Any god way to esperimentally determine the characteristic impedence of a backplane?

[Norm Dresner]

I have several old computer backplanes which were originally designed for
only modest (Typ Z-80) speed signals which I'd like to use with more
vigorous signals and I'd like to have the lines properly terminated. Is
there any simple experimental procedure to determine the characteristic
impedence of the backplane?

TIA
Norm

 

[Joerg]

Depends on how wide they are. If you have a pulse generator with sub-nsec
rise times and a fast scope (doesn't have to be a single shot type) you could
drive a line. Then vary the terminating resistor until the reflection
disappear. That should be the impedance. Of course you could also saw one
apart, measure the dimensions and calculate it out using the formulas in an
older Motorola MECL databook. They used to have a nice section about
transmission lines.

Regards, Joerg.

 

[Tom Bruhns]

I suppose it depends on the length of the lines and what you have in
the way of test equipment. If you have a fast 'scope and a fast step
generator (e.g., just a nice low-output-impedance digital line driver
like perhaps a 74AC244), it should be quite easy. Set up the line
driver with a square wave input, perhaps 10kHz. Connect the output to
one of the lines you want to check, through a resistor. Anything
from, say, 33 ohms to 150 ohms would be fine--it's not critical,
though matching the line impedance is best. Connect the other end of
the line to a variable resistance. 100 ohms max might work, but 200
or 250 would be safer to be sure you can adjust it to the line's
impedance. Connect a scope to the input end of the line; set the
sweep speed to about 5 nanoseconds/division. Adjust the resistor to
get the best looking single step, with minimum evidence of
reflections. (Adjust it up and down to see what an improperly
terminated line does, and why that's bad at high speeds!) The
resistor value will be the line impedance. That assumes it's a
reasonably constant impedance. If there are discontinuities in the
line, they will cause wiggles reflected back to the input depending on
how far away from the input end they are. All this is just
time-domain reflectometry. You can also do an equivalent analysis in
the frequency domain with a network analyzer...adjust the resistor to
make the impedance versus frequency as constant as possible. You need
to cover frequencies up to where the line is about half a wavelength
long, accounting for its propagation velocity, for best sensitivity.
The resistor, of course, should not be wirewound for either of these
tests! A small cermet trimmer should work fine. You can also
estimate the impedance fairly accurately from the dimensions, if you
know the spacing to the backplane ground plane. (I've assumed in all
this that you actually do have a ground plane in there; otherwise,
it's probably a nonsense question in the first place!)

Cheers,
Tom

 

[Winfield Hill]

Norm Dresner asked "Any god way to esperimentally determine..."

Not even I know what God's way would be to determine a backplane's
characteristic impedance.

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Norm Dresner]

I suppose it depends on the length of the lines and what you have in
the way of test equipment. If you have a fast 'scope and a fast step
generator (e.g., just a nice low-output-impedance digital line driver
like perhaps a 74AC244), it should be quite easy. Set up the line

[SNIP]

Thanks to all who answered with the detailed procedure for experimentally
determining the characteristic impedence. From what I've gathered, the
answer to the question in the subject is a sort of qualified "Yes" but
there's no trivial way [I kind of knew that but there's always hope]. All
of that said, one poster mentioned the Motorola MECL data book but I can't
find mine ... it may be long gone to the dump ... is there any on-line info?

Thanks
Norm

 

[Joerg]

Hi Norm,

Missouri-Rolla University has a pretty cool "electronic cheat sheet" with the
basics of transmission line calcs. I believe you must enable Java script to run
it:

http://www.emclab.umr.edu/pcbtlc/

Hint: I you ever see an old Moto MECL databook being thrown out somewhere, or on
a yard sale, grab it. No matter how weathered it is. I would never part with
mine.

Good luck,

Joerg.

 

[John Woodgate]

I read in sci.electronics.design that Winfield Hill
about 'Any god way to esperimentally determine the characteristic
impedence of a backplane?', on Fri, 9 Apr 2004:

Not even I know what God's way would be to determine a backplane's
characteristic impedance.

He wrote 'any god'. We have a choice. But I can't think of any god whose
aptitudes extend to this sort of thing. (;-)

 

[Mac]

I have several old computer backplanes which were originally designed for
only modest (Typ Z-80) speed signals which I'd like to use with more
vigorous signals and I'd like to have the lines properly terminated. Is
there any simple experimental procedure to determine the characteristic
impedence of the backplane?

TIA
Norm

Let's say you drive the transmission line with a realy fast logic chip in
series with a known resistor. Until the reflection comes back from the far
end of the transmission line, it looks like a resistor of its
characteristic impedance. So, if you can generate a fast enough edge, and
have a fast enough scope, you can use the simple voltage divider rule to
calculate the characteristic impedance.

For example lets say the resistor is 49.9 Ohms. If you are driving with
some kind of CMOS logic at 3.3 Volts, you should see (roughly) a single
step up to almost 3.3 Volts on the driver side of the resistor. But on the
transmission line side of the resistor, you will see some other voltage.
Maybe 1.65 Volts, for example, if the backplane is 50 Ohms. But when the
reflection comes back from the transmission line, the 1.65 Volts will
change to something else (probably 3.3 Volts.).

So the waveform on the transmission line side of the resistor has a step
in it like this:

------------
/
V1 -----/
/
----------------/

time --->


You can use V1/V2 = Z0/(Z0 + R)

Where V1 is the lower step level (on the transmission line side), V2 is
the first step level on the driver side of the resistor, Z0 is the
characteristic impedance, and R is the known resistor.

I'll let you solve for Z0.

HTH

--Mac

 

[Winfield Hill]

John Woodgate wrote...

Winfield Hill wrote


He wrote 'any god'.

Yes, you're right, there are many gods.

We have a choice. But I can't think of any god whose
aptitudes extend to this sort of thing. (;-)

Not even the god of backplane characteristic impedance?

Thanks,
- Win

whill_at_picovolt-dot-com

 

[Norm Dresner]

I have access to just about any type of bipolar TTL gates, 74, 74LS, 74S, a
few 74H, some 74ALS, and a few CMOS parts. I have an old but quite
serviceable TEK 465B 'scope. Is this adequate for the determination?

TIA
Norm

I suppose it depends on the length of the lines and what you have in
the way of test equipment. If you have a fast 'scope and a fast step
generator (e.g., just a nice low-output-impedance digital line driver
like perhaps a 74AC244), it should be quite easy. Set up the line

[SNIP]

Thanks to all who answered with the detailed procedure for experimentally
determining the characteristic impedence. From what I've gathered, the
answer to the question in the subject is a sort of qualified "Yes" but
there's no trivial way [I kind of knew that but there's always hope]. All
of that said, one poster mentioned the Motorola MECL data book but I can't
find mine ... it may be long gone to the dump ... is there any on-line info?

Thanks
Norm

 

[John Woodgate]

I read in sci.electronics.design that Winfield Hill
about 'Any god way to esperimentally determine the characteristic
impedence of a backplane?', on Sat, 10 Apr 2004:

John Woodgate wrote...

Yes, you're right, there are many gods.


Not even the god of backplane characteristic impedance?

Who would that be? Zo-roaster?

 

[Charles W. Johnson Jr.]

I read in sci.electronics.design that Winfield Hill
about 'Any god way to esperimentally determine the characteristic
impedence of a backplane?', on Sat, 10 Apr 2004:
Who would that be? Zo-roaster?
--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Z-eus I do believe

Charles

 

[Joerg]

Norm, 74S or ALS may do but ECL would be better. The 465B is way too slow, you
need a GHz scope, at least for regular sized backplanes. Maybe a friend has one?

Regards, Joerg.

 

[Tim Hubberstey]

... one poster mentioned the Motorola MECL data book but I can't
find mine ... it may be long gone to the dump ... is there any on-line info?

Fortunately for all of us, the good folks at ON Semiconductor have made
the entire text of the "MECL System Design Handbook" available online at:

http://www.onsemi.com/pub/Collateral/HB205-D.PDF

Be forewarned, it is a hefty 27 MB as it appears to have been scanned
from a paper copy.

 

[Spehro Pefhany]

Fortunately for all of us, the good folks at ON Semiconductor have made
the entire text of the "MECL System Design Handbook" available online at:

http://www.onsemi.com/pub/Collateral/HB205-D.PDF

Be forewarned, it is a hefty 27 MB as it appears to have been scanned
from a paper copy.

Wow. Look at the picture on PDF page 73 "Production area for MECL
Integrated Circuits". I count 3 or 4 blondes. Mid-sixties USofA?
(guessing from the hairstyles).


Best regards,
Spehro Pefhany

 

[Jim Thompson]

Wow. Look at the picture on PDF page 73 "Production area for MECL
Integrated Circuits". I count 3 or 4 blondes. Mid-sixties USofA?
(guessing from the hairstyles).


Best regards,
Spehro Pefhany

Yep, looks like it. I can remember a gal, nicknamed "the silver
bullet", who used to stop the world when she entered the cafeteria.
The hair was silver... I'll leave it to you to determine what the
"bullets" were ;-)

...Jim Thompson

 

[Spehro Pefhany]

Yep, looks like it. I can remember a gal, nicknamed "the silver
bullet", who used to stop the world when she entered the cafeteria.
The hair was silver... I'll leave it to you to determine what the
"bullets" were ;-)

...Jim Thompson

ROTFL. "Torpedo T*ts". I guess they've fallen by the wayside..

Best regards,
Spehro Pefhany

 

[John Miller]

ROTFL. "Torpedo T*ts". I guess they've fallen by the wayside..

Most of those from the '60s probably have.

--
John Miller
Email address: domain, n4vu.com; username, jsm

H. L. Mencken suffers from the hallucination that he is H. L. Mencken --
there is no cure for a disease of that magnitude.
-Maxwell Bodenheim

 

[Active8]

Yep, looks like it. I can remember a gal, nicknamed "the silver
bullet", who used to stop the world when she entered the cafeteria.
The hair was silver... I'll leave it to you to determine what the
"bullets" were ;-)

...Jim Thompson

"Silver bullet", according to Howard the 7th Med Com pharmacist
atached to Pendleton Barracks, Giessen, BRD around 1986, referred to
a shot of penecillin. I'll leave it to you to guess what the shot
was for and why his unit shouted "588th - silver-bullet" at the
maintainance battalion everytime they passed them on their PT run.

No pictures of any H2O2 users in my MotDL122/D rev5, though. But it
was free

 

[Mac]

I have access to just about any type of bipolar TTL gates, 74, 74LS, 74S, a
few 74H, some 74ALS, and a few CMOS parts. I have an old but quite
serviceable TEK 465B 'scope. Is this adequate for the determination?

TIA
Norm

It all depends on how long the backplane traces are. 1 foot is about 2 ns,
or 4ns round trip. So if the traces are around 12 inches (30 cm) long,
then you will need somewhere around 1GHz or so of bandwidth, I would
guess, to see things clearly. But if the traces are longer, like say 24"
(61 cm) then maybe your scope will pick it up. The 465 is a 400 MHz scope,
isn't it?

Oh, and you will still need very fast rise times. If the rise time is 5 ns
or something, you don't have a chance. The step has to reach its plateau
before the reflection comes back, so the rise time should be shorter than
the round-trip flight time of the signal.

You've got very little to lose from hooking up something and trying it
out. I would start by measuring the rise time of all your available logic
chips when driving a heavy load. I'm guessing that the backplane impedance
will be somewhere in the range of 50-100 Ohms, so you could measure the
rise times when driving a 75-Ohm load. Whichever chip has the shortest
rise time is your winner.

[snip]

--Mac