ACAM GP22 Chip
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for “work” purposes – I work for a company active in the weighing and force measurement world.
I should say from the start that I am new to time and frequency measurements and not even an electronics engineer – but then I have been exposed to high-precision electronics for the last 25 years hence have picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s experiment with the coaxial delay line from the HP5370b – but as my interest is in “measuring range 2” of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start). This is the equivalent of a 200 m length of cable. I fear trouble with this: Am I not getting unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far with 1 µs and are there other (simple/reliable) ways to achieve this kind of delay? I have tried it with a shorter piece of cable (around 2 ns which is measured in “range 1”), there I seem to get consistency virtually to within 100 ps. But I need to know if the device sticks to this level of performance when the periods are much longer, and thus measured in “range 2”.
Thanks and best regards,
Thomas.
Thomas
Welcome to the group. I am sure others will comment.
Many of us have a very wide range of experience and expertise so you should
feel comfortable with any question.
To the coax delay question. You are not pushing the limits.
But its important to understand the impacts of such long lines.
They need to be driven and terminated and the rise time will suffer from
the line capacitance. Essentially a fast rise time will become a slow
risetime on teh other end. There are lumped lc network delay lines. I have
experimented with them. They have the same effect. But you can cascade them
and use an inverter or buffer between each one.Each inverter also adds
delay. This helps the rise time issue. But the buffers add jitter and each
also adds delay thats temperature sensitive.
For cascaded delays of very short duration I have actually used 74LS244s
74HC244 line drivers cascaded and they work really well but only good for
each drivers delay.
Others will have better answers.
Regards
Paul
WB8TSL
On Tue, Nov 24, 2015 at 9:04 AM, Thomas Allgeier th.allgeier@gmail.com
wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at
for “work” purposes – I work for a company active in the weighing and force
measurement world.
I should say from the start that I am new to time and frequency
measurements and not even an electronics engineer – but then I have been
exposed to high-precision electronics for the last 25 years hence have
picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around
13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b – but as my
interest is in “measuring range 2” of the GP22 I need a delay of 500 ns or
more (actually 1 µs sounds a better start). This is the equivalent of a 200
m length of cable. I fear trouble with this: Am I not getting unwanted
inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far
with 1 µs and are there other (simple/reliable) ways to achieve this kind
of delay? I have tried it with a shorter piece of cable (around 2 ns which
is measured in “range 1”), there I seem to get consistency virtually to
within 100 ps. But I need to know if the device sticks to this level of
performance when the periods are much longer, and thus measured in “range
2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
my 2 cents:
- hack a RC delay + comparator (isolate well the R and C for temp
variations) - use a proper delay line (can be bought at digikey/mouser/etc)
Daniel
Em 24/11/2015 12:04, Thomas Allgeier escreveu:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for “work” purposes – I work for a company active in the weighing and force measurement world.
I should say from the start that I am new to time and frequency measurements and not even an electronics engineer – but then I have been exposed to high-precision electronics for the last 25 years hence have picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s experiment with the coaxial delay line from the HP5370b – but as my interest is in “measuring range 2” of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start). This is the equivalent of a 200 m length of cable. I fear trouble with this: Am I not getting unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far with 1 µs and are there other (simple/reliable) ways to achieve this kind of delay? I have tried it with a shorter piece of cable (around 2 ns which is measured in “range 1”), there I seem to get consistency virtually to within 100 ps. But I need to know if the device sticks to this level of performance when the periods are much longer, and thus measured in “range 2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Moin,
On Tue, 24 Nov 2015 14:04:32 -0000
"Thomas Allgeier" th.allgeier@gmail.com wrote:
We want to use this chip to measure the period of a square wave, of
around 13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
That's some modest requirement and should be doable with the GPS22
quite easily (or any other TDC for that matter). BTW: when specing
something like this, please make sure to mention whether 1ns is
1sigma, 3sigma or worst case/peak-to-peak. These 3 are quite different
requirements.
May I ask why you want to verify the specs of the GP22?
The specs say that it does something between 39ps and 70ps (1sigma),
which is probably way better than what you need. And as Acam is
a german company, I expect the datasheet to be accurate.
BTW: 1ns over 70us is approximately 14ppm. The GP22 uses the attached
crystal for absolute calibration. Please be aware that 14ppm will
require at least a TCXO to reach that level over the whole temperature
range, and depending on what TCXO you use, you might need to calibrate
the TCXO post-production and again after a couple of years of use.
Even if you don't need calibrate, I would add a TCXO frequency measurement
to the production test.
BTW2: the "we have a x ppm TCXO" value is usually misleading,
as that's the best-case, pre-soldering, pre-aging, pre-anything value.
The end-value can be 3 times as large... easily.
(unless you happen to choose one of the more honest manufacturers,
for example, like Abracon)
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b
For those wondering: "John A" is John Ackermann and the experiment
in question is documented at http://www.febo.com/pages/hp5370b/
– but as my interest
is in “measuring range 2” of the GP22 I need a delay of 500 ns or more
(actually 1 µs sounds a better start). This is the equivalent of a 200 m
length of cable. I fear trouble with this: Am I not getting unwanted
inductivities if I use a coil of that size?
The coax is a transmission line. Yes you have inductance and capacitance,
but it does not make that much sense to talk about that anymore,
the impedance is the right thing to talk about.
Your output will not be as sharp as your input due to dispersion,
but that can be easily recovered using some buffer gate.
Please make sure that your coil is reasonably temperature stabilized
and, if it's cheaper cable, also humidity stabilized, as both parameters
change your delay. (putting it into an isolated box should be good
enough for this kind of measurment).
Attila Kinali
--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson
If the goal is to create two signals consistently spaced near 70us apart why not use a good, fast 8-bit serial-in, parallel out shift register, clocked cleanly at 100kHz? Using the outputs from stages 1 and 8 would result in a 70us delay between signals. The data in would be fed 100KHz divided by 10 (or 16, or anything greater than 8) at whatever duty cycle is available. This allows the GP22 to see the combined instabilities of the clock and the shift register, which could be down in the nanosecond range, possibly less since the shift register delays would inherently cancel all but their differences.
Bob LaJeunesse
Sent: Tuesday, November 24, 2015 at 9:04 AM
From: "Thomas Allgeier" th.allgeier@gmail.com
To: time-nuts@febo.com
Subject: [time-nuts] ACAM GP22 Chip
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for “work” purposes – I work for a company active in the weighing and force measurement world.
...
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
Hi
If you head down to your local big box store, they will happily sell you a thousand foot spool
of RG-6 coax for next to nothing. If their prices are still to high, the auction sites will sell it for
even less. It has a 75 ohm impedance and a bandwidth of several GHz. The rather convent
formula of RT = 0.35 / BW then comes in. A 3.5 GHz cable will limit you to a 100 ps rise time.
In all likelihood, you will be unable to generate a signal with this fast a rise time.
You also will have some loss effects in the cable that are frequency dependent. The calculation above
assumes you have done a few tricks to take care of this. If not, to get a 10 ns rise time, you need to maintain
a 35 MHz bandwidth. That works fine if you have a buffer every 500 feet. No tricks, just a CMOS buffer
chip.
As noted by others, it is coax. You need to drive it and terminate it with 75 ohms. At 35 MHz, a cheap
75 ohm resistor will do the trick just fine. At 3.5 GHz you may need to get a bit more careful.
So is the 500’ limit an issue? I’d suggest that it’s not. Consider chopping up the spool in a binary series of
400, 200,100,50,25,12.5, 6.5, 3.25 feet. You now have a set of buffered lines that can be arranged to give you
a nice set of 256 time steps. Yes, the delay of the buffers will get in the way a bit. The actual line lengths will
be a bit shorter as the lengths drop.
So how much delay do you get from a 400’ line? Velocity factor comes in here. Best guess is that
your foam RG-6 has a 0.78 velocity factor. The "speed of light” in the coax is 78% of the speed of light
in vacuum. Your 400 foot coax has about a 520 ns delay. Your stack comes out just a bit over 1 us.
Bob
On Nov 24, 2015, at 9:04 AM, Thomas Allgeier th.allgeier@gmail.com wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for “work” purposes – I work for a company active in the weighing and force measurement world.
I should say from the start that I am new to time and frequency measurements and not even an electronics engineer – but then I have been exposed to high-precision electronics for the last 25 years hence have picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s experiment with the coaxial delay line from the HP5370b – but as my interest is in “measuring range 2” of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start). This is the equivalent of a 200 m length of cable. I fear trouble with this: Am I not getting unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far with 1 µs and are there other (simple/reliable) ways to achieve this kind of delay? I have tried it with a shorter piece of cable (around 2 ns which is measured in “range 1”), there I seem to get consistency virtually to within 100 ps. But I need to know if the device sticks to this level of performance when the periods are much longer, and thus measured in “range 2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Dear Paul,
Thanks for the reply and thanks to all other contributors. Seems I have
subscribed to the right list!
As it happens this is a sideline project. So I have the luxury to ask for
advice and even consult books before advancing with care.
I will probably try and go down the coax route, starting with a shorter
length first, and reading up a bit.
If any useable results can be obtained I will post them for future
reference.
Best regards,
Thomas.
----- Original Message -----
From: "paul swed" paulswedb@gmail.com
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, November 24, 2015 3:17 PM
Subject: Re: [time-nuts] ACAM GP22 Chip
Thomas
Welcome to the group. I am sure others will comment.
Many of us have a very wide range of experience and expertise so you
should
feel comfortable with any question.
To the coax delay question. You are not pushing the limits.
But its important to understand the impacts of such long lines.
They need to be driven and terminated and the rise time will suffer from
the line capacitance. Essentially a fast rise time will become a slow
risetime on teh other end. There are lumped lc network delay lines. I have
experimented with them. They have the same effect. But you can cascade
them
and use an inverter or buffer between each one.Each inverter also adds
delay. This helps the rise time issue. But the buffers add jitter and each
also adds delay thats temperature sensitive.
For cascaded delays of very short duration I have actually used 74LS244s
74HC244 line drivers cascaded and they work really well but only good for
each drivers delay.
Others will have better answers.
Regards
Paul
WB8TSL
On Tue, Nov 24, 2015 at 9:04 AM, Thomas Allgeier th.allgeier@gmail.com
wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at
for “work” purposes – I work for a company active in the weighing and
force
measurement world.
I should say from the start that I am new to time and frequency
measurements and not even an electronics engineer – but then I have been
exposed to high-precision electronics for the last 25 years hence have
picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of
around
13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b – but as my
interest is in “measuring range 2” of the GP22 I need a delay of 500 ns
or
more (actually 1 µs sounds a better start). This is the equivalent of a
200
m length of cable. I fear trouble with this: Am I not getting unwanted
inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too
far
with 1 µs and are there other (simple/reliable) ways to achieve this kind
of delay? I have tried it with a shorter piece of cable (around 2 ns
which
is measured in “range 1”), there I seem to get consistency virtually to
within 100 ps. But I need to know if the device sticks to this level of
performance when the periods are much longer, and thus measured in “range
2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Moin Attila,
You guessed my origin right from my name.
It is not that I don't trust ACAM (we are one of their customers) but from
the source I am measuring I get a lot of what you guys probably call jitter.
So I want to be sure the jitter comes from the oscillator I am measuring and
not from the GP22.
Absolute accuracy is not the problem here, unless the 32768 Hz clock on the
eval board actually drifts. In other words if we are 10 ns out on the period
measurement nobody cares if it remains the same 10 ns all the time. As in
many applications in the weighing industry what really matters is the change
in weight, not how heavy the mass is in absolute terms.
It sounds like the coil must be wound what we used to call "bifilar" in my
school days, i.e. self-cancelling the inductivity. That should be doable,
and on the other stuff I will read up a bit first.
Thanks,
Thomas.
----- Original Message -----
From: "Attila Kinali" attila@kinali.ch
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, November 24, 2015 4:56 PM
Subject: Re: [time-nuts] ACAM GP22 Chip
Moin,
On Tue, 24 Nov 2015 14:04:32 -0000
"Thomas Allgeier" th.allgeier@gmail.com wrote:
We want to use this chip to measure the period of a square wave, of
around 13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
That's some modest requirement and should be doable with the GPS22
quite easily (or any other TDC for that matter). BTW: when specing
something like this, please make sure to mention whether 1ns is
1sigma, 3sigma or worst case/peak-to-peak. These 3 are quite different
requirements.
May I ask why you want to verify the specs of the GP22?
The specs say that it does something between 39ps and 70ps (1sigma),
which is probably way better than what you need. And as Acam is
a german company, I expect the datasheet to be accurate.
BTW: 1ns over 70us is approximately 14ppm. The GP22 uses the attached
crystal for absolute calibration. Please be aware that 14ppm will
require at least a TCXO to reach that level over the whole temperature
range, and depending on what TCXO you use, you might need to calibrate
the TCXO post-production and again after a couple of years of use.
Even if you don't need calibrate, I would add a TCXO frequency measurement
to the production test.
BTW2: the "we have a x ppm TCXO" value is usually misleading,
as that's the best-case, pre-soldering, pre-aging, pre-anything value.
The end-value can be 3 times as large... easily.
(unless you happen to choose one of the more honest manufacturers,
for example, like Abracon)
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b
For those wondering: "John A" is John Ackermann and the experiment
in question is documented at http://www.febo.com/pages/hp5370b/
– but as my interest
is in “measuring range 2” of the GP22 I need a delay of 500 ns or more
(actually 1 µs sounds a better start). This is the equivalent of a 200 m
length of cable. I fear trouble with this: Am I not getting unwanted
inductivities if I use a coil of that size?
The coax is a transmission line. Yes you have inductance and capacitance,
but it does not make that much sense to talk about that anymore,
the impedance is the right thing to talk about.
Your output will not be as sharp as your input due to dispersion,
but that can be easily recovered using some buffer gate.
Please make sure that your coil is reasonably temperature stabilized
and, if it's cheaper cable, also humidity stabilized, as both parameters
change your delay. (putting it into an isolated box should be good
enough for this kind of measurment).
Attila Kinali
--
It is upon moral qualities that a society is ultimately founded. All
the prosperity and technological sophistication in the world is of no
use without that foundation.
-- Miss Matheson, The Diamond Age, Neil Stephenson
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hello Bob,
Thanks for the suggestions - these sound a bit beyond my current level of
skills and kit. But I do have people to call on in the office who may be
able to rig something like that up for me should the coax route fail.
Best regards,
Thomas.
----- Original Message -----
From: "Robert LaJeunesse" lajeunesse@mail.com
To: time-nuts@febo.com
Sent: Tuesday, November 24, 2015 6:25 PM
Subject: Re: [time-nuts] ACAM GP22 Chip
If the goal is to create two signals consistently spaced near 70us apart
why not use a good, fast 8-bit serial-in, parallel out shift register,
clocked cleanly at 100kHz? Using the outputs from stages 1 and 8 would
result in a 70us delay between signals. The data in would be fed 100KHz
divided by 10 (or 16, or anything greater than 8) at whatever duty cycle
is available. This allows the GP22 to see the combined instabilities of
the clock and the shift register, which could be down in the nanosecond
range, possibly less since the shift register delays would inherently
cancel all but their differences.
Bob LaJeunesse
Sent: Tuesday, November 24, 2015 at 9:04 AM
From: "Thomas Allgeier" th.allgeier@gmail.com
To: time-nuts@febo.com
Subject: [time-nuts] ACAM GP22 Chip
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at
for “work” purposes – I work for a company active in the weighing and
force measurement world.
...
We want to use this chip to measure the period of a square wave, of
around 13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hello Bob,
That kind of approach is what I had in mind and as others have commented if
done carefully (which for me means a bit at a time) should get me there.
The buffers may present a bit of a challenge to a mechanical engineer
(unless they are the kind that can be salvaged from railway waggons) but the
AoE book as suggested by Hal should set me straight.
Thanks again,
Thomas.
----- Original Message -----
From: "Bob Camp" kb8tq@n1k.org
To: "Discussion of precise time and frequency measurement"
time-nuts@febo.com
Sent: Tuesday, November 24, 2015 11:24 PM
Subject: Re: [time-nuts] ACAM GP22 Chip
Hi
If you head down to your local big box store, they will happily sell you a
thousand foot spool
of RG-6 coax for next to nothing. If their prices are still to high, the
auction sites will sell it for
even less. It has a 75 ohm impedance and a bandwidth of several GHz. The
rather convent
formula of RT = 0.35 / BW then comes in. A 3.5 GHz cable will limit you to
a 100 ps rise time.
In all likelihood, you will be unable to generate a signal with this fast
a rise time.
You also will have some loss effects in the cable that are frequency
dependent. The calculation above
assumes you have done a few tricks to take care of this. If not, to get a
10 ns rise time, you need to maintain
a 35 MHz bandwidth. That works fine if you have a buffer every 500 feet.
No tricks, just a CMOS buffer
chip.
As noted by others, it is coax. You need to drive it and terminate it
with 75 ohms. At 35 MHz, a cheap
75 ohm resistor will do the trick just fine. At 3.5 GHz you may need to
get a bit more careful.
So is the 500’ limit an issue? I’d suggest that it’s not. Consider
chopping up the spool in a binary series of
400, 200,100,50,25,12.5, 6.5, 3.25 feet. You now have a set of buffered
lines that can be arranged to give you
a nice set of 256 time steps. Yes, the delay of the buffers will get in
the way a bit. The actual line lengths will
be a bit shorter as the lengths drop.
So how much delay do you get from a 400’ line? Velocity factor comes in
here. Best guess is that
your foam RG-6 has a 0.78 velocity factor. The "speed of light” in the
coax is 78% of the speed of light
in vacuum. Your 400 foot coax has about a 520 ns delay. Your stack comes
out just a bit over 1 us.
Bob
On Nov 24, 2015, at 9:04 AM, Thomas Allgeier th.allgeier@gmail.com
wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at
for “work” purposes – I work for a company active in the weighing and
force measurement world.
I should say from the start that I am new to time and frequency
measurements and not even an electronics engineer – but then I have been
exposed to high-precision electronics for the last 25 years hence have
picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of
around 13 kHz i.e. in the 70 µs range. As the application is potentially
high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b – but as my
interest is in “measuring range 2” of the GP22 I need a delay of 500 ns
or more (actually 1 µs sounds a better start). This is the equivalent of
a 200 m length of cable. I fear trouble with this: Am I not getting
unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too
far with 1 µs and are there other (simple/reliable) ways to achieve this
kind of delay? I have tried it with a shorter piece of cable (around 2 ns
which is measured in “range 1”), there I seem to get consistency
virtually to within 100 ps. But I need to know if the device sticks to
this level of performance when the periods are much longer, and thus
measured in “range 2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
I don't want to muddy the water too much, but I'd second going
all-digital, along the lines of what Bob L. recommended earlier. It
seems like you just need to make some specific delay times, so various
logic counting or shifting circuits should be just fine - and
well-defined and understood.
If you go with coaxial delays, the loss and dispersion may cause much
analog grief to determine how much compensation and amplification is
needed, and the thresholds for comparison for squaring it up. This is
especially aggravated with long lines.
In the hundreds of meters range, it may be better to go with optical
fiber instead. The bandwidth can be huge, and losses will be tiny, even
into km lengths, and all kinds of E/O and O/E conversion parts are
available for the ends. There can be some timing jitter due to the noise
of the conversions, and AGC issues, but probably less than the noise
associated with adapting long coax cables to this task.
Another option may be acoustic glass delay lines. I don't know if any
are made nowadays, but they were common in TVs and VCRs, with time delay
somewhere in the horizontal line duration range, and very compact. It
could be that there are many more kinds for various applications (or
maybe all obsolete - replaced by digital).
Ed
Hi
To me the easy way to do the buffers:
!) Grab some buffers or inverters in a reasonable package. The 74ACT14
in a PDIP-14 package seems to sell for about 50 cents.
-
Power the gate off of 5V, it’s TTL input so it will trigger at 1.4 V
-
Hook the inputs of 5 inverters to the output of the 6th. The input to
the gizmo will be the input to this 6th inverter. -
Terminate the input to the 6th inverter to ground with 75 ohms.
-
Run a (75 x 5 = 375 ohm) resistor from the output of each gate to a common
point. That point will drive your coax.
Other than connectors, that’s it. You will have roughly 2.5V logic on the “far end”
of the coax. that will trigger the TTL inputs nicely. Each buffer has a parts cost of
maybe $1. Each buffer will add about 14 ns to the delay of the coax. If you buffer
both the input and out of the line, that would give you 28 ns.
I realize that may not make much sense at this point. It is a recipe you can point
somebody at and they can say “yes, I can do that”. It’s also a way to give you
a rough cost for the parts.
Bob
On Nov 25, 2015, at 9:27 AM, Thomas Allgeier th.allgeier@gmail.com wrote:
Hello Bob,
That kind of approach is what I had in mind and as others have commented if done carefully (which for me means a bit at a time) should get me there.
The buffers may present a bit of a challenge to a mechanical engineer (unless they are the kind that can be salvaged from railway waggons) but the AoE book as suggested by Hal should set me straight.
Thanks again,
Thomas.
----- Original Message ----- From: "Bob Camp" kb8tq@n1k.org
To: "Discussion of precise time and frequency measurement" time-nuts@febo.com
Sent: Tuesday, November 24, 2015 11:24 PM
Subject: Re: [time-nuts] ACAM GP22 Chip
Hi
If you head down to your local big box store, they will happily sell you a thousand foot spool
of RG-6 coax for next to nothing. If their prices are still to high, the auction sites will sell it for
even less. It has a 75 ohm impedance and a bandwidth of several GHz. The rather convent
formula of RT = 0.35 / BW then comes in. A 3.5 GHz cable will limit you to a 100 ps rise time.
In all likelihood, you will be unable to generate a signal with this fast a rise time.
You also will have some loss effects in the cable that are frequency dependent. The calculation above
assumes you have done a few tricks to take care of this. If not, to get a 10 ns rise time, you need to maintain
a 35 MHz bandwidth. That works fine if you have a buffer every 500 feet. No tricks, just a CMOS buffer
chip.
As noted by others, it is coax. You need to drive it and terminate it with 75 ohms. At 35 MHz, a cheap
75 ohm resistor will do the trick just fine. At 3.5 GHz you may need to get a bit more careful.
So is the 500’ limit an issue? I’d suggest that it’s not. Consider chopping up the spool in a binary series of
400, 200,100,50,25,12.5, 6.5, 3.25 feet. You now have a set of buffered lines that can be arranged to give you
a nice set of 256 time steps. Yes, the delay of the buffers will get in the way a bit. The actual line lengths will
be a bit shorter as the lengths drop.
So how much delay do you get from a 400’ line? Velocity factor comes in here. Best guess is that
your foam RG-6 has a 0.78 velocity factor. The "speed of light” in the coax is 78% of the speed of light
in vacuum. Your 400 foot coax has about a 520 ns delay. Your stack comes out just a bit over 1 us.
Bob
On Nov 24, 2015, at 9:04 AM, Thomas Allgeier th.allgeier@gmail.com wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for “work” purposes – I work for a company active in the weighing and force measurement world.
I should say from the start that I am new to time and frequency measurements and not even an electronics engineer – but then I have been exposed to high-precision electronics for the last 25 years hence have picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John A’s experiment with the coaxial delay line from the HP5370b – but as my interest is in “measuring range 2” of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start). This is the equivalent of a 200 m length of cable. I fear trouble with this: Am I not getting unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far with 1 µs and are there other (simple/reliable) ways to achieve this kind of delay? I have tried it with a shorter piece of cable (around 2 ns which is measured in “range 1”), there I seem to get consistency virtually to within 100 ps. But I need to know if the device sticks to this level of performance when the periods are much longer, and thus measured in “range 2”.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
N'abend
On Wed, 25 Nov 2015 14:19:37 -0000
"Thomas Allgeier" th.allgeier@gmail.com wrote:
You guessed my origin right from my name.
It is not that I don't trust ACAM (we are one of their customers) but from
the source I am measuring I get a lot of what you guys probably call jitter.
So I want to be sure the jitter comes from the oscillator I am measuring and
not from the GP22.
Ah, so it's about verification of the results you already got.
Then I would rent some TIC for a day or two from one of the many shops
that rent out measurement instruments. That's probably cheaper (especially
if you include your work time) and you don't have to guestimate the noise
that your reference circuit makes, but get the noise floor values from
the instrument manual.
Absolute accuracy is not the problem here, unless the 32768 Hz clock on the
eval board actually drifts. In other words if we are 10 ns out on the period
measurement nobody cares if it remains the same 10 ns all the time. As in
many applications in the weighing industry what really matters is the change
in weight, not how heavy the mass is in absolute terms.
I don't really see how a 32kHz crystal is used in a scale, but there are
many way to do things :-)
It sounds like the coil must be wound what we used to call "bifilar" in my
school days, i.e. self-cancelling the inductivity. That should be doable,
and on the other stuff I will read up a bit first.
Again, please do not think of a coax in terms of wires. It's a transmission
line. 99% of the wave energy is between the inner and the outer conductor.
There is (almost) no magnetic field that you will be able to cancel by
building a bifilar coax spool. How your signal looks like depends on the
intrinsic properties of the cable. There is very little dependence of the
electrical parameters on the way how you arange the cable, whether you use
a straight line or roll it up.
On Wed, 25 Nov 2015 14:23:40 -0000
"Thomas Allgeier" th.allgeier@gmail.com wrote:
I will report back any outcome that looks plausible / presentable.
Please also report back when you have negative results. It's almost
always more instructive to see what went wrong and figure out why.
Attila Kinali
--
Reading can seriously damage your ignorance.
-- unknown
On Wed, 25 Nov 2015 14:19:37 -0000
"Thomas Allgeier" th.allgeier@gmail.com wrote:
It is not that I don't trust ACAM (we are one of their customers) but from
the source I am measuring I get a lot of what you guys probably call jitter.
So I want to be sure the jitter comes from the oscillator I am measuring and
not from the GP22.
Absolute accuracy is not the problem here, unless the 32768 Hz clock on the
eval board actually drifts.
Oh.. there is one thing i forgot to mention: 32kHz oscillators are usually
optimized for low power. Modern 32kHz oscillators can get below 1uA current
consumption. This optimization for power has of course consequences, one
of them is that these oscillators are fairly noisy. People using 32kHz
oscillators usually don't care as they use these to count seconds anyways,
where the instability is dominated by the temperature coefficient of the
crystall and noise mostly averages out. I wouldn't expect the noise to
be in the ns range, but i wouldn't surprised if it was a few 10ps.
I have never done any measurements though, and I don't think i've
ever seen any jitter measurements for 32kHz oscillators, so take
this value as rough guestimate.
Attila Kinali
--
Reading can seriously damage your ignorance.
-- unknown
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b
For those wondering: "John A" is John Ackermann and the experiment
in question is documented at http://www.febo.com/pages/hp5370b/
Maybe I misunderstand, but I would not suggest testing a time interval counter by using a fixed ns delay -- that's almost never how the real world works and those tests tend to produce bogus ADEV plots that have -1 slope forever (a clue that something's wrong with the test).
A selection of fixed delays is slightly better. But best, and much easier, is to use uncorrelated A, B, and LO (ext ref) signals. A fixed delay may land on a sweet spot or honey bucket. Linear sweeping the range covers all spots, and gives you best case / worst case / rms statistics as a bonus. In other words, what you want is a set of random (but known, or knowable) delays; not a set of hardcoded delays.
of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start).
Are you sure you want a hardcoded delay of N ns or N us? Or is a variable or even varying delay sufficient?
What I use in cases like this is two stable oscillators that slowly drift apart (i.e., close, but not the same frequency). For example, if they differ in frequency by 1e-12 your signals drift 1 ps/s. Or if they differ by 1e-10 your signals drift by 1 ns / 10 s. You get an uncorrelated, very low-noise, linear phase sweep "for free".
This sort of slow varying phase relationship is ideal when making counter tests; much better than a fixed delay. You can use a laboratory counter to monitor their exact phase difference in parallel with your DUT. That is, you then compare TIC "truth" against what your DUT reports.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range.
As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
I may have missed it in the thread -- but how quickly do you need your measurements? Is one measurement every 1 or 10 or 100 seconds ok? (in which case an ACAM chip is total overkill). Or is this some sort of sub-second real-time application that require both modest resolution (1 ns / 70 us = 15 ppm, easy) and fast response (hard)?
/tvb
Hi Thomas,
Another option for generating a range of delays would be to use a
stable oscillator like an ovenised one. This is actually similar to
what the GP22 does for measuring on mode2.
If you gate the output of the oscillator you can get a start and stops
at 1us or 70us or whatever. It's useful because it allows you to test
the whole measurement system rather than just the TDC noise which
should be way below 1ns peak-peak.
Angus.
On Tue, 24 Nov 2015 14:04:32 -0000, you wrote:
Hello,
I have an ACAM GP22 TDC chip and evaluation board which I am looking at for work purposes I work for a company active in the weighing and force measurement world.
I should say from the start that I am new to time and frequency measurements and not even an electronics engineer but then I have been exposed to high-precision electronics for the last 25 years hence have picked up some dangerous degree of half-knowledge.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range. As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
In order to evaluate the chip I was planning to replicate John As experiment with the coaxial delay line from the HP5370b but as my interest is in measuring range 2 of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start). This is the equivalent of a 200 m length of cable. I fear trouble with this: Am I not getting unwanted inductivities if I use a coil of that size?
So, to come to the point: Am I pushing the concept of a coax delay too far with 1 µs and are there other (simple/reliable) ways to achieve this kind of delay? I have tried it with a shorter piece of cable (around 2 ns which is measured in range 1), there I seem to get consistency virtually to within 100 ps. But I need to know if the device sticks to this level of performance when the periods are much longer, and thus measured in range 2.
Thanks and best regards,
Thomas.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
I wouldn't expect the noise to
be in the ns range, but i wouldn't surprised if it was a few 10ps.
I have never done any measurements though, and I don't think i've
ever seen any jitter measurements for 32kHz oscillators, so take
this value as rough guestimate.
On the GP21 board I used it was more like 10ns+. I don't know if
that's typical, but it wasn't a even particularly cheapo crystal as
the original one was faulty and I had to replace it. The official acam
board might be better, but the data sheet specifically talks about the
32K osc having a lot of jitter.
One of the problems of the data sheet is that it is really tailored
for those using it as a flow converter - some of the text does not
even make sense otherwise. That was one of the reasons I was planning
to test it more as a TDC, but just getting back to doing that now.
Angus.
Hello Tom,
One of the problems here (embarrassingly) is that we have a lack of kit in
this particular respect. And I am not yet prepared to throw too much money
at it before I can judge the real potential.
(Our forte so far is high precision in terms of voltage and current, which
is what you traditionally need for strain gauge measurements, which is what
we have been traditionally doing. nV instead of ns or ps so to speak.) Hence
the low-tech approach, and having read the tests on the HP5370 I thought
"hey, I can do that".
Anyway we need the results a lot quicker than 1 Hz, ideally at least every
10 ms or so. The reason is that in weighing much is achieved by filtering
and if you want a reasonable response time after the filter then you have to
feed in the raw stuff quite quickly. The GP22 seems to be pretty much what
is needed, and we happen to buy from ACAM already.
I played with it last night and I think I should shortly be in a positions
to share "early" findings.
Best regards,
Thomas.
-----Original Message-----
From: Tom Van Baak
Sent: Thursday, November 26, 2015 12:02 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] ACAM GP22 Chip
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b
For those wondering: "John A" is John Ackermann and the experiment
in question is documented at http://www.febo.com/pages/hp5370b/
Maybe I misunderstand, but I would not suggest testing a time interval
counter by using a fixed ns delay -- that's almost never how the real world
works and those tests tend to produce bogus ADEV plots that have -1 slope
forever (a clue that something's wrong with the test).
A selection of fixed delays is slightly better. But best, and much easier,
is to use uncorrelated A, B, and LO (ext ref) signals. A fixed delay may
land on a sweet spot or honey bucket. Linear sweeping the range covers all
spots, and gives you best case / worst case / rms statistics as a bonus. In
other words, what you want is a set of random (but known, or knowable)
delays; not a set of hardcoded delays.
of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a
better start).
Are you sure you want a hardcoded delay of N ns or N us? Or is a variable or
even varying delay sufficient?
What I use in cases like this is two stable oscillators that slowly drift
apart (i.e., close, but not the same frequency). For example, if they differ
in frequency by 1e-12 your signals drift 1 ps/s. Or if they differ by 1e-10
your signals drift by 1 ns / 10 s. You get an uncorrelated, very low-noise,
linear phase sweep "for free".
This sort of slow varying phase relationship is ideal when making counter
tests; much better than a fixed delay. You can use a laboratory counter to
monitor their exact phase difference in parallel with your DUT. That is, you
then compare TIC "truth" against what your DUT reports.
We want to use this chip to measure the period of a square wave, of around
13 kHz i.e. in the 70 µs range.
As the application is potentially high-accuracy we need to know the period
to within 1 ns or better.
I may have missed it in the thread -- but how quickly do you need your
measurements? Is one measurement every 1 or 10 or 100 seconds ok? (in which
case an ACAM chip is total overkill). Or is this some sort of sub-second
real-time application that require both modest resolution (1 ns / 70 us = 15
ppm, easy) and fast response (hard)?
/tvb
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hello Angus,
That is roughly what I want to do, a plausibility check if what I see is
caused more by the GP22 or already present on the signal I am trying to
measure.
Very clearly there is no principal problem, the data I get from the GP22 is
pretty decent and with enough filtering already useable. I have the feeling
the GP22 does not add very much noise/jitter and all I want is get a handle
on how much in reality.
If we take this approach further we will have to invest in some additional
kit, but apart from that I am also rather taken by the capabilities of the
GP22 as a timer, and my own curiosity wants to find out how good a homebrew
counter/timer could be built from it.
As I said I will feed back anything I can establish in this respect.
Best regards,
Thomas.
-----Original Message-----
From: Angus
Sent: Thursday, November 26, 2015 1:54 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] ACAM GP22 Chip
I wouldn't expect the noise to
be in the ns range, but i wouldn't surprised if it was a few 10ps.
I have never done any measurements though, and I don't think i've
ever seen any jitter measurements for 32kHz oscillators, so take
this value as rough guestimate.
On the GP21 board I used it was more like 10ns+. I don't know if
that's typical, but it wasn't a even particularly cheapo crystal as
the original one was faulty and I had to replace it. The official acam
board might be better, but the data sheet specifically talks about the
32K osc having a lot of jitter.
One of the problems of the data sheet is that it is really tailored
for those using it as a flow converter - some of the text does not
even make sense otherwise. That was one of the reasons I was planning
to test it more as a TDC, but just getting back to doing that now.
Angus.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to
https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
Hi
The issue of “technology to back this up” is not a trivial point at all. There is a
lot of detailed knowledge in each of these areas. There is also a lot of gear
that is specific to this or that kind of work. The gear is fairly easy to take care
of and to estimate. The “how long will the learning curve be” part is often the
biggest issue.
Bob
On Nov 26, 2015, at 5:12 AM, Thomas Allgeier th.allgeier@gmail.com wrote:
Hello Tom,
One of the problems here (embarrassingly) is that we have a lack of kit in this particular respect. And I am not yet prepared to throw too much money at it before I can judge the real potential.
(Our forte so far is high precision in terms of voltage and current, which is what you traditionally need for strain gauge measurements, which is what we have been traditionally doing. nV instead of ns or ps so to speak.) Hence the low-tech approach, and having read the tests on the HP5370 I thought "hey, I can do that".
Anyway we need the results a lot quicker than 1 Hz, ideally at least every 10 ms or so. The reason is that in weighing much is achieved by filtering and if you want a reasonable response time after the filter then you have to feed in the raw stuff quite quickly. The GP22 seems to be pretty much what is needed, and we happen to buy from ACAM already.
I played with it last night and I think I should shortly be in a positions to share "early" findings.
Best regards,
Thomas.
-----Original Message----- From: Tom Van Baak
Sent: Thursday, November 26, 2015 12:02 AM
To: Discussion of precise time and frequency measurement
Subject: Re: [time-nuts] ACAM GP22 Chip
In order to evaluate the chip I was planning to replicate John A’s
experiment with the coaxial delay line from the HP5370b
For those wondering: "John A" is John Ackermann and the experiment
in question is documented at http://www.febo.com/pages/hp5370b/
Maybe I misunderstand, but I would not suggest testing a time interval counter by using a fixed ns delay -- that's almost never how the real world works and those tests tend to produce bogus ADEV plots that have -1 slope forever (a clue that something's wrong with the test).
A selection of fixed delays is slightly better. But best, and much easier, is to use uncorrelated A, B, and LO (ext ref) signals. A fixed delay may land on a sweet spot or honey bucket. Linear sweeping the range covers all spots, and gives you best case / worst case / rms statistics as a bonus. In other words, what you want is a set of random (but known, or knowable) delays; not a set of hardcoded delays.
of the GP22 I need a delay of 500 ns or more (actually 1 µs sounds a better start).
Are you sure you want a hardcoded delay of N ns or N us? Or is a variable or even varying delay sufficient?
What I use in cases like this is two stable oscillators that slowly drift apart (i.e., close, but not the same frequency). For example, if they differ in frequency by 1e-12 your signals drift 1 ps/s. Or if they differ by 1e-10 your signals drift by 1 ns / 10 s. You get an uncorrelated, very low-noise, linear phase sweep "for free".
This sort of slow varying phase relationship is ideal when making counter tests; much better than a fixed delay. You can use a laboratory counter to monitor their exact phase difference in parallel with your DUT. That is, you then compare TIC "truth" against what your DUT reports.
We want to use this chip to measure the period of a square wave, of around 13 kHz i.e. in the 70 µs range.
As the application is potentially high-accuracy we need to know the period to within 1 ns or better.
I may have missed it in the thread -- but how quickly do you need your measurements? Is one measurement every 1 or 10 or 100 seconds ok? (in which case an ACAM chip is total overkill). Or is this some sort of sub-second real-time application that require both modest resolution (1 ns / 70 us = 15 ppm, easy) and fast response (hard)?
/tvb
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.
time-nuts mailing list -- time-nuts@febo.com
To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts
and follow the instructions there.