Scope attenuators 1,2,5, why bother?

[Jan Panteltje]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V / div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

 

[Joerg]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

Like with all things that is what engineers have become used to. Change
it and you'll hear complaints. After all a foot still has 12 inches,
most currencies go 10-20-50-100, there are 12 months in a year and so
on. Remember when they wanted to change a full circle from 360 to 400
degrees? Nobody bit.

That 11V setting is a bit unorthodox though

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Most serious HW people do not think binary. Some of them think culinary ;-)

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V / div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

If you want to market the device my advice would be: Don't try to fix a
system that works.

 

[amdx]

So, while waiting for parts to arrive, I have been diddling sketches on an
A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC
scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits
shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V
/ div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per
division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even
if it
switched ranges in octal ;-)

Multiplying in your head is easy with the numbers 1, 2 or 5 but
1.28, 2.56 or 5.12 would tax my patience and make me question
why I bought that @$^*@ binary scope.
3.3 units x 200mv is a lot easier than 4.125 x 160mv.
Just my one, two or four cents worth.
Mike

 

[Jan Panteltje]

Like with all things that is what engineers have become used to. Change
it and you'll hear complaints. After all a foot still has 12 inches,
most currencies go 10-20-50-100, there are 12 months in a year and so
on. Remember when they wanted to change a full circle from 360 to 400
degrees? Nobody bit.

That 11V setting is a bit unorthodox though

Sorry, typo, it is 2 x 5.5 tough.

Most serious HW people do not think binary. Some of them think culinary ;-)

Well, once we had a little together in a large hall where we
transferred a million dollar electronics project I did.
I was there with Mr Big Boss, and we were all casually dressed.
Then there was this guy in black suit that came up to me, and asked if I wanted
some champagne.
I politely refrained, asked Mr Big Boss, who is that, the waiter?
No, he replied, that is the architect of this building.

If you want to market the device my advice would be: Don't try to fix a
system that works.

No, I do not want to market it, although anybody will be free to order some.
If it really works out OK I can have some PCBs made perhaps.
It is just an experiment.

 

[Jan Panteltje]

Multiplying in your head is easy with the numbers 1, 2 or 5 but
1.28, 2.56 or 5.12 would tax my patience and make me question
why I bought that @$^*@ binary scope.
3.3 units x 200mv is a lot easier than 4.125 x 160mv.
Just my one, two or four cents worth.
Mike

Very wise words.
Yes, but the idea is also to read voltages from the digital display
(as 1.55Vpp for example).
In such a case one could even use continuous variable gain to set sensitivity.

 

[D from BC]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V / div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

Sometimes the fastest way to get things done is to break the rules,
norms, standards, traditions, customs, codes, trends and styles.

Be electropunk!

Based on:
http://en.wikipedia.org/wiki/Cyberpunk


D from BC
British Columbia
Canada.

 

[Arlet Ottens]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

You could make the attenuator steps user programmable through some setup
menu.

BTW, why torture yourself with a PIC, when you can get a cheap LPCxxxx
ARM that can be programmed in C ? The ARM also has a fast multiplier
that makes scaling easy.

 

[Jan Panteltje]

On a sunny day (Wed, 16 Jan 2008 12:31:49 -0800) it happened D from BC

norms, standards, traditions, customs, codes, trends and styles.

Be electropunk!

Based on:
http://en.wikipedia.org/wiki/Cyberpunk

It is nice to see the word 'cybernetics' again.
I got a lot out of reading a book about cybernetics, when I was a kid.
Do not remember who wrote it, but it was very clear, and the best intro to
electronics trouble shooting perhaps.

 

[Jan Panteltje]

You could make the attenuator steps user programmable through some setup
menu.

mm, but the bit shifts will always be 1,2,4,8, etc..

BTW, why torture yourself with a PIC, when you can get a cheap LPCxxxx
ARM that can be programmed in C ? The ARM also has a fast multiplier
that makes scaling easy.

It is just an experiment, I may do it in FPGA again, as that at least gets
me into the > 50 MHz sampling, while PIC will be limited to some kHz (think 25kHz).

 

[[email protected]]

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

Trace? Real hackers just need the raw hex from the ADC.

 

[Arie de Muynck]

Very wise words.
Yes, but the idea is also to read voltages from the digital display
(as 1.55Vpp for example).
In such a case one could even use continuous variable gain to set
sensitivity.

I assume you will need to multiply values somewhere in the chain by a
calibration value anyway, so the input attentuators can be loosely coupled
with the readout steps. The gain switching could do 1/2/4/8/16/... while the
display would readout 1/2/5/10 sequence. For each display scale setting you
would use the best available gain range.

Arie de Muijnck

 

[John Larkin]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V / div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

My TDS2012 vertical steps 1-2-5, but the timebase steps 1-2.5-5.
Bizarre.

John

 

[Joerg]

My TDS2012 vertical steps 1-2-5, but the timebase steps 1-2.5-5.
Bizarre.

They pretty much all do that including mine (not a Tek).

 

[Joerg]

My TDS2012 vertical steps 1-2-5, but the timebase steps 1-2.5-5.
Bizarre.

Oh wait, I correct my previous post. Not all do that. The trusty old Tek
7000 in the lab goes 1-2-5 for the time base.

 

[John Devereux]

mm, but the bit shifts will always be 1,2,4,8, etc..


It is just an experiment, I may do it in FPGA again, as that at least gets
me into the > 50 MHz sampling, while PIC will be limited to some kHz (think 25kHz).

Analog Devices ADUC70xx are ARM7 based flash microcontrollers. They
will let you sample at 1MHz (in fact you can "overclock" the ADC to
2MHz), at 12 bits. It also has built-in 12 bit DACs - so it could be
an arbitrary waveform generator too! Or you could "play back" the
captured trace. There is a version with a 20MHz DDS and they all have
a (very simple) FPGA.

 

[Jan Panteltje]

Trace? Real hackers just need the raw hex from the ADC.

I have done that debugging some audio stuff

 

[Jim Backus]

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

It's a nominally log scale. The other commonly seen one is 1, 3, 10,
30 etc. 1, 2, 5 are very approximately 10^(1/3) and 1, 3, 10 are
approximately 10^(1/2)

 

[Joel Koltner]

Oh wait, I correct my previous post. Not all do that. The trusty old Tek
7000 in the lab goes 1-2-5 for the time base.

Have you ever posted a picture of your lab in one of the perennial "bench
photos" threads, Joerg? I'd certainly like to see all those boat anchors
I suspect you have in there!

 

[Joerg]

Have you ever posted a picture of your lab in one of the perennial "bench
photos" threads, Joerg? I'd certainly like to see all those boat anchors
I suspect you have in there!

I haven't but it ain't easy do do this in a photo. The main area
consists of the usual RF gear, things you'd see in any lab where serious
analog work is going on. The other stuff is used occasionally and thus
tucked into cabinets. The old (huge) Rohde&Schwarz SMF generator, the
Boonton megacycle meter, various self-builts and so on.

 

[john jardine]

So, while waiting for parts to arrive, I have been diddling sketches on an A4 piece
of paper, on how to design the the minimal cost attenuator, for the PIC scope I am planning.

Now I peaked at one scope I have here, and it has an 11 position switch
10mv, 20 mV, 50 mV, 100 mV, 200mV, 500mV, 11V, 2V, 5V, 10V, 20V

Now that sucks in a way, as counting to 5 is very difficult,
as we all think binary these days ;-)

The other thing that is interesting, is that the PIC ADC is 10 bits,
and I have a 64 pixel high LCD.
So I actually only need 6 bits (2^6 = 64 if it escaped you) for display.
That is still better then 2/100 or 2% accuracy, so fine with me.

So 10 bits in, and use only 6, then we can bit-shift, and use the 4 bits shift
to make attenuator steps of 2, 4, 8, 16
This reduces hardware (switching) people!

Then I was thinking:
Why not use binary on the settings? much easier.
So then you get sensitivity of:
10mV, 20mV, 40mV, 80mV, 160mV per division.
And maybe then continue, after switching gain ONCE:
320mV, 640mV, 1.28V, 2.56V, 5.12V per division.

Looks like a need for 2 more steps for higher voltage, 10.24V, and 20.48V / div.

For 64 pixels high and 8 divisions vertical that leaves 8 pixels per division.
makes a max voltage of 163.84V full screen (at 1x probe).

We can have a cursor on the trace, and a volts display, so who cares even if it
switched ranges in octal ;-)

Yep. Why bother with all those faffy ranges when you've massive dynamic
coverage on the PIC ADC.
Go something like 1:10:100