Digital Signal Integrity and Bandwidth: Signal’s Risetime is Important, Period is Irrelevant

What bandwidth should your scope and scope probe have? When do you have to worry about “high frequency” effects? This is basic theory relating the time and frequency domains, but too often people quote rules of thumb that, while strictly accurate, are very misleading. Typical misleading statements are:

• the scope’s bandwidth should be at least 5 times the frequency, so you can see a reasonable waveform
• I only have a 1kHz (i.e. 1ms period) signal, which isn’t high frequency (certainly not RF), and an audio frequency scope is sufficient

This post is a basic illustration of the “maximum frequency” in a 1kHz signal, to help dispel some misconceptions.

Advertisements

Hints and Tips for Renovating an early Tektronix 475: Timebase, Switches, Potentiometers, and HF Response

I recently picked up an early Tek 475 scope plus service manual plus probes at a hamfest. I asked what was wrong, and the seller indicated the timebase was unreliable. Fair enough and, given the low price, at worst it would be a parts mule.

It was a bit mucky on the outside, as would be expected of something that had been in storage. A quick look the scope showed it was cleaner inside than I expected, and playing with it made me feel it was worth resuscitating. If nothing else, the puzzles involved would be more interesting and worthwhile than crosswords and soduku.

Overall the renovation involved some traditional problems and solutions, plus a couple that are less well-known – and a few diversions down blind alleys. Maybe the hints and tricks outlined below will save other people some time.

What could kids do in the summer holidays?

For no good reason other than avoid the >30C heat, I browsed some of the old usenet postings I’ve saved over the years. I’ve reproduced one below to illustrate how things have changed, maybe for the better, maybe not. I have a couple of such books, and can confirm that they do have chapters on building and using your own X-Ray machine, including such gems as noting that if your skin reddens you are probably using it too much.

Now I must get back to doing something more up-to-date, such as using the rather interesting XMOS xCORE processors to implement a 12.5MHz ratiometric reciprocal frequency counter purely in software. That’s possible since the have the best hard realtime architecture I’ve seen in decades. Highlights – which enable deterministic timings calculated by the IDE tools before execution – include:

• many 100MIPS cores
• no caches/interrupts
• switching fabric for i/o ports plus inter-core and inter-chip comms
• programming in xC, i.e. C with parallel extensions based on CSP with timing
• FPGA-like i/o ports: SERDES, strobed and master/slave ports,  timed input and timed output (4ns resolution), multiple programmable clocks

Great fun, and I don’t know of any other processor with those attributes.

A Weston standard cell – an introduction to voltnuttery?

Weston standard cells are pretty, but can they be more than a broken antique? And yes, that is liquid mercury visible at the bottom of the left leg.

One Cell

I remember using Weston cells at school during physics lessons to measure voltages more accurately that could be done with a standard analogue meter (DMMs being years in the future). The measurements are based on using a 1m ruler plus resistance wire plus an uncalibrated NiFe cell as a bridge, and use a Weston cell to calibrate the bridge. That technique (which is a good example of using simple materials, understanding and imagination – rather than merely throwing money at the problem) is outlined below.

Anyway, I have vaguely wanted a Weston cell ever since, a little more so since I got hold of an HP3468 and have wanted to check whether it is still “accurate”. I’ve wondered about buying one on ebay, but they are unjustifiably expensive and sufficiently fragile that I doubt they would survive shipping.

Unfortunately I’ve now found a cheap one, and I fear it might lead me down the expensive path of voltnuttery. Memo to self: do not become a voltnut.

Scope Probe Accessory: Higher Frequency Results

This takes the scope probe accessory described previously, and repeats the measurements with a higher frequency oscilloscope (>350MHz) and higher frequency probes (250MHz). The results are even more impressive at higher frequencies.

Renovating a Tektronix P6013A 12kV 100kHz Scope Probe

I’ve recently had to debug several scope’s 2kV-3kV HV supply and the CRT’s Z-axis waveforms at 2.5kV. So far I’ve got away with using a homemade 1000:1 voltage divider and a multimeter. Since that is crude and not particularly safe, I’m not going to mention the details in order to avoid someone apeing me and hurting themselves.

Then, at a recent auction I managed to pick up:

• a 40kV meter for measuring 17kV anode voltages, but which barely registers 2kV
• a Tek P6013A 12kV 1000:1 100kHz scope probe

The probe was functional but missing part of the handle. While not strictly necessary, I wanted to have a little fun fabricating the missing part…

Measuring Input Protection Components’ Effects With a TDR

My previous post used a Tektronix 1502 to examine discontinuities in cables. This post examines the discontinuity introduced by a “nominally invisible” protection diode on a PCB; it is clearly visible with the TDR, but probably won’t affect the final application.

A Traditional TDR Cable Tester With <2cm Resolution

What can you see, test and measure with a traditional time domain reflectometer (TDR)? The answer is “more than you might expect”:

• measure impedance variations in connectors/filters/antennas/PCBs
• locate short/open circuits and damage in cables
• locate intermittent faults in cables and connectors
• locate connectors in cables

and can resolve discontinuities around 2cm apart. That resolution is at least 10 times better than can be obtained with the typical homebrew logic pulse + oscilloscope combination.

I recently bought a couple of cheap 1970s Tektronix 1502s in the hope that I could make a single working frankenmachine.  My initial assessment was depressing: one had a cracked and broken case (so I assumed the CRT was also broken), the other’s electrolytic caps had spewed acid across the PSU and had a faulty 2kV PSU, and both had defective NiCd batteries – and it won’t even start without a working battery. But eventually I managed to get both working: I recapped the PSUs, rewelded the case with methylene chloride, used my “new” 12kV scope probe and 40kV meter to repair the HV PSU, created a “NiCd emulator”, and the CRT wasn’t damaged after all. Later reading of a TekScope magazine indicates it isn’t surprising the CRT survived: it is mechanically completely isolated from the chassis to protect against up to 26 12″ drops.

So I am now the proud possessor of two nice little portable waterproof instruments, literally designed for field use – one of the service manuals indicates they were used with Patriot missile defence systems.

Tektronix 1502 TDR Cable Tester

Rescuing a Broken Tektronix 465 CRT

A non-functional Tektronix 465 oscilloscope with a dented case and several serious faults was recently donated to my local HackSpace. Since they didn’t want another broken scope, I took it to see what was wrong. Eventually, after learning about CRT theory, with the help of very knowledgable people on the TekScopes forum, and gingerly measuring waveforms at -2450V, I came to the conclusion the CRT’s grid was faulty. Reluctantly removing the CRT, I found that it could not be repaired. Since parts of the CRT are rather beautiful, I decided to salvage those parts for display. This note outlines the original tube, the fault, and the process of turning it into something fit for a display cabinet.

SDR Utilities

I’ve been experimenting with an SDR dongle to see how it can be used as a 1.5GHz scalar network analyser, as a time domain reflectometer, and to estimate digital signal edge speeds. While doing that I’ve developed several general-purpose command-line utilities which capture and post-process power spectrums. These utilities are distributed with a MIT licence, and the code is at the bottom of this note.

• rtlscan: scans, measures and displays a raw scalar power spectrum, and saves the power spectrum in a CSV file
• rtlplot: displays one or more raw power spectrums
• rtldiff: displays one or more calibrated power spectrums, i.e. the difference between multiple raw spectrums and a single reference or calibration spectrum
• rtltdr: displays the impulse response implicit in a calibrated power spectrum

The graphs can have linear or logarithmic frequency axis, can be zoomed, panned and printed, and have text annotations added. The interaction with the SDR dongle is via Kyle Keen’s rtl_power program, so its capabilities and limitations are reflected in these utilities.

As is traditional, these are an unfinished work; I already know of the next change I would make, but I have not verified that it would be an improvement. Nonetheless, the utilities are usable.