The excursions from the DC value go at least 60mV plus and minus from the DC value. They're really skinny if that's any clue; so narrow they're sometimes hard to see. But they ARE there nevertheless. I have two questions here:
1. Could these spikes cause the YTO unlock condition?
2. What could cause these spikes, given that they're present on ALL the DC rails: -40V, -12V, -5.2V, +5.2, +12V, +22V and so on?
I'm pretty sure this isn't operator-induced error, but I guess I can't altogether rule out the vague possibility it may be an internal fault in the oscilloscope itself. But let's assume for now the spikes do not originate within the 'scope itself. As Harke said a few minutes ago, look at probe grounding. This is often dismissed as one of those "That can't happen!" things, but it's a very common and widely-misunderstood problem. Always connect your probe's ground as close to the monitored circuit node as possible, and keep the probe's ground wire as short as possible. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
I'm not too sure what Harke means. I don't see what shorting the tip to the ground is going to achieve.
I'm probing A17 and A18 boards and there is simply nowhere other than nearby aluminium chassis to clip the ground lead to. I appreciate what you say about keep the loop as short as possible, but the only alternative would be to pull the boards and solder fly-leads directly to the closest ground pours on the PCBs themselves. It's a pity HP didn't provide convenient, accessible ground pins for this purpose!
|
On 2/27/22 12:41, Jinxie wrote: I'm not too sure what Harke means. I don't see what shorting the tip to the ground is going to achieve.
I'm probing A17 and A18 boards and there is simply /nowhere/ other than nearby aluminium chassis to clip the ground lead to. I appreciate what you say about keep the loop as short as possible, but the only alternative would be to pull the boards and solder fly-leads directly to the closest ground pours on the PCBs themselves. It's a pity HP didn't provide convenient, accessible ground pins for this purpose! (It would be helpful to quote the email that you're replying to. Hint: This is a mailing list, not a web forum.) I don't mean this the way it's going to sound, but most people don't see what it's going to achieve. Think of it this way. Visualize the probe tip, ground clip, and ground wire as a "pickup loop", an antenna of sorts. Anything nearby can induce a current in that loop. "Ground" is a "virtual" entity in electronics; it means nothing other than "reference point of the moment". Everything is relative to "ground", but what "ground" means is relative. Your oscilloscope is showing a graph of voltage against "ground". Any voltage at the probe tip will cause Y-axis deflection on the oscilloscope, but variations in the voltage at the ground lead, with constant voltage at the probe tip, will also cause Y-axis deflection, because it's graphing the relative voltage between the two. If current is induced in the ground wire, the oscilloscope's idea of "ground" changes, and you'll see Y deflection on the oscilloscope. I'm not explaining things very well today, so I hope the concept is clear. -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
Might not be a bad idea to look up and review Bob Pease's article(s) on this. Grounding was a particular concern of his and he wrote several very good articles explaining the concepts.
Cheers!
Bruc
Quoting Dave McGuire <mcguire@...>:
toggle quoted message
Show quoted text
On 2/27/22 12:41, Jinxie wrote:
I'm not too sure what Harke means. I don't see what shorting the tip to the ground is going to achieve.
I'm probing A17 and A18 boards and there is simply /nowhere/ other than nearby aluminium chassis to clip the ground lead to. I appreciate what you say about keep the loop as short as possible, but the only alternative would be to pull the boards and solder fly-leads directly to the closest ground pours on the PCBs themselves. It's a pity HP didn't provide convenient, accessible ground pins for this purpose! (It would be helpful to quote the email that you're replying to. Hint: This is a mailing list, not a web forum.)
I don't mean this the way it's going to sound, but most people don't see what it's going to achieve. Think of it this way.
Visualize the probe tip, ground clip, and ground wire as a "pickup loop", an antenna of sorts. Anything nearby can induce a current in that loop. "Ground" is a "virtual" entity in electronics; it means nothing other than "reference point of the moment". Everything is relative to "ground", but what "ground" means is relative. Your oscilloscope is showing a graph of voltage against "ground". Any voltage at the probe tip will cause Y-axis deflection on the oscilloscope, but variations in the voltage at the ground lead, with constant voltage at the probe tip, will also cause Y-axis deflection, because it's graphing the relative voltage between the two.
If current is induced in the ground wire, the oscilloscope's idea of "ground" changes, and you'll see Y deflection on the oscilloscope.
I'm not explaining things very well today, so I hope the concept is clear.
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
|
On Sun, Feb 27, 2022 at 07:49 AM, Dave McGuire wrote:
As Harke said a few minutes ago, look at probe grounding. This is
often dismissed as one of those "That can't happen!" things, but it's a
very common and widely-misunderstood problem.
Always connect your probe's ground as close to the monitored circuit
node as possible, and keep the probe's ground wire as short as possible.
Jinxie, Your spikes do sound a lot like what you'd see from excessive probing inductance caused by a large loop area from the ground lead. One simple test I often do to determine whether my probing is creating high-frequency content in the measured signal is to manually decrease the loop area by pushing the ground lead wire up against the oscilloscope probe body (basically, you want to minimize the area created by the main part of the probe and ground lead). If this decreases the amplitude of the spikes, that's a good indication that the high-frequency content is a result of poor probing and not a signal present in the DUT. Matt
|
Okay, that's all understood now and thanks for the clarification, guys.
In order to properly eliminate the possibility of these spikes being a red herring I'm going to have to get creative with the probe grounding, clearly. This is where an extender board would be really useful. What a pity I don't have one!
|
On 2/27/22 16:59, Jinxie wrote: Okay, that's all understood now and thanks for the clarification, guys.
In order to properly eliminate the possibility of these spikes being a red herring I'm going to have to get creative with the probe grounding, clearly. This is where an extender board would be really useful. What a pity I don't have one! You could try slipping a few ferrite beads on the probe's ground lead. I'm not sure that'd help, but it might; does anyone else have any thoughts on that? -Dave -- Dave McGuire, AK4HZ New Kensington, PA
|
If the spikes are at line frequency, then I'd suggest turning off
LED or Fluorescent lamps and see if it makes a difference.?
You could always remove the ground wire, and use the probe as a
sniffer to see if the radiation is more obvious at any place in
the equipment.? Or shop, for that matter.
an experiment would be to wrap the ground line around the probe
body and then ground the tip.
Another might be to wrap the probe (with grounded tip) in the old
standard aluminum foil, that may or may not tell you something.
If, of course, you put a shorting plug on the scope's input
instead of a probe and still get those spikes, that tells you
something quite different.
Harvey
On 2/27/2022 4:59 PM, Jinxie wrote:
toggle quoted message
Show quoted text
Okay, that's all understood now and thanks for the clarification,
guys.
In order to properly eliminate the possibility of these spikes
being a red herring I'm going to have to get creative with the
probe grounding, clearly. This is where an extender board would be
really useful. What a pity I don't have one!
|
First, make sure that you don¡¯t have any ground loops.? I suggest you to power the 8566 and your scope from the same outlet.? ? Best, Francesco K5URG ?
toggle quoted message
Show quoted text
From: [email protected] < [email protected]> On Behalf Of Jinxie Sent: Sunday, February 27, 2022 4:00 PM To: [email protected]Subject: Re: [HP-Agilent-Keysight-equipment] Scoping the Power Rails [8566B] ? Okay, that's all understood now and thanks for the clarification, guys.
In order to properly eliminate the possibility of these spikes being a red herring I'm going to have to get creative with the probe grounding, clearly. This is where an extender board would be really useful. What a pity I don't have one!
|
Many thanks, guys. Seems like the consensus is to look into the ground issues. It's getting very late in my part of the world right now so I shall retire and take a fresh look at these things Tuesday (as I'm busy all day Monday). I'll let y'all know how it goes.....
|
Dave,
Adding ferrite beads to the ground lead itself will probably make things worse.??
However, threading the entire probe cable several times through a large ferrite can often help reduce noise due to ground differences between the oscilloscope and the unit under test.? A large, short, separate ground strap between the chassis of the two units also helps.
The "ferrite-turns-on-the-probe-cable" method was shown to me by a designer of multi-kilowatt vehicle drive inverters.
--John Gord
toggle quoted message
Show quoted text
On Sun, Feb 27, 2022 at 02:05 PM, Dave McGuire wrote:
You could try slipping a few ferrite beads on the probe's ground
lead. I'm not sure that'd help, but it might; does anyone else have any
thoughts on that?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
|
I've be watching this chain with someinterest. The thing that occurs to me is why are we all helping him go down the rabbit hole?
I've owned or worked on something like 25 8566 A and B instruments, many with unlock problems and never seen a case in which low level power supply noise caused unlock - in might aggravate the problem but never seen anything like the noise discussed affecting YO Lock If the voltages are within 5% the power supply is unlikely the problem.
Here are some questions that should ve asked (Forgive me if they have already been asked)
1) Are the loops MN and 20/30 subsections all locked?
2) Is the YO loop locked (There is a green LED on the YO loop interconnect board - all lock indicator are visible with the cover removed - no other disassembly required).
3) Does the Unlock message appear at all frequencies (Set scan width to 0 and use RPG to vary center frequency - see if loop as always unlocked or if sometimes it locks)
4) Have you used the frequency diagnostics and measured the YIG output frequency and power? (somm dissesembly required)
5) IMO, If all the LEDS in the M/N and 20/30 sections are lit and the YO LED is not, suspect the coupler (directly on top of the YO interconnect board.
6) Assuming M/N and 20/30 are OK, it would be a good idea to check the voltages AT the YO interconnect board before starting any disassembly.
7) It would also be a good idea to confirm the M/N, 20/30, and YIG frequencies (available using the front panel frequency diagnostic) before resorting to disassembly.
Unlock problems are generally relatively simple to tack down if you follow the procedure in the diagnostics section.
IMO using a scope on the DC rails can be problematic for exactly the reason you are having problems - I generally use a good RMS voltmeter to confirm PS ripple -there is a lot of noise floating around the environment and separating real from induced can, as you are finding out, can be a difficult problem. One thing you might do to eliminate chassis ground problems is to use the ground on the sweep card (forgotten wheather it is the sweep control or YIG driver, but one of them has a connection to the backplane ground plane and doesnot require connection to the chassis.
My 2 cents worth and I'll defer to Don B if he cares to comment because he has had WAY MORE experience than I have.
Cheers!
Bruce
Quoting Matt Huszagh <huszaghmatt@...>:
toggle quoted message
Show quoted text
On Sun, Feb 27, 2022 at 07:49 AM, Dave McGuire wrote:
As Harke said a few minutes ago, look at probe grounding. This is
often dismissed as one of those "That can't happen!" things, but it's a
very common and widely-misunderstood problem.
Always connect your probe's ground as close to the monitored circuit
node as possible, and keep the probe's ground wire as short as possible.
Jinxie,
Your spikes do sound a lot like what you'd see from excessive probing inductance caused by a large loop area from the ground lead. One simple test I often do to determine whether my probing is creating high-frequency content in the measured signal is to manually decrease the loop area by pushing the ground lead wire up against the oscilloscope probe body (basically, you want to minimize the area created by the main part of the probe and ground lead). If this decreases the amplitude of the spikes, that's a good indication that the high-frequency content is a result of poor probing and not a signal present in the DUT.
Matt
|
Ah ok, thanks for this info! I've made a note of this.
-Dave
toggle quoted message
Show quoted text
On 2/27/22 19:32, John Gord via groups.io wrote: Dave,
Adding ferrite beads to the ground lead itself will probably make things worse.
However, threading the entire probe cable several times through a large ferrite can often help reduce noise due to ground differences between the oscilloscope and the unit under test.? A large, short, separate ground strap between the chassis of the two units also helps.
The "ferrite-turns-on-the-probe-cable" method was shown to me by a designer of multi-kilowatt vehicle drive inverters.
--John Gord
On Sun, Feb 27, 2022 at 02:05 PM, Dave McGuire wrote:
You could try slipping a few ferrite beads on the probe's ground
lead. I'm not sure that'd help, but it might; does anyone else have any
thoughts on that?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
--
Dave McGuire, AK4HZ
New Kensington, PA
|
For what it's worth. Here was my problem 
|
That's what I was thinking.? A ferrite bead is an inductor that is purposefully made to be lossy.? It absorbs high frequency energy while passing DC and low frequency energy.
Assuming the probe cable can be bent into loops without damage, you're probably onto something, John.
Jim Ford
Sent from my T-Mobile 4G LTE Device
toggle quoted message
Show quoted text
-------- Original message -------- From: "John Gord via groups.io" <johngord@...> Date: 2/27/22 4:32 PM (GMT-08:00) Subject: Re: [HP-Agilent-Keysight-equipment] Scoping the Power Rails [8566B]
Dave, Adding ferrite beads to the ground lead itself will probably make things worse.?? However, threading the entire probe cable several times through a large ferrite can often help reduce noise due to ground differences between the oscilloscope and the unit under test.? A large, short, separate ground strap between the chassis of the two units also helps. The "ferrite-turns-on-the-probe-cable" method was shown to me by a designer of multi-kilowatt vehicle drive inverters. --John Gord On Sun, Feb 27, 2022 at 02:05 PM, Dave McGuire wrote:
You could try slipping a few ferrite beads on the probe's ground
lead. I'm not sure that'd help, but it might; does anyone else have any
thoughts on that?
-Dave
--
Dave McGuire, AK4HZ
New Kensington, PA
|
?I'm not too sure what Harke means. I don't see what shorting the tip to the ground is going to achieve. Wideband oscilloscope measurements seen magically selective to exclude all interference. You¡¯re not only measuring the voltage you are after, but also whatever voltage got induced in the exposed probe circuit, that the clip wire is a large part of. By shorting the probe tip to the ground clip, you know what voltage you expect to see: a flat line at zero. My hunch is that you won¡¯t see a flat line at zero. That¡¯s where real life differs from idealizations. The wider the measurement bandwidth, the crazier it can be. So, one problem is induction in the exposed loops of circuit attached to the probe, including ground clip lead. Another problem is common-mode to differential-mode conversion. Probes aren¡¯t well balanced differential circuits at all. A classical demonstration is to remove the ground clip wire and wrap copper or aluminum tape around the head of the probe to short the tip to the grounding sleeve immediately above it. You¡¯ll see a nice flat line without spikes as long as you are away from strong pulsed magnetic fields. And then touch this shorted probe tip to an earthed point on the device under test. You¡¯ll be surprised. And this CM to differential conversion can happen just fine with good mains isolation transformer used to power the DUT, even with electrostatic shielding between the windings. When isolated, you can touch one such wrapped probe tip to any point in the circuitry and observe varying levels of CM->diff conversion. It¡¯s almost like magic the first time you see it. At least now you know what to Google to learn more about it :) Cheers, Kuba
|
@Bruce
I only had very little time this evening to investigate further, I'm afraid. Not even enough time to try shortening the ground clip lead! I will try out everyone's suggestions in due course, however.
As things stand, I was able to observe that the screen is showing three errors in all now: Battery, YTO Unlock and REF Unlock. I don't recall that last one being there previously. There are three LEDs lit on the stack of modules criss-crossed by the press-on patch leads. I think these are the ones you were referring to:
ON=PLL3 Lock
ON=PLL1 Lock
ON=HET
As I say, all three are lit. Can't see any LED for PLL2, though!
After trying everyone's suggestions I plan to work through the steps suggested in the troubleshooting & diagnostics manual. It's probably going to be above my pay-grade, but nothing to lose by at least trying I guess.
|
So the REF unlock will be a major problem - you need to fix that before addressing the YTO unlock. Be sure the internal REF (10 MHz) oscillator is on frequency and that the INT/EXT switch is set properly and there as a good BNC connection between the INT and EXT reference BNCs (this assumes you are using the internal reference. I thought there was an LED for the M/N section (I'll have to check - don't have the cover off an 8566 right now) - or you could post a close up of the REF + M/N modules - As I remember, there is at least one LED there. The REF unlock is much more likely the problem than any power supply noise. Cheers! Bruce Quoting Jinxie <paul666@...>:
@Bruce
I only had very little time this evening to investigate further, I'm afraid. Not even enough time to try shortening the ground clip lead! I will try out everyone's suggestions in due course, however.
As things stand, I was able to observe that the screen is showing three errors in all now: Battery, YTO Unlock and REF Unlock. I don't recall that last one being there previously. There are three LEDs lit on the stack of modules criss-crossed by the press-on patch leads. I think these are the ones you were referring to:
ON=PLL3 Lock
ON=PLL1 Lock
ON=HET
As I say, all three are lit. Can't see any LED for PLL2, though!
After trying everyone's suggestions I plan to work through the steps suggested in the troubleshooting & diagnostics manual. It's probably going to be above my pay-grade, but nothing to lose by at least trying I guess.
|
Jinxie
Dave McGuire