HP 8590A weird digital? bug/user mixer reference level error

[coppercone2]

So my HP8590A works as a spectrum analyzer but I caught it being funny a few times before.

1) i was looking at signals and for some reason a bunch of TV-band frequencies turned into a noise for some reason.. as if the SNR of the instrument suddenly got horrible until it was reset, so it looked like a clean but noisy display wheras before there was quite a few peaks there
2) The mixer level adjustment seems to 'latch'. If I set the dB/division low (say 1-2 dB/div) and I scroll the mixer too hard (i.e. the signal trace goes off scale on top of screen), then the mixer level cannot be moved back up, and I need to hit the reset button.

I thought that #2 might be something to do with a bad DAC? I am not sure why it would work fine moving the mixer level up and down fine, but if the signal goes off screen the mixer knob becomes unresponsive.

I got this instrument at a bargin price so there is probobly a fault with it.... but I am not even sure if its a memory problem, some kind of intrinsic problem with the analyzer, or some kind of analog problem with 'latch up' of some kind in the DAC..

So I turn on the instrument, turn on attenuation to 0dB in amplitude screen, put on 2 db / division, using keypad, so the signal is under the bottom of the display. Then I press the reference level button and adjust the trace using the knob, until the trace is shown on screen. Now, when I adjust the knob,m the number on the display changes but the trace does not move. If my velocity on the knob is high, then I can totally hide the trace before it 'locks control out'. If I click the reset button, it works fine.

But if I don't play with the attenuator setting, when I change the mixer level, I hear relays clicking and it never locks out, but it changes the attenuator level. It only locks out if I first set the attenuator. But if the trace gets frozen, I can move it up further, just cannot move it down.

Is this a intrinsic property of this spectrum analyzer? I thought I caught another one doing something like this before but I don't have that much experience with it. It feels like there are forbidden zones of operation. My more analog spectrum analyzer has some kind of rotary linkage that makes sure the knobs have the correct states in reference to each other, so you can't run it on silly settings.

But I do find it weird that the system locks out instead of warning you or whatever... (it feels broke to the user if you can suddenly only further hide a trace/have a lying display, since clearly the mixer level is not changing anything despite the numbers changing) Can anyone confirm with a 85xx instrument with a hardset attenuator? You don't need a signal or antenna for this, a bare instrument will do. If this is a design, they really should have put a 'invalid setting' notification pop up on the display. 

[SoundTech-LG]

So I turn on the instrument, turn on attenuation to 0dB in amplitude screen, put on 2 db / division, using keypad, so the signal is under the bottom of the display. Then I press the reference level button and adjust the trace using the knob, until the trace is shown on screen. Now, when I adjust the knob,m the number on the display changes but the trace does not move. If my velocity on the knob is high, then I can totally hide the trace before it 'locks control out'. If I click the reset button, it works fine.

But if I don't play with the attenuator setting, when I change the mixer level, I hear relays clicking and it never locks out, but it changes the attenuator level. It only locks out if I first set the attenuator. But if the trace gets frozen, I can move it up further, just cannot move it down.

This is the sequence you want tested out to confirm? When you say "the signal" what signal are you using, or no input at all?

[coppercone2]

by the signal I mean the trace shown on the screen. The analyzer is not hooked up to any source (the bug is the same with or without a source).

[SoundTech-LG]

OK, so that should be easy to confirm for you. I have both a 8560E, and 8562A, both functional as far as I know. Trouble is I can't access them until later today. Someone can step up to help you before that I bet.

[coppercone2]

no rush, this has been a problem for 5 years I just decided to do something now

[SoundTech-LG]

Hi,

So checking this for you.

"So I turn on the instrument, turn on attenuation to 0dB in amplitude screen, put on 2 db / division, using keypad, so the signal is under the bottom of the display."

That is not what I see. The TRACE, or as you say "The SIGNAL" does not move below the bottom of the display.

[SoundTech-LG]

I'll mention I used PRESET before your keystrokes.

[coppercone2]

Hmm.. so you have some kind of auto scaling?

I can hide the trace under neath the display if I go low enough.. it simply moves right on down as I crank the knob

[SoundTech-LG]

Using PRESET should put the unit in a known state. Did you do it?

[SoundTech-LG]

So you are saying when you are in the dB/Div mode you can move the TRACE up and down?

[SoundTech-LG]

So only the SIGNAL meaning the spectral noise, moves below the display. the TRACE is displayed at the bottom graticule line.

[SoundTech-LG]

"adjust the trace using the knob, until the trace is shown on screen. Now, when I adjust the knob,m the number on the display changes but the trace does not move."

This does not seem quite clear. First you say you CAN move the TRACE or SIGNAL (not sure which) THEN, you say the TRACE does NOT move...  Which is it???

[SoundTech-LG]

Down at about -60dB I can get a pretty full screen of noise. You do not get that?

[coppercone2]

the preset button makes it return to default settings so it can happen all over again.

When I turn it to 2db/div the spectrum trace disappears under the bottom of the display (not in view).

When I press the mixer level button, it says -10dbM after a preset. When it gets around -66dbm I can see the noise spectrum trace again

However, at this point, when I am at approx -60, I can bring it down lower, to say -70, but I cannot bring it back up to -60 again. It latches. But, on the screen numerical display, I can bring it back to -60, but.. the spectrum trace does not follow it anymore.So If I spin the knob too fast, the spectrum trace can get stuck near the top 20% of the screen and you can't see any signals... I have to dial it in slowly, or press the preset/reset button and try again.

Its like the up/down functionality breaks at a particular Dbm. Hell, I can bring it back to -10dbm, but... the spectrum trace (I don't know what the term is, the electron beam display of frequency vs power) is stuck on the screen wherever it landed lowest. I would say it was a broken encoder normally, but the damn on screen numerical display follows the knob just fine.... so IDK

I thought maybe its some kind of problem with a control DAC 'latching' in one direction. But its not like its totally stuck, if it gets stuck at -66dbm, I can bring it to -10dbm with no change, but if I go lower to say -80dbm, it does go down and hide under the screen again... so it seems like something to do with code more then hardware.


Its just like if your offset knob on the oscilloscope decided to work on only 1 direction after it got to a certain point, despite the offset indicator telling you that it should be moving up.


The mechanical analogy is that it turned into a ratchet for some reason. Like you have god damn lathe knob that when you get it to a certain point it activates a fucking ratchet. It seems totally bootleg. I dont understand electrically why this happens unless its some kind of protection circuit.. but then the fucking display indicator should not lie.


If its still unclear:
the knob is a ratchet on a bolt set to the middle setting where the ratchet mechanism is locked so you can spin it both ways. When you tighten it enough, for some reason the ratchet turns on, so you can only further tighten it but not loosen it, but the display numbers show that it is supposed to be getting looser (this would be a fancy ratchet with a turn indicator on it).....

[SoundTech-LG]

Part of the problem is your describing of the SIGNAL as the TRACE and vice versa. In this case, the SIGNAL is only broadband noise. The TRACE should always be visible, even if the SiGNAL goes completely off screen. If it's far enough off screen you will have to increase or reduce attenuation. The TRACE will be at the bottom graticule line with a weak, or over attenuated signal. The TRACE will be at the top graticule line with an overload, or under attenuated signal.

[coppercone2]

Ok, but why is it acting like a ratchet. If its supposed to work this way I don't think that HP would have you look at numerical lies (nomenclature aside).

I can move it up and down until it gets to that dbm level and then it only moves down.

[SoundTech-LG]

Not really sure at all about this ratchet bolt description. It's a Rotary Pulse Generator RPG... and maybe that is the issue. Seen a lot of those go bad.

[coppercone2]

Well I am not familiar with that circuit I thought there was some kind of DAC biasing something up and down. When I google Rotary pulse generator it looks like a rotary encoder to me.

The knob is fine because the display number in the left hand upper corner changes as I rotate it, but its lying. If it was the knob it would get not change the display number. The digital system is registering a change but it seems to be able to only bias the mixer in one direction after a certain point. I can't even imagine how that could happen really. I suppose I can tap a analog control line and bypass the DAC in there to see if its a RF problem or a digital problem, I should be able to move the trace up and down with a control signal connected to the right point, but obviously thats some what a pain in the ass. How does a DAC develop a ratchet type behavior?

[T3sl4co1l]

So you're doing...?

Initial power-on state / [PRESET]:
1500MHz SPAN, 750MHz CENTER, RBW 3MHz, SWP 20ms, etc., 10dB/div, 10 dB ATTEN.

[AMPLITUDE]
[LOG dB/DIV]
--> 2 dB/

[MAX MXR LEVEL]
--> below -56 dBm?

When I make this adjustment, I notice nothing shows (only the 0Hz pilot tone, which tends to drift on screen on my unit anyways) until -56.1.  Then nothing much changes until some other threshold, which, hmm maybe I'm seeing the sticky behavior you describe?  It seems to not go back down at the same rate.  After dipping it lower, now the visible threshold is -55.5dBm.  The next step occurs at -65.5dBm, towards the top of the screen.  It seems to push up smoothly now by -70dBm, and going back up, it seems to stay there.

Notice the REF (top left corner below hp) now shows, whatever, all the way to -90dBm if you cranked MXR LEVEL to minimum (-100dBm) it seems.  Press [AMPLITUDE] and set it back to 0.  Trace looks normal again.  (Well, as normal as it will, still at 2dB/.)

Tim

[coppercone2]

Yeah thats it, it must be a bug in the design.

I don't see any velocity stuff, it just gets stuck.

I wonder if there is a SMB connector some where that connects a voltage bias to the mixer that I can plug external supply into. I bet its like 1V full scale, or if its buried in a RF can.

[coppercone2]

Put it on 0 db attenuator first. If you set the attenuator manually when you scroll through the mixer it will no longer click the relays. When I don't mess with attenuator it behaves better.

Without the manual attenuator setting (the way you posted it), it will scale the attenuator depending on the mixer level and you can hear it click when you turn the knob.

I think when you manually set the attenuator it disables the auto-scaling attenuator feature.

[SoundTech-LG]

Here is a mention of the Rotary Encoder for the 8590...

https://www.eevblog.com/forum/testgear/agilentkeysight-esa-synthesizer-shaft-encoders/msg798255/#msg798255

[coppercone2]

how could it do anything with the encoder when the encoder makes the dbm display change?? the numbers track both ways just fine

[SoundTech-LG]

TRUE.

[SoundTech-LG]

But it sounded so weird to hear it described as a ratchet and bolt...