Maker Pro
Maker Pro

DIY sampling "scope"

P

PFC

Jan 1, 1970
0
Hi guys !

I'm looking for a solution to measure PCBs, power supply decoupling
performance, input impedance of stuff that has inputs, etc...
I've got a 60 MHz network analyzer, works great, but I'd like to go a bit
higher in frequency, without spending megabucks.
And suddenly, this post from 2008 :
If it did TDR, you could sell bundles of them to PCB houses, and to
engineers and QC people who care about trace impedances. Especially if
it were cheap and interfaced to a PC, so it could document pcb test
coupons.
I'd design the fast stuff if somebody else handles the USB part and
the PC software. I have a slick deconvolution algorithm that would run
on the pc side and really beautify the TDR response.

So, if someone here has interest in designing the fast stuff (or just
throwing some schematics), I'd gladly volunteer for the rest (usb, micro,
adc, software, gui, etc).

Basically a board which :
- outputs a fast pulse or edge
- has an input with a sampler running synchronous to the output, with a
delay sweep
- ADC, usb, pc, display

With a Cypress FX2 ez-usb chip, streaming USB2 at full bandwidth is
simple, I've got a board with that running from a previous project.

Anyone interested ?
 
N

Nico Coesel

Jan 1, 1970
0
PFC said:
Hi guys !

I'm looking for a solution to measure PCBs, power supply decoupling
performance, input impedance of stuff that has inputs, etc...
I've got a 60 MHz network analyzer, works great, but I'd like to go a bit
higher in frequency, without spending megabucks.
And suddenly, this post from 2008 :





So, if someone here has interest in designing the fast stuff (or just
throwing some schematics), I'd gladly volunteer for the rest (usb, micro,
adc, software, gui, etc).

Basically a board which :
- outputs a fast pulse or edge
- has an input with a sampler running synchronous to the output, with a
delay sweep
- ADC, usb, pc, display

With a Cypress FX2 ez-usb chip, streaming USB2 at full bandwidth is
simple, I've got a board with that running from a previous project.

Thats the easy part. I've been playing around with a similar idea. The
biggest problem is triggering on signals with frequencies in the GHz
range.
 
N

Nico Coesel

Jan 1, 1970
0
John Larkin said:
If it's just TDR, it doesn't need a trigger. TDR is probably the best,
concentrated market to sell to. The old Tek 11801-series scopes are failing, and
the Polar Instruments stuff is slow and ancient. There's room for a new player.

Aren't there a lots of TDR kits out there already? If I enter 'usb
tdr' into Google I get lots of companies selling USB TDR kits.
What will be a bunch of work would be a good PC GUI application that PCB houses
or receiving inspection folks could use to acquire and archive PCB impedance
tests.

Software is an important selling point these days.
 
P

PFC

Jan 1, 1970
0
Thats the easy part. I've been playing around with a similar idea. The
biggest problem is triggering on signals with frequencies in the GHz
range.

I wasn't thinking about building a GHz scope, that would be way too
ambitious.
What I'd like would be a small part of what such a machine can do, for a
very small fraction of the price.
I'd just like to send a pulse into stuff and look at what comes out, at
hobby prices (ie, couple hundred bucks not thousands).
That would be an extremely useful tool. With separate output and input, I
could also send the stimulus to logic circuits or mosfet drivers, and see
what comes out, probe the supplies with a wide bandwidth, etc.

BTW it's not a commercial project, I live in a mostly communist country
anyway, so profits ... lol.
 
S

Spehro Pefhany

Jan 1, 1970
0
NNTP-Posting-Host: 88.168.96.119
ip_to_country -i 88.168.96.119
ip=88.168.96.119 (1487429751) "FR" "FRANCE"
??????????????

FR is more communist than CN if you look at the percentage of GDP that
is government controlled.
 
N

Nico Coesel

Jan 1, 1970
0
PFC said:
I wasn't thinking about building a GHz scope, that would be way too
ambitious.

It would make sense to make a GHz scope. There are so many TDR boxes
out there already that it would be hard to come up with an ultra low
cost solution that adds something to the current landscape.
 
What I'd like would be a small part of what such a machine can do, for a

very small fraction of the price.

I'd just like to send a pulse into stuff and look at what comes out, at

hobby prices (ie, couple hundred bucks not thousands).

That would be an extremely useful tool. With separate output and input, I

could also send the stimulus to logic circuits or mosfet drivers, and see

what comes out, probe the supplies with a wide bandwidth, etc.

Sounds like good fun. Do keep us informed!
 
P

PFC

Jan 1, 1970
0
The fast stuff isn't all that hard at, say, 50 ps or so rise time.

That's way faster than I'd need, but if you can pull this off on FR-4
without megabucks, then I'm drooling.
Some thought
will have to be put into connectors, fixturing, interfacing to PCB test
coupons.

What about using SATA, USB3, or other cheap, mass-produced, highspeed
cables for signal ?
What will be a bunch of work would be a good PC GUI application

I'll try to make a quick mockup.

How many bits are realistic in the ADC ? 8 ? or is it possible to hope for
more ?

For a TDR, there is :

- a clock, used to generate a pulse
- and a controlled delay, used to trigger the sampler

The delay could be a fast ramp + comparator, the ECL delay chip whose ref
I don't remember, or one of the new, interesting fast clock buffer chips
with embedded PLL and adjustable delay.

Using 2 oscillators doesn't seem too good, since it relies on long-term
jitter being low, a delay relies only on period-to-period jitter, which is
easier to get low.

For a scope,

- a trigger (ie, highspeed low jitter comparator)
- and a controlled delay

An equivalent-time scope can only examine perfectly periodic signals, with
a very short period, and for triggering, it needs an edge on the signal
that is always at the same spot... that's a pretty large restriction...
observing more kinds of signals means more trigger complication.

OK, feature creep kills projects, so a possibility of plugging in a
trigger/delay daughterboard (replacing the default clock/delay) would be
nice.
 
T

Tim Williams

Jan 1, 1970
0
John Larkin said:
12 is easy, but if you do a slideback sampler, you essentially make your
own
1-bit ADC. That's in the 1964 GE transistor manual.

I'm looking at the manual but don't see anything of the sort. Cite?

Tim
 
M

mike

Jan 1, 1970
0
That's way faster than I'd need, but if you can pull this off on FR-4
without megabucks, then I'm drooling.


What about using SATA, USB3, or other cheap, mass-produced, highspeed
cables for signal ?


I'll try to make a quick mockup.

How many bits are realistic in the ADC ? 8 ? or is it possible to hope
for more ?

For a TDR, there is :

- a clock, used to generate a pulse
- and a controlled delay, used to trigger the sampler

The delay could be a fast ramp + comparator, the ECL delay chip whose
ref I don't remember, or one of the new, interesting fast clock buffer
chips with embedded PLL and adjustable delay.

Using 2 oscillators doesn't seem too good, since it relies on long-term
jitter being low, a delay relies only on period-to-period jitter, which
is easier to get low.

For a scope,

- a trigger (ie, highspeed low jitter comparator)
- and a controlled delay

An equivalent-time scope can only examine perfectly periodic signals,

That's restrictive. All you really need is jitter statistics such
that you get enough samples falling in the view window to produce
a trace in the time you're willing to wait.
with a very short period, and for triggering, it needs an edge on the
signal that is always at the same spot... that's a pretty large
restriction... observing more kinds of signals means more trigger
complication.

OK, feature creep kills projects, so a possibility of plugging in a
trigger/delay daughterboard (replacing the default clock/delay) would be
nice.

It's important to realize the difference between a hobby project and a
business plan.
A typical example is the spectrum analyzer.
Yes, there are lots of hobby projects that take a TV tuner and create
some blips on the screen.
But 30 Hz. resolution bandwidth and -124 dBm noise floor and 21GHz.
range is 40 year old technology.

It's easy to think that you can engineer/produce/support a TDR system
for less than an existing company with established
marketing/service/manufacturing infrastructure.

It's easy to envision simpler ways to do stuff...unitl you try to
implement it.
The devil is in the details.

And, yes, innovation happens...but you won't likely get the secret
sauce in a newsgroup...at least not from anyone who know how the
patent system works.
 
B

Bill Sloman

Jan 1, 1970
0
FR is more communist than CN if you look at the percentage of GDP that
is government controlled.

Communism is dead. France is moderately socialist, but elitist and
more inclined than most to believe in the advantages of central
planning, but they aren't silly enough to let the highly educated
bureaucrats in the central planning office execute any of the plans.
That's left to regular commercial companies operating in the free
market. Quite a few of them have significant state share-holdings, but
French productivity is respectable and they have been known to do good
stuff, though their "experts" do tend to be a little impractical.

The French bourgeoisie hate income taxes, and the French collect most
tax as valued-added tax, which is not progressive and in fact collects
a rather lower proportion of income from the rich than from the rest
of the population. Their income inequality - Gini index of 32.7% - is
marginally worse than Canada's at 32.6%

http://en.wikipedia.org/wiki/List_of_countries_by_income_equality

but it's way lower than China (42.5%), the US (40.8%) or
Russia(40.1%), where the ruling oligarchies reserve most the goodies
for themselves.
 
B

Bill Sloman

Jan 1, 1970
0
That's way faster than I'd need, but if you can pull this off on FR-4
without megabucks, then I'm drooling.


What about using SATA, USB3, or other cheap, mass-produced, highspeed
cables for signal ?


I'll try to make a quick mockup.

How many bits are realistic in the ADC ? 8 ? or is it possible to hope for
more ?

For a TDR, there is :

- a clock, used to generate a pulse
- and a controlled delay, used to trigger the sampler

The delay could be a fast ramp + comparator, the ECL delay chip whose ref

Motorola/ON-Semiconductor MC100EP195

http://www.onsemi.com/pub_link/Collateral/MC10EP195-D.PDF

It does 10nsec, so you have to use it with a 100MHz oscillator,
counting clock edges for longer delays. Temperature stability isn't
great (see figure 4 in the data sheet) and when I was planning on
using it's predecessor I figured on automatic recalibration every few
minutes - you use it to set up a pulse-width modulated waveform and
measure the DC level of the filtered output with a decent A/D
converter - 16-bits IIRR.

It might be easier to measure it's temperature with a thermistor and
use a resistor to keep it's temperature stable.
I don't remember, or one of the new, interesting fast clock buffer chips
with embedded PLL and adjustable delay.

DDS chip?
Using 2 oscillators doesn't seem too good, since it relies on long-term
jitter being low, a delay relies only on period-to-period jitter, which is
easier to get low.

For a scope,

- a trigger (ie, highspeed low jitter comparator)
- and a controlled delay

Protecting the fast comparator, and giving it a balanced signal -
external trigger versus external ground - can keep busy for a while.

<snipped the obvious>
 
F

Fred Abse

Jan 1, 1970
0
Page 372.

That's the full 7th edition, cost $2. There was also a "compact" edition.

PDF scan posted to a.b.s.e.

Where you'd get the TDs at sensible prices, I don't know.
 
T

Tim Williams

Jan 1, 1970
0
John Larkin said:
That's the full 7th edition, cost $2. There was also a "compact"
edition.

Ah, found it. But this is self-triggered, so it can't be used for
equivalent time sampling a waveform, only the highest point on the waveform.

You could add a pulse generator to it, but then you need the timing
generation, and the pulse generator, which again necessitates an accurate
trigger generator. I don't see how this helps.

Tim
 
F

Fred Abse

Jan 1, 1970
0
TDs were cool. Some would generate 25 ps steps, in 1964.

I have a Tek TDR/sampler plugin that does exactly that.

Stated to be <25ps. rise. As far as I can measure, it is.
 
F

Fred Abse

Jan 1, 1970
0
Which model? The SD-24 does 17 ps, but uses SRDs.

7S12. I don't have the manual for the pulser, only the main TDR, but the
waveform looks very TD, and it's only 300mV.
 
M

mike

Jan 1, 1970
0
7S12. I don't have the manual for the pulser, only the main TDR, but the
waveform looks very TD, and it's only 300mV.
Don't know which head you have.
The S-50 is a 50mA TD.
The S-52 is a TD, but couldn't find the spec in the time I was willing
to spend.
Probably 50 or 100mA.
 
Top