Tektronix 2445B - Noise on calibrator squarewave, is this normal?

[CZ101]

Hi all, first post here.

I recently scored a nice looking 2445B w/ 4 probes for a decent price. I'm a Tektronix newbie btw, most of my oscilloscope experience has been with a 50 Mhz Hitachi V-550B.

So anyway, I just wanted to ask:

Is it normal for the calibrator square wave to be so noisy without the 20 Mhz bandwidth limit enabled? See pics for what I'm talking about.

The probe I'm using here is a Tek P6108A, and I'm getting basically the same results with a Tek P6133 probe as well.

One curious thing going on here is the noise is visibly reduced as the probe lead is brought near the rear of the oscilloscope, which I assume is the location of the power supply?

Any help would be greatly appreciated!

[KJDS]

Looks like the intensity is set a bit high and the probes need adjusting.

[miguelvp]

Probe needs to be adjusted at the base of the probe with a plastic flat head until the signal is square.

Other than that it seems you have the intensity set too high.

LOL KJDS you beat me to it

[eurofox]

Looks like the intensity is set a bit high and the probes need adjusting.

I was writing almost the same at the same time but add focus and astig (astigmatism) and of course intensity. 

[T3sl4co1l]

If it's periodic, you can probe the ground pin and move the cable around, see if there's anything to trigger on.

Most scopes I've seen have an extra maybe 0.1-0.2 div noise on the highest gain setting at high bandwidth.

Tim

[MarkL]

The screen is showing 10mV/div which makes the calibrator output 40mV.  It should be 400mV.  Perhaps something is wrong with the calibrator output.

But another possibility is there is something wrong with your probe ground, and maybe the BNC ground since you tried a different probe.  This could explain the increased noise when you move the probe around, and also that the x10 ID ring for the probe is not working.  (When you plug a x10 probe in, it should automatically adjust the V/div.)

[CZ101]

Thanks for the replies. The intensity I think is being exaggerated in the photos.

Regarding the compensation adjustment, yes the probe is clearly a little "under" compensated in the first post's pics.

But the noise at full bandwidth seems excessive, no?

I've attached pics this time with a Pomona 4550B 100Mhz probe - same results.

The screen is showing 10mV/div which makes the calibrator output 40mV.  It should be 400mV.  Perhaps something is wrong with the calibrator output.

But another possibility is there is something wrong with your probe ground, and maybe the BNC ground since you tried a different probe.  This could explain the increased noise when you move the probe around, and also that the x10 ID ring for the probe is not working.  (When you plug a x10 probe in, it should automatically adjust the V/div.)

The Tek 6108A probes don't have the little retractable pin that contacts the x10 ID ring. The Tek P1633 probes however do have the spring loaded pin and I've confirmed the x10 ID feature to be working with them (automatically switching to 100mV scale) so in this sense the scope appears to be working.

If it's periodic, you can probe the ground pin and move the cable around, see if there's anything to trigger on.

Most scopes I've seen have an extra maybe 0.1-0.2 div noise on the highest gain setting at high bandwidth.

Tim

The last pic attached shows ground probed at 100ns per division. Clearly something is wrong, yes?

[David Hess]

Thanks for the replies. The intensity I think is being exaggerated in the photos.

Regarding the compensation adjustment, yes the probe is clearly a little "under" compensated in the first post's pics.

But the noise at full bandwidth seems excessive, no?

I've attached pics this time with a Pomona 4550B 100Mhz probe - same results.

High bandwidth oscilloscopes can be tricky as far as noise.  What you are seeing could very well be a local interference source or even a radio transmitter.  To test for this see if the noise level changes when you ground the probe, move its cable, or move your hand along it.

This is another reason coaxial probe connections are so important with high bandwidth oscilloscopes.  I can clip the ground lead to the probe tip on mine and use it as an RF sniffer.  Compact fluorescent lamps and LED lamps are particularly bad about spewing RF interference everywhere.

Some of the older Tektronix oscilloscopes provide the calibrator output through a BNC so it is possible to use a BNC to coaxial probe tip adapter and get a very clean calibration signal.

The Tek 6108A probes don't have the little retractable pin that contacts the x10 ID ring. The Tek P1633 probes however do have the spring loaded pin and I've confirmed the x10 ID feature to be working with them (automatically switching to 100mV scale) so in this sense the scope appears to be working.

So the probe and calibrator are working correctly.  The displayed 10 millivolts per division is actually 100 millivolts per division and the calibration signal is 0.4 volts peak to peak.

If it's periodic, you can probe the ground pin and move the cable around, see if there's anything to trigger on.

Most scopes I've seen have an extra maybe 0.1-0.2 div noise on the highest gain setting at high bandwidth.

This is my experience as well but I do not think that is what is going on here.  10 millivolts per division should be pretty clean.

The last pic attached shows ground probed at 100ns per division. Clearly something is wrong, yes?

It shows a frequency comfortably higher than the switching power supply operates at so I think you are seeing something external.  If you want to double check, switch the channel coupling to ground or use a coaxial short on the end of the probe.

[T3sl4co1l]

You don't happen to have an FM radio station nearby do you.....

That looks comfortably like 100MHz, though if it has any shape is dubious (how would you know in a 200MHz BW?).

Tim

[CZ101]

High bandwidth oscilloscopes can be tricky as far as noise.  What you are seeing could very well be a local interference source or even a radio transmitter.  To test for this see if the noise level changes when you ground the probe, move its cable, or move your hand along it.

Yes, the noise signal amplitude is sensitive to the positioning of the probe cable and/or my physical proximity. Touching the scope on its sides (on the plastic stand end caps) has quite a dramatic effect and sends the triggering haywire. When the oscilloscope inputs are set to DC it is particularly sensitive.

This is another reason coaxial probe connections are so important with high bandwidth oscilloscopes.  I can clip the ground lead to the probe tip on mine and use it as an RF sniffer.  Compact fluorescent lamps and LED lamps are particularly bad about spewing RF interference everywhere.

This has been my experience as well. Just for kicks I tried to see if I could isolate the source - unplugging fluorescent lights and other items with switching power supplies, but to no avail. I also tried the scope on a power conditioner that I've observed to reduce EMI on connected DC power supplies, but with no effect in this case.

It shows a frequency comfortably higher than the switching power supply operates at so I think you are seeing something external.  If you want to double check, switch the channel coupling to ground or use a coaxial short on the end of the probe.

Channel coupling to ground almost completely squashes the interference.

Attached is a pic showing Channel 1 and 2 in "differential mode" (channel two inverted and added to channel 1).

Seeing as though the two almost cancel (with some finagling, albeit), would this suggest the signal is common mode and therefore, external, being picked up by the probes w/ their leads?

You don't happen to have an FM radio station nearby do you.....

That looks comfortably like 100MHz, though if it has any shape is dubious (how would you know in a 200MHz BW?).

Tim

There are radio stations within a 5 mile radius towering over me (I live in Cincinnati, OH if you are familiar). First pic attached shows frequency measurement.

[T3sl4co1l]

Channel coupling to ground almost completely squashes the interference.

Attached is a pic showing Channel 1 and 2 in "differential mode" (channel two inverted and added to channel 1).

Seeing as though the two almost cancel (with some finagling, albeit), would this suggest the signal is common mode and therefore, external, being picked up by the probes w/ their leads?

If the probe leads are in the same general location (ideally, twisted or tied together), they'll be picking up the same signal (phase and amplitude).

In essence, what you've got is an inductive loop antenna; 3-6 foot probe leads are probably a good rough match to the FM band.

There are radio stations within a 5 mile radius towering over me (I live in Cincinnati, OH if you are familiar). First pic attached shows frequency measurement.

Probably about right.  You'd have to check which ones are closest and strongest.  I don't notice much here, but the listing for my area shows most are 10 miles away.  If you have a good strong 50kW station blaring within a few miles (or a smaller one even closer, e.g. a college radio station measured on or near campus -- a common gotcha for newbies!), that would be no problem.

Might also help to move your equipment about three feet in some lucky direction.  Use multipath to your advantage for once!   The best solution of course would be a shield in the direction of the antenna (or better still, a full Faraday cage).

A stack of ferrite beads / rings on the probes helps, but signals still get in (consider the capacitance of the probe cable and handle as the antenna).  Tends to be a matter of dB rather than tens of dB.

Tim

[CZ101]

Might also help to move your equipment about three feet in some lucky direction.  Use multipath to your advantage for once!   The best solution of course would be a shield in the direction of the antenna (or better still, a full Faraday cage).

A stack of ferrite beads / rings on the probes helps, but signals still get in (consider the capacitance of the probe cable and handle as the antenna).  Tends to be a matter of dB rather than tens of dB.

Tim

I took the scope down to the basement and low and behold there was a substantial reduction in EMI pickup, thanks for the advice.

By all appearances, the Tek 2445B is still noisier than my Hitachi 550B, and I am not sure if this is simply due to the extra bandwidth; something tells me it is not (but I may be wrong).

I've attached pics of some tests I did with my signal generator (BK 3017A). I used a straight 50 ohm coax connection (terminated) and compared this with a Tek P6108A probe attached to the signal gen's output with a BNC-probe tip adapter. The difference in rise time is noticeable.

One thing that strikes me in comparing these two scopes is how much sharper the trace is on the Hitachi. Even in 20MHz bw limit mode the Tek's trace is far less well defined.

I've read various posts recently about the role of the Tek's switching power supply in all of this and how recapping could be of some benefit.

Any thoughts?

[T3sl4co1l]

Just looks to me like the trace is thicker.  I would guess the Tek has one of those special high-falootin' CRTs, which the Hitachi doesn't?  By which I mean, the tube in the 475 at least has a mesh screen following the deflection plates, which is used to improve intensity * deflection sensitivity (the beam deflection is greater before reaching the EHV section), but necessarily reduces beam current and focus in the process (cuz, well, it's a friggin screen!).  So... a Tek with a slightly thicker beam than most probably isn't unusual.

Or maybe your intensity is just a little high (a dimmer beam is thinner.. space charge effect).

If it's periodic noise, you can adjust the signal frequency (and trigger holdoff to some extent) veeeery slowly and see if anything zooms past (i.e., the signal gets some harmonic relation to the interference within a human-readable time scale, maybe +/-20Hz to be able to see it line up).  Example: before I recapped my 475, I would notice the trace had an intensity modulation (dips and peaks every 120Hz).  It will be harder to spot if it is switching interference.

Tim

[David Hess]

The difference in bandwidth is enough to explain why the interference shows up so much better on the 150 MHz 2445B than the 50 MHz 550B.  I hardly notice anything on my 100 MHz oscilloscopes but on my faster ones local RF interference is significant unless I use coaxial probing techniques.  Short ground leads are usually not enough.

The extra bandwidth will also make the trace of the 2445B fuzzier but there could be other reasons for this.  The power supply is a possibility and worth checking but also the CRT designs are significantly different.  To support enough deflection at a high bandwidth, the Tektronix CRT has a mesh expansion dome which as a side effect increases the trace size.  Some of their older oscilloscopes are preferred because they lacked that.  My 50 MHz 547 from the tube era has an especially crisp CRT.

The 2445B also has some CRT calibrations which affect the display quality.

[leppie]

Do you have a flourescent light with an electronic ballast near it?

That's what causes issues for me if the probe is too close to it.

[MarkL]

...
One thing that strikes me in comparing these two scopes is how much sharper the trace is on the Hitachi. Even in 20MHz bw limit mode the Tek's trace is far less well defined.
...

If you think it could be sharper, there are several internal adjustments that affect the focus.  They are small pots reachable through the HV power supply shield.  It might be worth trying if adjusting the astig or focus gives you sharper traces and sharper dots in one axis or the other.  Take a look at the service manual for details.

If you decide to move any adjustments, make sure to mark their positions first in case there's no improvement.  Some of them will slightly affect the screen calibration, and you may have to tweak the vert or horiz centering and gain pots.

But honestly speaking, your screen already looks pretty good to me.

[kevinpt]

+1 on the internal adjustments. I like to set my 2465 up with the crispest trace possible with the readouts a little blurred to help combat burn in.

[David Hess]

I would not worry about burn in on a normal CRT unless you are abusing it.  The only ones where I have seen this be a problem are storage, MCP, specific non-P31 phosphor, and non-aluminized screen CRTs.

[CZ101]

Wow, a lot of good info here. I've done some soul searching and I'm leaning in the direction of exchanging the 2445B and getting something else like the Hitachi V-1565.

In taking measurements on the 2445B I find it annoying that it always defaults to a preset timebase and attenuation rather than remain as set beforehand. Also, having a more realtime frequency counter would be useful.

Thanks everyone for the insights, I learned a lot!

[David Hess]

In taking measurements on the 2445B I find it annoying that it always defaults to a preset timebase and attenuation rather than remain as set beforehand. Also, having a more realtime frequency counter would be useful.

What do you mean by "more realtime" frequency counter?  The frequency counter in the Tektronix 2247A is pretty good being a reciprocal frequency counter and I believe the one in the 2445B is similar.

The Tektronix oscilloscopes which are contemporaries to the 2445B including the 2232, 2247A series, and 2440 series remember their settings when power is removed so I would be surprised if the 2445B did not unless it was specifically configure not to like the 2440 can be.

[CZ101]

In taking measurements on the 2445B I find it annoying that it always defaults to a preset timebase and attenuation rather than remain as set beforehand. Also, having a more realtime frequency counter would be useful.

What do you mean by "more realtime" frequency counter?  The frequency counter in the Tektronix 2247A is pretty good being a reciprocal frequency counter and I believe the one in the 2445B is similar.

The Tektronix oscilloscopes which are contemporaries to the 2445B including the 2232, 2247A series, and 2440 series remember their settings when power is removed so I would be surprised if the 2445B did not unless it was specifically configure not to like the 2440 can be.

By "more realtime" I mean free running and updating with changes in frequency. On the 2445B you have to enter frequency measurement mode and then let it calculate - "measurement in progress, please wait". When the calculation is finished (after a couple seconds), the 2445B does not return to its previous horizontal or vertical settings, but defaults to the "auto setup" configuration. This is the case for all the functions in the measurements menu (rise time, voltage, etc).

The 2445B does remember its settings after being powered off, just not after being put into measurement mode.

[David Hess]

In taking measurements on the 2445B I find it annoying that it always defaults to a preset timebase and attenuation rather than remain as set beforehand. Also, having a more realtime frequency counter would be useful.

What do you mean by "more realtime" frequency counter?  The frequency counter in the Tektronix 2247A is pretty good being a reciprocal frequency counter and I believe the one in the 2445B is similar.

The Tektronix oscilloscopes which are contemporaries to the 2445B including the 2232, 2247A series, and 2440 series remember their settings when power is removed so I would be surprised if the 2445B did not unless it was specifically configure not to like the 2440 can be.

By "more realtime" I mean free running and updating with changes in frequency. On the 2445B you have to enter frequency measurement mode and then let it calculate - "measurement in progress, please wait". When the calculation is finished (after a couple seconds), the 2445B does not return to its previous horizontal or vertical settings, but defaults to the "auto setup" configuration. This is the case for all the functions in the measurements menu (rise time, voltage, etc).

The 2445B does remember its settings after being powered off, just not after being put into measurement mode.

I see.  The frequency counters in the 2247A and 2236 oscilloscopes just free run with the readout updating continuously.  Of course they also support gated measurements as well.  In a lot of respects they duplicate the functions of the 7D15 frequency counter plug-in for the 7000 series mainframes but operate as a reciprocal counters so resolution depends only on gate time and not on input frequency.