Tek 468 and quantization?

[John Smith]

Hi -

I bought a Tek 468 which I returned today. I thought it had a problem, but
the seller disagreed. I would like to know if I am wrong.

The 468 is an analog/digital scope. In the digital mode, I noticed that the
trace had rectangular pulses on it of about .1 cm amplitude and apparently
random in frequency and width. This was with the inputs grounded using the
switch on the front panel. The amplitude of these pulses did not change when
I changed the attenuator setting except at the three most sensitive settings
of the attenuator (2 mV, 1 mV and .5 mV). Those three most sensitive
settings were active for digital storage operation only and were
non-functional in the analog mode.

When I complained about it, the seller said that it was caused by
quantization, it met Tek's specifications, and that .1 cm was entirely
reasonable. I disagreed because all waveforms appeared noisy so one would
never be able to tell whether the observed waveform was at fault or the
scope was at fault. In addition, .1 cm is 1 part in 80 for the vertical
displacement (8 cm vertical grid). It didn't seem right that maximum
resolution would be 1 in 80 or even 1 in 100. Logically, I felt that the
resolution would be 1 in 256 so that quantization noise would be much less
than actually observed.

Maybe I understand nothing about scopes and digital stuff. But, I could not
believe that Tek would sell such a product. Am I wrong?

Thanks for any comments.

John

 

[Frank Bemelman]

Hi -

I bought a Tek 468 which I returned today. I thought it had a problem, but
the seller disagreed. I would like to know if I am wrong.

The 468 is an analog/digital scope. In the digital mode, I noticed that the
trace had rectangular pulses on it of about .1 cm amplitude and apparently
random in frequency and width. This was with the inputs grounded using the
switch on the front panel. The amplitude of these pulses did not change when
I changed the attenuator setting except at the three most sensitive settings
of the attenuator (2 mV, 1 mV and .5 mV). Those three most sensitive
settings were active for digital storage operation only and were
non-functional in the analog mode.

When I complained about it, the seller said that it was caused by
quantization, it met Tek's specifications, and that .1 cm was entirely
reasonable. I disagreed because all waveforms appeared noisy so one would
never be able to tell whether the observed waveform was at fault or the
scope was at fault. In addition, .1 cm is 1 part in 80 for the vertical
displacement (8 cm vertical grid). It didn't seem right that maximum
resolution would be 1 in 80 or even 1 in 100. Logically, I felt that the
resolution would be 1 in 256 so that quantization noise would be much less
than actually observed.

It is normal. The ADC's span is bigger than the 8 squares on the grid.
One-tenth of a grid square 'uncertainy' doesn't sound alarming. Of course
that 1-bit noise rides on top of everything. You also see this very clear
on a new LCD scope like a TDS220 or something. Perhaps less annoying with
these scopes, because it's fast realtime update tends to average it out,
sort of. But when you stop the aqcuisition, it is all over the place.

Maybe I understand nothing about scopes and digital stuff. But, I could not
believe that Tek would sell such a product. Am I wrong?

Tek has sold more rubbish TAS220 for instance.

 

[John Larkin]

Hi -

I bought a Tek 468 which I returned today. I thought it had a problem, but
the seller disagreed. I would like to know if I am wrong.

The 468 is an analog/digital scope. In the digital mode, I noticed that the
trace had rectangular pulses on it of about .1 cm amplitude and apparently
random in frequency and width. This was with the inputs grounded using the
switch on the front panel. The amplitude of these pulses did not change when
I changed the attenuator setting except at the three most sensitive settings
of the attenuator (2 mV, 1 mV and .5 mV). Those three most sensitive
settings were active for digital storage operation only and were
non-functional in the analog mode.

When I complained about it, the seller said that it was caused by
quantization, it met Tek's specifications, and that .1 cm was entirely
reasonable. I disagreed because all waveforms appeared noisy so one would
never be able to tell whether the observed waveform was at fault or the
scope was at fault. In addition, .1 cm is 1 part in 80 for the vertical
displacement (8 cm vertical grid). It didn't seem right that maximum
resolution would be 1 in 80 or even 1 in 100. Logically, I felt that the
resolution would be 1 in 256 so that quantization noise would be much less
than actually observed.

Maybe I understand nothing about scopes and digital stuff. But, I could not
believe that Tek would sell such a product. Am I wrong?

Thanks for any comments.

John

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

John

 

[Paul Burridge]

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

Sounds like a heap of shit. Why don't you use a good ol' classic
analogue scope like this one I rescued from a dumpster and shoved into
my garage for posterity:

http://www.burridge8333.fsbusiness.co.uk/oscilloscope.jpg

No quantization noise guaranteed!

 

[Frank Bemelman]

Sounds like a heap of shit. Why don't you use a good ol' classic
analogue scope like this one I rescued from a dumpster and shoved into
my garage for posterity:

http://www.burridge8333.fsbusiness.co.uk/oscilloscope.jpg

No quantization noise guaranteed!

You need to be a bit more convincing to make John want to
swap oscilloscopes with you

 

[Georg Acher]

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

The "cheaper" scopes use CCD shift registers and multiple ADCs for
sampling. That saves the exponentially increasing price for faster ADCs, but
introduces additional noise and common mode problems...

 

[John Smith]

It is normal. The ADC's span is bigger than the 8 squares on the grid.
One-tenth of a grid square 'uncertainy' doesn't sound alarming. Of course
that 1-bit noise rides on top of everything. You also see this very clear
on a new LCD scope like a TDS220 or something. Perhaps less annoying with
these scopes, because it's fast realtime update tends to average it out,
sort of. But when you stop the aqcuisition, it is all over the place.


Tek has sold more rubbish TAS220 for instance.

It is disappointing to learn they produced such a piece of equipment.
Thanks, Frank, for the information.

John

 

[John Smith]

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

John

Hi, John -

I used a late-model Tek digital storage scope in my last job less than a
year ago. I did not see such trash on the scope. That's why I was so
surprised and convinced something was wrong with it. Well, thanks for your
input (and Frank's).

John

 

[John Larkin]

Sounds like a heap of shit. Why don't you use a good ol' classic
analogue scope like this one I rescued from a dumpster and shoved into
my garage for posterity:

http://www.burridge8333.fsbusiness.co.uk/oscilloscope.jpg

No quantization noise guaranteed!

Why not?

Color
Infinite storage
Expansion and panning of stored traces
Measurement cursors
A 545 wouldn't fit on my desk
A 545 would be hard to carry up the stairs
It's warm in my office already

I like the 545B. I have one or two, I forget. It's a classic 545 with
transistors replacing many of the tubes, especially the vertical
distributed amplifier.

See pic in abse.

John

 

[Bob Stephens]

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

John

I had a TDS2012 on my desk and the quiescent noise was so bad I called up
the metrology lab we use for a repair quote. They said that this is a known
problem with this series of 'scopes and that no one including Tektronix
will fix them. He maintained that it is a condition that worsens with age
rather than an inherent design flaw.

Bob

 

[John Larkin]

I had a TDS2012 on my desk and the quiescent noise was so bad I called up
the metrology lab we use for a repair quote. They said that this is a known
problem with this series of 'scopes and that no one including Tektronix
will fix them. He maintained that it is a condition that worsens with age
rather than an inherent design flaw.

Bob

This one is a couple of years old, and I haven't noticed the noise
increasing. 1 mv p-p noise ain't at all bad for a 100 MHz scope; it
*does* go down to 2 mv/cm, which few analog bench scopes will do.
Heck, a cheap analog scope won't focus to better than 0.1 cm, and the
Tek will signal average up to x128, too.

I like it, anyhow.

John

 

[scada]

Hi -

I bought a Tek 468 which I returned today. I thought it had a problem, but
the seller disagreed. I would like to know if I am wrong.

The 468 is an analog/digital scope. In the digital mode, I noticed that the
trace had rectangular pulses on it of about .1 cm amplitude and apparently
random in frequency and width. This was with the inputs grounded using the
switch on the front panel. The amplitude of these pulses did not change when
I changed the attenuator setting except at the three most sensitive settings
of the attenuator (2 mV, 1 mV and .5 mV). Those three most sensitive
settings were active for digital storage operation only and were
non-functional in the analog mode.

When I complained about it, the seller said that it was caused by
quantization, it met Tek's specifications, and that .1 cm was entirely
reasonable. I disagreed because all waveforms appeared noisy so one would
never be able to tell whether the observed waveform was at fault or the
scope was at fault. In addition, .1 cm is 1 part in 80 for the vertical
displacement (8 cm vertical grid). It didn't seem right that maximum
resolution would be 1 in 80 or even 1 in 100. Logically, I felt that the
resolution would be 1 in 256 so that quantization noise would be much less
than actually observed.

Maybe I understand nothing about scopes and digital stuff. But, I could not
believe that Tek would sell such a product. Am I wrong?

Thanks for any comments.

John

I have two Tek Digital scopes in the shop. One is color with a floppy drive,
the other just a two channel B&W. I have noticed the same problem on both.
I've been told this is inherent! I just figure its a trade off for the
benefits of going digital. I still keep a couple of old analog scopes around
to measure the small stuff.

 

[Frank Bemelman]

It is disappointing to learn they produced such a piece of equipment.
Thanks, Frank, for the information.

Well, it is still useful to view single events. Typically, a 1/10th
square of noise is not interesting. You look at the general shape of
the waveform, in such cases anyway.

A much more annoying aspect of older digital scopes is their low
sampling frequency, which gives all kinds of aliasing problems.
You always have to ask yourself if what you see actually makes
sense

 

[Pooh Bear]

I have a TDS2012 on my desk, displaying RS232 poll/reply sequences for
a uP thing I'm testing. With no input, it has about 0.1 cm p-p noise
at most gain settings, increasing to 0.5 cm p-p at 2 mv/cm. The noise
appears, to me, to be a mixture of real front-end analog noise, adc
quantization, and lcd pixellization. Welcome to the digital age!

Thank heavens for my old 465 !

I even had some old valve scopes before that such as the 545 !


Graham

 

[John Larkin]

I had a TDS2012 on my desk and the quiescent noise was so bad I called up
the metrology lab we use for a repair quote. They said that this is a known
problem with this series of 'scopes and that no one including Tektronix
will fix them. He maintained that it is a condition that worsens with age
rather than an inherent design flaw.

Bob

I wonder if the increasing noise is an artifact of using multiple
interleaved ADCs to get the high sample rate. If the ADCs drift, it
will look like noise. At max gain, 2 mv/cm, the noise on my 2012 does
look bimodal, so it may be an adc interleave problem. There's likely a
cal factor somewhere that would fix that. An ideal design would
automatically synchronize the ADCs.

Interleaving fast ADCs is a non-trivial problem.

John

 

[john jardine]

Sounds like a heap of shit. Why don't you use a good ol' classic
analogue scope like this one I rescued from a dumpster and shoved into
my garage for posterity:

Looks like the same one I threw in 10 years ago
regards
john

 

[Paul Burridge]

See pic in abse.

That thing ain't no scope! It's a pocket calculator. It ain't a real
scope unless you can tip it out of an office block window and kill
half a dozen people queuing for a bus. My ol' boat anchor qualifies;
your calculator doesn't.

 

[Paul Burridge]

Looks like the same one I threw in 10 years ago

Mine's actually in full working order (although I don't use it; just
admire it from a safe distance). It's quite formidable to see it all
fired-up, actually. They don't make 'em like that no mo' (global
warming concerns etc.)