开云体育

I am trying to correct a scan non-linearity in the 492 spectrum analyzer.
This is a picture of the display using the calibration procedure for the horizontal sweep.
/g/TekScopes/photo/270207/3340785
The 10ms pulses are expanded on the left of center, OK to the right up to the last two which are compressed.
The calibration has a single gain setting (R1055) but my problem is a linearity issue.
The deflection amplifiers (schematic 26) have a couple of trim capacitors (C5021 & C4057). Would these be the adjustments or is it likely that another component is failing?

The 'horzizontal out' sawtooth (generated before the deflection amplifier) appears to be a linear ramp but that is just by eyeballing.

Help appreciated!

Michael

Michael,

See if your board has two 620,000 1/2W carbons. If your has these, replace them with 2W types. Using a 1% will make sure better accuracy and low thermal drift. These are available on ebay. These two resistors are known to go bad. This site will explain more if you need it: . The board photo is on page 9 of the pdf. I replaced these in my 494AP. I replaced the two 300,000 ohms on the same board with 2W 1% types. These resistors I mounted off the board to allow better air flow around them. I would like a schematic of the one I have.

Mark

Actually, when doing a previous repair I had bought these parts but not put them in. It makes sense that this might be an issue since they are in the feedback path. Thanks for the reminder, I'll try that.

I replaced the resistors... but it didn't help 8-(

Michael,

DISCLAIMER: I know nothing about the instrument in question, and am just reading the service manual PDFs on the TekWiki, so my comments may be completely off base.

The problem, if I understand your description (and the associated image) is that the sweep is too fast at the start (the time marks are expanded), and too slow at then end (time marks compressed)? That sounds to me as if the sweep signal has "too much" of the normal charge curve of a capacitor, rather than just the early portion that should look very linear. If the sweep signal were being generated in a very simple way (just charging a capacitor with a constant voltage through a current limiting resistor) we would expect that this would be caused either by the resistor having decreased in value, and thus charging the capacitor faster, or by the capacitor having decreased in value (and charging up to the non-linear region sooner).

The description of the sweep generator in the service manual indicates that the current source for the integrator is more complex than just a current limiting resistor, so I doubt that's the issue (but I'd try to check it, nevertheless). It's possible that the timing capacitors are off (C2094 2ms-100ms, C2098 20us-1ms, and C3079 200ms-10s), but it seems more likely that only one of them would be bad. Since you are seeing this with 10ms pulses I would suspect C2094 may be bad (or maybe Q3095 which feeds it?).

I also notice that there is a trim pot (R5105) in the current source labeled "SWP ACCURACY" which sounds like it may be of interest.

-- Jeff Dutky

On Sat, Nov 27, 2021 at 01:09 AM, Jeff Dutky wrote:

R5105

I'm pretty convinced that it is an issue with the deflection amplifier. According to the calibration setup, in manual sweep, the voltage at the test point TP1061 is -5V when the cursor is at the left most graticule, 0 in the middle and right most graticule. I can adjust the center and right to be correct but this means that when I move the dot to the right most graticule it measures -4.7V.

I can get the 'horizontal out signal' (which is tapped before the deflection amplifier) and apply it to to the X drive of an oscilloscope in XY mode and it is linear in time. (externally triggering the 492 with the waveform generator). This implies that the sweep generator is linear and
X deflection amplifier (or the tube!) is non linear. I'll try to plot the input voltage against the right and left HV drives to see if this shows me what side of the driver is faulty.

..that should read +5V at the right most graticule.

That sounds much more likely than my suggestion, but analyzing the amplifier is outside my skill / expertise. Would there need to be trimming circuits to adjust the linearity of the horizontal amplifier, or to adjust for nonlinearity in the CRT?

— Jeff Dutky

Have you tried replacing U6061? TP1061 is the output of that op amp, and there's no component in the feedback path that could have drifted, which implicates the op amp itself. An LM318 would be a cheap and easy fix, if that's the failed component.

There's also a voltage divider (R6058/R6052) and bypass cap (C5058) on the non-inverting input of U6061. I wonder what the effect would be if any of those resistors or the capacitor were off spec or failed.

-- Jeff Dutky

There's several ways an amplifier can go non-linear.

One takes a perfect amplifier and then rolls off the? frequency response, so not enough HF.? This could produce a sweep that is fine at low sweep speeds, and starts to get odd at higher sweep speeds.? If you can't do anything with the sweep speed, then that's a non-issue for now.

The other is more subtle.

What Tektronix did on some of the scopes (and likely they used it a lot) is a circuit that looks like this:

Somewhere you have a sweep generator, producing a sawtooth. Check that for linearity with another scope and look at the timing pulses.? If the sweep is linear, then....

Tektronix used an emitter coupled differential amplifier.? lt's an amplifier that has an A and B input, and produces at the output an out of phase and an in phase copy of the input.? Tek then ran this through an amplifier that had cross coupling to give you compensation and gain control.? The output of that amplifier drove each deflection plate, the + and the -.? They typically drove one input, and signal wise, grounded the other.? This allowed one single polarity sweep signal to be split into a + going and - going sawtooth.

One side of the screen is mostly driven by the + part of the amplifier, the other by the - part.? If one side of the amplifier chain is non-linear, the the waveform on that side will be non-linear.

One good reason for an amplifier to be non-linear is that it is being run into saturation.? Since the amount of non-linearity is small, it'll be subtle.

If you can turn off the sweep, turn down the intensity, then go through and see if the collector voltages on the + part and the - part of the amplifier match.? The horizontal positioning should be about centered.? If one stage or side is off then the trace will not center at the control's mid position.

You could put timing pulses in on a scope, then use the sweep waveform from each collector to drive the H input of another scope, and perhaps see where the nonlinearity begins.

there are typically centering and gain adjustments which ought to be at reasonable settings.

Harvey

Harvey,

Okay, I see the (I think) differential amplifier that drives the horizontal deflection plates (starting with Q4025/Q4038 and ending with Q1024/Q1031/Q1043/Q1049), but it sounds like the fault is already present at TP1061, which is the output of the op amp U6061, and is buffered from the horizontal deflection amplifier by two other op amps (U7073 and U2060), so I doubt that a fault in the deflection amplifier board could be causing the low voltage at TP1061 that Michael is seeing.

Maybe R7051 and R7081, if they were both below spec, could be dragging down the output of U6061, but U7073 would have to be in really bad shape to be letting anything further down the line have any effect at TP1061.

Still, I'm fully prepared to be wrong. I'm out of my depth, and I'm not entirely sure I understand what Michael has measured in the circuit.

Also, I don't see much trimming in the deflection amplifiers. There is a gain trim pot across U2060 (R2069), and two trim caps on the inputs to the differential amps driving the deflection plates (C4057 and C5021), but I don't really understand what the trim caps are doing.

-- Jeff Dutky

Interleaved:

On 11/28/2021 5:59 AM, Jeff Dutky wrote:

Harvey,

Okay, I see the (I think) differential amplifier that drives the horizontal deflection plates (starting with Q4025/Q4038 and ending with Q1024/Q1031/Q1043/Q1049), but it sounds like the fault is already present at TP1061, which is the output of the op amp U6061, and is buffered from the horizontal deflection amplifier by two other op amps (U7073 and U2060), so I doubt that a fault in the deflection amplifier board could be causing the low voltage at TP1061 that Michael is seeing.

Using another scope, if you grab the sweep at the U6061 output, run it to the horizontal input, and put timing pulses on the vertical inputs, that ought to tell you about the linearity.

Maybe R7051 and R7081, if they were both below spec, could be dragging down the output of U6061, but U7073 would have to be in really bad shape to be letting anything further down the line have any effect at TP1061.

The voltages at U6061 are OK?? supply wise?

If U6061 is easy to replace, I'd try that and take a look at the parts around it, if it's producing the sweep.? It's possible that bad supply voltages could cause amp nonlinearity in the circuit.

Still, I'm fully prepared to be wrong. I'm out of my depth, and I'm not entirely sure I understand what Michael has measured in the circuit.

Also, I don't see much trimming in the deflection amplifiers. There is a gain trim pot across U2060 (R2069), and two trim caps on the inputs to the differential amps driving the deflection plates (C4057 and C5021), but I don't really understand what the trim caps are doing.

Likely peaking the frequency response of the amplifier.? Checking the linearity per stage could be useful, or maybe just seeing what the manual says about the appropriate adjustments.

Harvey

-- Jeff Dutky

I don't think you can say that the fault is present at TP4061. -5V as measured here should correspond to the left graticule.
So far it seems that it's OK up to the output of U2060 (when I measure the voltage at pin 6 with the trace at the right graticule it is close to the inverse of the voltage at the left most graticule).
This is the older version of the board which have a couple of matched pair transistors.

I should have realized the implication of ... "when I measure the voltage at pin 6 with the trace at the right graticule it is close to the inverse of the voltage at the left most graticule".
This means that the push pull HV amplifier is working correctly and that the problem is either U2060 or the "shaper network". Presumably this 'shaper' is a way to introduce a nonlinearity to compensate for the tube but it is symmetrical and so should be mirrored around 0V.
I'll run a Spice simulation to see if I can better understand how it is supposed to work.

Yes, that was it. A bad diode (CR4051) in the 'shaper' network.

On Sun, Nov 28, 2021 at 11:16 PM, vk2bea wrote:

Yes, that was it. A bad diode (CR4051) in the 'shaper' network.

HUZZAH!