Re: SUCCESS! The "sick" 475A is now the "fixed" 475A!


Harvey White
 

responses interleaved and I'll add an anecdotal tale (but true).

On 12/8/2020 6:34 PM, Dave Peterson via groups.io wrote:
I'd say this deserves a new topic under calibration, but I wouldn't want to loose the connection to all this good info in this one.

I spent some time working through the calibration procedure from the 465 manual last night. I was in the "Vertical System Calibration" section of the "Short-Form Calibration". Uhg! The depth of my ignorance is further revealed!

I have to mention that in performing these sorts of procedures I find that I suffer from the same ailment I had in the Army. It's not good enough for me to follow the procedure. I have a personal flaw in which I need to understand what I'm doing via the procedure. What am I doing, why, what effect previous steps have on the current adjustments, and what effects the current adjustments have on the next steps. I have to "grok" it. So I'm slow.
Everybody learns differently, some is kinesthetic, (you have to do it with your hands to understand it), some is auditory (hear it and it's gold), some visual (doesn't make sense until you see the picture).  Slow is not a factor.

I now understand better some of the issues Raymond raised with the idea of a cheap function generator. I do find it useful, and it's working well for what it is. But it does have it's limitations regarding verifying the specs of the 465. I'll have to keep my eye on the recommendations here, like the fast edge pulse generator. As Raymond pointed out, sure enough, the procedure for vertical calibration calls for a 3.5ns edge in the High-Frequency Compensation, step 23. And the bandwidth of the vertical input section requires a 100MHz capable signal source - who's roll off is above that. The cheap FG starts rolling off at about 10MHz.
The problem on some function generators is not the frequency.  Many rise time measurements can be done with a 1 Khz wave, with the frequency only determining the overall brightness of the trace.  The real problem is the rise time.  To measure the rise time of a scope, you need a waveform that is significantly better than the rise time to be measured, perhaps at least by a factor of 10 so that the contributions of the generator are minimized.  Most function generators don't do that.  In fact, the PG506 has two specific outputs, one with a + risetime that's specified, one with a - risetime.  All the other waveforms are specified to amplitude (IIRC) and *not* risetime.  The risetime waveforms *only* are specified for risetime, you don't care about the amplitude.

I have to say, one of my motivations behind all of this is/was my interest in building some audio amps. My daughter is learning electric guitar and I thought it'd be fun to build some for both of our entertainment. She'd learn what's going on with amps, and I'd get to practice and learn the issues and effects impacting audio amp design and build. So the two-channel sweep FG I think will work well in an audio application, and helps get me started on the scope tuning front. And a 465 is a good scope to use for it as well. Why I finally pulled the trigger on it. Oops.
As has been mentioned, the essential difference in amplifiers is how they sound, and that's subjective.  One measure of hearing is distortion.  If an amplifier sharply clips, then it introduces a lot of odd order harmonics, and the sound is perceived as "harsh".  If it softly clips (a characteristic of tube amplifiers driving transformers), then the energy in those harmonics is reduced.  It's still clipping, but it's a "softer" clipping. Different sound.  Not so much subjective (since it can be seen and measured) but it is interpreted differently.


I suspect I'll have more questions the further I go through the calibration procedure. Something that I encountered last night was mentioned by Raymond in reply to Tom: a 20pf normalizer. What is a normalizer? Why are people using a 10x probe for it? The 465 manual leaves a lot to be desired in explaining some of these things. Like attenuators as needed? Needed for what precisely? Waveform characteristics? Noise?
This one is easy.  An oscilloscope input looks like a resistor in parallel with a capacitor, both connected to ground.  At DC, all you see is the resistor.  As the frequency goes up, the capacitive reactance (looks a bit like a resistor) goes down with frequency, so there's a frequency when you see 1 meg of resistance in parallel with 1 meg of reactance, and your signal sees 500K.  Your standard oscilloscope probe is a resistor in parallel with a capacitor.  Those form a voltage divider. At DC, you see a 9 meg resistor in series with a 1 meg resistor (DC = no capacitors), and get a 10:1 division ratio.  As the frequency goes up, the loading (due to the capacitors) goes up (lower effective reactance), and you need to adjust the capacitor in the probe for the appropriate 10:1 ratio, since the capacitance predominates.

It's not as simple as that, since you have the inductance of the cable, the capacitance of the cable, and a few other things to consider.

What the normalizer does is to be a 2:1 divider with the capacitor adjusted precisely to feed a network of exactly 1 meg paralleled with a 20 pf capacitor.  In other words, exactly what the scope input is specified to be.   Put the normalizer on, adjust the scope for proper response, and the scope input is 1 meg and 20 pf.  Since that's the standard, when you adjust one probe to make it exact (for 1 meg and 20 pf), it ought to be exact for any *other* 1 meg and 20 pf input.  Just for fun, the higher frequency scopes need less capacitance, HP has a different standard capacitance, and there are scopes with 50 ohm inputs where the capacitance plays a lot less of a role in the input impedance.

Hope this helps.



I have a long wish list of test equipment queued up on Amazon. Like 50ohm cables, BNC connectors of many sorts - like T's, and 50ohm BNC terminators. Those were called for often in test setups in the Army. Do I need other values? Do cap "normalizers" come as BNC connections? How are they connected? In-line? Shunt?
Normalizers are a small box with either BNC (for newer scopes) M and F on either side, series connection between the generator and the scope, or UHF (SO-239) connectors for older equipment.

Now for the anecdote.  In my early days, I had Heathkit equipment.  A sine/square wave generator and an oscilloscope (OM-3) that likely barely hit 500 Kc (as it was then).  I put the resistors in the generator by bending the leads, and using all the leads, letting the resistors have an inch or more lead.  No problem, the generator output looked fine.

Then I got a Tektronix scope (a 513D), which had a much higher bandwidth.  The excess lead length of the resistors caused the circuits to ring, lots.  What I thought was a square wave certainly wasn't a good one.

SO what you see is completely dependent on your measuring equipment, yet you do have, somewhere in there, the concept of "what's good enough".

so with a square wave generator that's got too slow a rise time, you just simply can't adjust a fast risetime circuit.  Best to leave it alone until you have the right equipment.

Always tons to learn, regardless of where you are.  Fun, though.

Harvey

Tons to learn!
Dave




Join TekScopes@groups.io to automatically receive all group messages.