Tektronix 2465B oscilloscope teardown

[Bryan]

The multimeter I have is a UTI-61E, below is a pic of the AC specs. I thought crest factor also had to be taken into consideration. Either way you are right, splitting hairs. Will compare the output of the HP to the  multimeter and on my Rigol1052e. The Tek is only rated for 2% vertical accuracy so I am probably well in the ballpark. Although it would be nice to have a calibrated 34401a.

[David Hess]

I thought crest factor also had to be taken into consideration.

The crest factor for a square wave is 1 so it is not a consideration.

[calin]

I also need to do a vertical cal on my 2465 and of course no PG506 .. I do have a nice working FG 504 very stable and nice, a 51/2 digit Fluke  multimeter and some scopes to compare ... but I am scratching a bit my head at how the vertical calibration signal should look like.


So it is clear, the calibration signal should be a 1kHz  square with 50% duty with the following values, 0.5V, 0.2V, 0.1V, 50mV, 20mV, 1V and 10V. Now here is what is not clear:


1) Are these voltages in 50 ohm or in high impedance ? I think is HiZ.
2) Yes voltages are peak-to-peak, are they 0 to 0.5V for or AC  +/- 2.5V ? I think they are 0 to +0.5 but better check first before I do something stoopid

[BravoV]

And since it was open (you need to calibrate the markers with pots), i also took a picture of my wannabee U800 cooler:

Guido, that heatsink looks like a stacked coins to me. 

[BravoV]

So it is clear, the calibration signal should be a 1kHz  square with 50% duty with the following values, 0.5V, 0.2V, 0.1V, 50mV, 20mV, 1V and 10V.

I wish if there is an easy and relatively cheap to build such circuit, any experienced or experts want to contribute on the circuit schematic ?

Don't need to be high precision right ? <5% accuracy is enough maybe ? 

[Bryan]

1) Are these voltages in 50 ohm or in high impedance ? I think is HiZ.
2) Yes voltages are peak-to-peak, are they 0 to 0.5V for or AC  +/- 2.5V ? I think they are 0 to +0.5 but better check first before I do something stoopid

Good question. I assumed they are 50 ohm impedance. Thea manual states the testing procedure switches to 50 ohm impedance in Cal 01, I just assume this carried over to Cal 02 (Vertical).

^{Upon entering CAL 01, the Input Coupling is
automatically set to 50 Q DC and the 50 Q OVERLOAD
protection is disabled. Before starting the procedure,
make sure any 50 Q OVERLOAD condition
has been cleared.}

[David Hess]

I also need to do a vertical cal on my 2465 and of course no PG506 .. I do have a nice working FG 504 very stable and nice, a 51/2 digit Fluke  multimeter and some scopes to compare ... but I am scratching a bit my head at how the vertical calibration signal should look like.

...

1) Are these voltages in 50 ohm or in high impedance ? I think is HiZ.

What matters is the peak to peak voltage at the oscilloscope's vertical input with or without termination.

2) Yes voltages are peak-to-peak, are they 0 to 0.5V for or AC  +/- 2.5V ? I think they are 0 to +0.5 but better check first before I do something stoopid

Only the peak to peak value matters as long as the waveform can be positioned on the CRT.  The PG506 for instance generates square waves from 0 to +x volts in calibrated amplitude mode.

Calibration of the 2465 cursors and any automatic measurements might require 0 to +x versus AC though.

[David Hess]

Good question. I assumed they are 50 ohm impedance. Thea manual states the testing procedure switches to 50 ohm impedance in Cal 01, I just assume this carried over to Cal 02 (Vertical).

It cannot always be 50 ohms because a PG506 will not generate 10 volts into a 50 ohm load.

Another reason it cannot be 50 ohms is that then the calibration amplitude would depend on the accuracy of the 50 ohm termination value.

[mwilson]

Good question. I assumed they are 50 ohm impedance. Thea manual states the testing procedure switches to 50 ohm impedance in Cal 01, I just assume this carried over to Cal 02 (Vertical).

It cannot always be 50 ohms because a PG506 will not generate 10 volts into a 50 ohm load.

Another reason it cannot be 50 ohms is that then the calibration amplitude would depend on the accuracy of the 50 ohm termination value.

Yep... also, to remove any remaining doubt, the PG 506 calibration procedure calls for calibrating its standard amplitude DC voltage ranges with an exact 1 megaohm load (within 0.1%).

[calin]

Perfect thanks guys. Looks like I am not a total noob and I've been reading right .  1kHz square, 50% duty, 0 to +X DC voltages into HiZ it is.  I will use the FG 504 and my trusty Fluke meter to measure RMS and try a vertical cal on my scope.

[guido]

Guido, that heatsink looks like a stacked coins to me. 

It is out of a 7A13 actually.

So it is clear, the calibration signal should be a 1kHz  square with 50% duty with the following values, 0.5V, 0.2V, 0.1V, 50mV, 20mV, 1V and 10V.

I wish if there is an easy and relatively cheap to build such circuit, any experienced or experts want to contribute on the circuit schematic ?

Don't need to be high precision right ? <5% accuracy is enough maybe ? 

Take a 10V reference chip, make the voltages with some precision dividers and use such a cmos analog switch to chop it up. Driven by a 555 or function generator. Maybe use a low offset opamp after the dividers or switch as buffer. I used a 6.8nF cap on the output to stop overshoot. You might check this with a good scope for a good value for your switch/coax cable.

Check the PG506 manual for the precision requirements...

I wonder if a normal function generator will work at the low voltages. What i looked at, some had offset figure specifcations of +/- 10mV... You might need to use a high voltage square as output and divide using precision resistors. See if you can get the 20mV nicely on the screen of a good scope.

[Bryan]

Good question. I assumed they are 50 ohm impedance. Thea manual states the testing procedure switches to 50 ohm impedance in Cal 01, I just assume this carried over to Cal 02 (Vertical).

It cannot always be 50 ohms because a PG506 will not generate 10 volts into a 50 ohm load.

Another reason it cannot be 50 ohms is that then the calibration amplitude would depend on the accuracy of the 50 ohm termination value.

What would my best option be for attempting to calibrate with my HP8116a. It as far as I know can only output the stated output voltages at 50 Ohm.

[calin]

The HP can output into HiZ ... just the voltage seen by the scope will be double. But then if you measure RMS with a mutimeter that is the same as seen by the scope . Of course if you have a square wave with 50% RMS value for 5V pk to pk is 2.5 .  Same is my FG504 .. 50 ohm.  At 1Khz reflections on the line are not something to worry too much about i think. Yeah many FG-s cannot go that low .. you don't need precision resistors .. just precision dividing any resistor of the right value and a multiturn pot/trimmer will do as ong as you have a good rms meter and a comparision scope you are set.  We are not looking for ultra high precision and low tempco here .. if is stable for 5 mins until the cal step is done is good enough i guess

[Bryan]

The HP can output into HiZ ... just the voltage seen by the scope will be double. But then if you measure RMS with a mutimeter that is the same as seen by the scope . Of course if you have a square wave with 50% RMS value for 5V pk to pk is 2.5 .  Same is my FG504 .. 50 ohm.  At 1Khz reflections on the line are not something to worry too much about i think. Yeah many FG-s cannot go that low .. you don't need precision resistors .. just precision dividing any resistor of the right value and a multiturn pot/trimmer will do as ong as you have a good rms meter and a comparision scope you are set.  We are not looking for ultra high precision and low tempco here .. if is stable for 5 mins until the cal step is done is good enough i guess

Right, double checked the HPmanual and that is what it suggests, double the output. I just wish I had a calibrated DMM to double check the output from the HP. 2.5v output reads 4.99v or 5.12 depending on the number of divisions on the scope. Odd in itself and may be a problem with my Rigol.

If I did not have the HP I would look at some precision reference voltage chips and use a couple LTC 1043 chips, supposedly very good , very accurate voltage divider, multipliers.  Page 8 of the datasheet

http://www.linear.com/product/LTC1043

One or two reference voltages and a couple LTC1043's and you could come up with all the required voltages needed Just wondering what circuit one could use to chop it into 1khz without any appreciable noise or loss.



 

[David Hess]

It cannot always be 50 ohms because a PG506 will not generate 10 volts into a 50 ohm load.

Another reason it cannot be 50 ohms is that then the calibration amplitude would depend on the accuracy of the 50 ohm termination value.

What would my best option be for attempting to calibrate with my HP8116a. It as far as I know can only output the stated output voltages at 50 Ohm.

The way I did it with a function generator at low voltages was to use a stepped attenuator and 50 ohm termination while measuring the peak to peak voltage at the oscilloscope vertical input.  At high voltages the stepped attenuator and termination are not needed.

When dealing with voltages lower than can be accurately measured with an AC voltmeter, the stepped attenuator may be calibrated at a higher level and then the voltage can be measured at the input to the attenuator.

[Bryan]

Hmm, just got me thinking, could I not before the calibration procedure compare a precision DC voltage reference and the HP outputs on Channel 1 and 2 of the scope. Note the difference if any between the two and then subsequently adjust the HP outputs during the calibration standards. For example if the difference between the precision 1v dc and the HP 1v 1Khz AC is +10mv on the scope then when actually calibrating set the HP output for 90mv.

Or just don't worry about it and rely on the accuracy of my Uni61E to test the HP output. Thinking the scope comparison method may be more accurate in determining the output accuracy of the HP.

[David Hess]

Using two channels of the oscilloscope to make the comparison will not work because the offset and gain calibration of the two channels is independent.  One of the reasons to do the calibration is to match them as closely as possible.

There is an old technique which does exactly what you are proposing using one oscilloscope channel and it could be used to calibrate a signal generator.  It involves using a "differential comparator" which adds a precision offset voltage to the signal to be measured.  The offset voltage is adjusted to bring two different vertical levels into alignment and the difference represents the difference in voltage between those two signal levels.

The old Tektronix 7A13 does exactly this to 4 significant digits and with a resolution down to 1 millivolt although better resolution is available if an external voltmeter is used to measure the comparison voltage.

This is also known as a slide-back measurement.  I would have gone this route if I had lacked a modern AC voltmeter which could accurately measure low frequency square waves.

[Bryan]

Thanks David;

I did a Google search on Slide back measurements did not find much info. Any additional details you could share would be appreciated.

[David Hess]

There is not much online about it.

Tektronix used and described the procedure in a couple of their service manuals where it was used for instrument calibration.

I suspect there is a detailed article about it in one of the TekScope publications but I have never seen an index of them to know where to look.  The 7A13 was first produced in 1969 so that would be the time period to search. 

There were earlier differential comparators as well and the operating manuals for them may describe how they can be used to make slide-back measurements.

The term also shows up in some old NIST publications.

Update: I did a search and found articles in November 1972 TekScope and January/Feburary 1973 TekScope but they are part 2 and 3 so there is an earlier one which I do not have.  They do not discuss the 7A13 differential comparator specifically but part 2 mentions slide-back measurement:

Let's see how we go about making a differential comparator measurement.  First, we establish a reference position on our display by grounding both inputs.  Then selecting the appropriate input (positive voltage source to the + INPUT, negative voltage source to the - INPUT), we switch the other input to the comparison voltage (Vc).  Next, the comparison voltage is adjusted until the trace "slides back" to the reference position.  What have we accomplished?  Using the "difference" principle we have introduced a "common-mode" condition in the form of the comparison voltage; that is, the comparison input voltage now equals the signal input.  We see that we now have the ability to measure any potential whether it be DC, complex in nature, or a combination of both - such as a complex wave superimposed on a DC potential.  Thus, we have an extremely versatile measuring tool.

The reason you would do this instead of reading the voltage directly off of the oscilloscope is increased precision and accuracy.

[Bryan]

Let's see how we go about making a differential comparator measurement.  First, we establish a reference position on our display by grounding both inputs.  Then selecting the appropriate input (positive voltage source to the + INPUT, negative voltage source to the - INPUT), we switch the other input to the comparison voltage (Vc).  Next, the comparison voltage is adjusted until the trace "slides back" to the reference position.  What have we accomplished?  Using the "difference" principle we have introduced a "common-mode" condition in the form of the comparison voltage; that is, the comparison input voltage now equals the signal input.  We see that we now have the ability to measure any potential whether it be DC, complex in nature, or a combination of both - such as a complex wave superimposed on a DC potential.  Thus, we have an extremely versatile measuring tool.

Hi David:

Looks like to perform the differential comparator measurement as above  a differential probe is needed?.

[David Hess]

Looks like to perform the differential comparator measurement as above  a differential probe is needed?.

A differential comparator is a type of differential amplifier and may be used in place of a differential probe.

[Bryan]

Well looked up on differential amplifiers and I am sure there is a way but way beyond me I am afraid, especially the complexity of one of the signals being AC. Ran my Rigol 1052E and a Tek 2465A side by side and compared the outputs of the HP, for the most part the Tek and the Rigol were very close in all the measurements. Under 1v outputs only a few mv difference between the two and 10's of mv above 2v. Where there was a large discrepancy was at the 10v range where the Rigol was almost .2v out compared to the tek. I think there may be a issue with the Rigol. The Rigol varies a fair bit in measurement depending on the vertical scale selected.

So in short I think the output accuracy of the HP is ok, at least for my hobbyist purposes to calibrate my 2465B.   

[David Hess]

Well looked up on differential amplifiers and I am sure there is a way but way beyond me I am afraid, especially the complexity of one of the signals being AC.

Differential comparator slide-back measurements are one of those techniques which have been lost in time.  They are especially useful for measuring complex AC signals but as you observe, it would help to see how it is done in action and I know of no good examples showing this.

When first did oscilloscope calibration, I used an AC voltmeter to measure the peak to peak signal values instead of making a slide-back measurement which would have taken longer.

[Phaedrus]

I finally found the time to open up my 2465A I bought a couple months back. I was actually quite surprised; the date codes run up to middle of 1986. According to Wikipedia this model wasn't offered until 1987. So I think I might have gotten one of the first production batches. 

All the electrolytics look good visually, but there's probably been some degradation anyway. I'm going to look into replacing them. Also probably stick a San Ace fan in there. How would you test to see if the U800 is having any issues?

[David Hess]

How would you test to see if the U800 is having any issues?

If U800 is having problems then it should be apparent on the display with horizontal drift or gain changes.