| Pages:
1
2 |
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
Ok that gives an idea what your dealing with, i didnt expect the current bridge. Makes things a bit easier in some respects. 10 bit resolution is all
you need. The machine itself is effectively 10 bit. So going with a ADC higher than 10 bits wont give you more accuracy, it may at a push be needed,
not for the accuracy but to extend the range.
Your dealing with a narrow range here, 1mV separations. It seems to have a 300mV range unless I have misread it, i will read again more carefully
later. But just giving you a quick view from a read through.
My take at the moment is, the bridge while being run at 10 bits, is attenuated. Not that it matters much, but you are going to need a instrument grade
rail to rail opamp.
Dont expect these to be super cheap, while not major expensive, to me they seem alot of money for what they are.
But keep in mind this was designed for a pen chart, the detection hardware (opamp bit) would of been inside that. All the charts recorders I have
taken apart (4 only), didnt actually have great opamps in. Also what confuses me a little is the bridge regulation is by zener diode, in todays world
a zener is not exactly cutting edge voltage regulation.
TBH a zener is the quick n dirty goto these days. So this makes me a bit skeptical of the bridge specs, hence on the reasons even a crap scope on a
usb might be worth getting to look at the output.
While I read up on this manual, make sure you dont know anyone with a real oscilloscope, not a crappy usb one, but a real one. If you do happen to
know someone, 10 mins on that sucker will give you all the info needed to build the output stage bridge.
The actual circuit and electronics, on this is experienced hobby level to sort out. The hard bit is knowing what you got to build around.
We what we know so far
1) 10 bit is plenty
2) its run in 1mV resolution
3) its a current bridge
4) 300mV max adjustment
5) with the above I would expect 0-1V levels tops
6)......You will need an opamp
I will come back and explain the opamp later, as we are now 100% sure its a current bridge, so your going to read the voltage drop off a resistor.
You could do it with a higher bit ADC but it will cost more to build the power source that it would to do it this way. Trying to measure uV is fine if
you got zero ripple, in the real world even the bridge gives a ripple spec.
So rather than try and read uV, we boost the output with a opamp. I dont know yet how much, but likely 10X or 100X. This is actually controled by two
simple resistors connected to the opamp.
For clarity, we are talking 1% resistors. i would go smd as they tend to be more stable and accurate in my experience. The bigger size smd resistors
are not hard to solder with practice.
Go hunt down a chap with a scope, and I will read the manual again. Actually I am going to hunt down the service manual for a suitable pen recorder
 . That will tell me what kind of front end they use on it. Then kind of copy the
front end bit with modern bits.
But the good good news is..........1000% you can run that with matlab and serial data into a usb cabel, you can use GC software to calculate and draw
the the output, but obviously we are not talking adjusting the machine via software!!!
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
The USB scope is out, its 5mV resolution. Still cant find a schematic for a decent pen chart!! Dont expect great accuracy with this thing, the built
in errors are fairly large compared to what the software normally works with.
I am more leaning with a simple approach and something like matlab or maybe simple graphs in C#. Doing alot of calcs on area isnt going to give much
info. The accuracy just isnt there, if you go up to 12 bit bit ADC you will simply pic up the noise.
I might have the front of a pen chart in my spares box, I will try and find out what was used as the front end pick up. The more I read the less
expensive i would be inclined to make it. Sure we could capture everything coming out, but you wont get a peak, or likely to get a mass of needle like
spikes doing that.
Damping it down might actually give you better results. If your mapping the contours of a mountain, dosnt help you much if you measure Everest every
mm. Some times better to work in feet than inches.
|
|
|
SHADYCHASE54
Hazard to Others
Posts: 150
Registered: 16-12-2010
Location: CaNaDay!
Member Is Offline
Mood: No Mood
|
|
I just wondered if this might have what your looking for.
Attachment: phpw5wPtA (7.7MB)
This file has been downloaded 354 times
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
If its got a schematic in it then yes its exactly what I am after. The file you gave just shows me the front page.
See if you can get a decent sized png of the schematic, if its got more than, then you after the one that shows from the input. We just want to know
how they deal with the input and shape it.
Once we know that you can decide what you want to do. I would suggest you use Nitrogen, loads of reasons but mainly, helium to me shouldnt go into
party balloons, and shouldnt be used when you have a reasonable alternative.
Its likely in the final design, you will have a load of blank spaces on the board. I suggest you spend the $20 and get the board made in china.
I will leave a section in the design but i suggest you dont actually populate it with parts unless you have too. What I am thinking is.....
It costs nothing to design in a damping system, with no oscilloscope to hand, and no way of me knowing what signal you get. Prudence makes me want to
design in a damping system.
My gut says you wont need it, but if you do then you can add the caps and resistors and solder in a link wire. Then on the arduino/pic side of things,
I will add in the software that looks for the link.
In a nutshell in psuedo code it goes a bit like this...
Is pin X high or low
if pin high then sample Y times in W ms and divide by Z
If pin X Low then just read and output
So software wise and hardware wise its no big deal to cover your arse, will save you doing it all twice.
These jobs are so much harder to do without the bloody machine in front of you. I am sure I could have done this over a weekend had you lived near me!
Probably for about £10
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
Ok worked out the file!!! Didnt twig i needed to save as pdf sorry.
First glance.... isnt organized nicely to go straight to the information. They seem to of been coy about some the modules. But dosnt matter, is really
old tech so most of it you wont need.
Surprised how much they jack the voltage up!! 400V! No idea why this is done, but then again i have no intention of looking up the datasheets for most
the components.
I need some time to work out how they designed it, I know roughly what they done, but I need to check what the front end expects from the GC.
From the quick glance.....
GC passes current to the front mods, these simply use a resistor to give a voltage drop. One reason they might be using insane voltage is too do away
with complex amplification, no idea how old this is, but probably circa 7014 opamps.
That would mean nastiness, alot of it. So we might not have to much signal conditioning at all, simply drop the current over a high precision resistor
(ok not that high...1%).
I would still use a reasonable rail to rail instrument opamp to boost the signal. From that point on its almost single chip and software.
Going by gut feeling, I would take a couple long routes. Normally with a pic I tend to use the internal OSC, With this for what they cost I would go
crystal OSC. That gives you a pretty accurate time base.
Next I would normally the internal voltage ref on the chip, if you use a good psu this is spot on. But in this case for a couple £, i would use a
precision voltage ref for the ADC on the chip to read against.
The rest consist's of stuff you find in just about every micro circuit. ISCP setup for programing, max 232 for serial data, although with the usb
cables these days you can use the serial output from the chip directly. But i still use the max chip.
Not much more to it..........
So recap
Turn the current to a voltage so the ADC can read it, this is done via a resistor. Ohms law connects resistance, current and Voltage. So if you got a
Voltage you drop that over a known resistance and read the Voltage. This is directly proportional to the current.
In our case its a similar thing. The micro wont read uV directly, so you boost this with a opamp. every uV will be boosted to 1mV or maybe 100mV per
uV (depends what the max output is). If the range is small enough then you will get high resolution with 10 bit. 1024 steps to be precise.
The rest is simple software, read the ADC 64 times in ~10ms average it, take that number and divide by 1024. This gives you the mV, output that over
the serial line, plot it on a graph or other software.... Job done
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
from the GC manual:
The output from the GC appears to be the raw bridge signal which is a voltage with a source impedance of between 32 and 64ohms depending on source
impedance of the current drive to the bridge.
It recommends a chart recorder with a span of 1mV for maximum sensitivity and 1s response. and max chart rate of 2in/min.
The GC has a noise floor of 10uV rms max and a drift of 40uV/hour max.
If you assume the max sensitivity is achieved with the attenuator set at zero then you can decreases the signal amplitude but not increase it.
So how many uV should a bit/step represent and what should maximum voltage be ?.
Well ideally a bit should be no bigger than the noise floor and it could be smaller with filtering. Lets say no more than 10uV/ bit with at least a
maximum range voltage of 1mv.. So an 8bit ADC with 3.9uv/bit would provide 1mV max. There is little point in increasing the resolution (reducing the
uv/bit) unless the filtering reduces the noise and then you still have the drift. Its probably more useful to increase the dynamic range so you
don’t have to repeat a measurement with the attenuator set lower to see the large peaks. So a 10bit DAC would be set 4mV range. For nitrogen or
argon you might want +/- 2mV. to decode the positive and negative peaks. Idealy the DAC voltage reference would be referenced to bridge voltage
reference but that would requre a new connection.
A HX711 is a single two channel bridge amplifier and 24 bit DAC with gains of +/-20mV or +/-40 and +/-80mV with 10sps or 80 sps . You can get one
ready to go on small board for 99p from Ebay. Note it has only has 18 noiseless bits. It also has built in power line frequecy filters for 50 and
60Hz. Its intended use is for a bridge strain gauge type of electronic scales. It would provide a large dynamic range.
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
WHAT??? I would edit but its been up a while, ignore the bits between the lines. I see your on about the amp side instead of the opamp use the DAC
into the micro.
I edited the post but seems the first bit was missed.
----------------------------------------------------------------------------IGNORE HERE
How does it help? How are you then going to get the display on the pc? EVEN with that chip you have to connect a micro or something and then spit out
raw USB or better still serial at the rate the GC works.
A opamp and 10bit is doing the same job as that chip except it can also get the data displayed.
I cant see the advantage, the reason for using a opamp is because its exactly how the detector in the pen chart works. Boost the signal and filter,
mains noise is a simple 100nf ceramic cap across the rails. What you need to filter is the high frequency noise.
This is why i said a band pass and boost the signal so you just need to filter mains hum.
But if you can get that chip to display the signal.....
TO HERE
-----------------------------------------------------------------------------
You could connect that to a arduino, but 24 bit is overkill. I except it would give you a massive range, but I think thats asking for trouble.
I cant answer this yet, Some the GC manual is contradictory, it talks about the bridge in terms of 1024 steps, this would be 10 bit were it digital.
On 1mV range I think the penchart is 8 bit!! But dont quote that I havnt had chance to go through the schematics.
Mains hum is easy to take care of, a 10uF electrolytic cap and a 100nf ceramic cap job done. But these days its the high frequency stuff thats an
issue.
Dont add any more ideas! Thats another data sheet i got to read! 99p is a tempting price though. But is it buying cheap problems?
Couldnt use it with a PIC 18f but it would leave us the arduino option. My gut feeling is its cheap but too sensitive. Worth reading the sheet though.
[Edited on 18-2-2018 by NEMO-Chemistry]
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Nemo: Why do you care that bridge amp and DAC I suggested is too sensitive or has too many bits. It not like its too expensive. Does it have
sufficient sensitivity, at least the number bits to represent the voltage range required with sufficiently low nose and sufficient accuracy is what
you need consider. Actually the accuracy is poor but it’s a ready to go solution for 99p. Any extra bits are bonus but they don’t have to be used
but why not.
The number steps on the bridge current attenuator has very little to do with number of bits required on the DAC if your only digitising the voltage
range of a paper chart recorder.
I am guessing but I would expect a modern GC to digitise most of not all of the dynamic range of the bridge sensor (about 18 bits) and store the
result. Then only one run would required.
It probably better to pay more attention to your frontal lobes than your elementary canal or limbic system LOL and if you don’t want to read my
suggestions don’t.
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
LOL, I left the the bit I originally edited up, mainly because it had been up several hours before I spotted it.
Going to your question on bits, why do I care about it being 24 bits. I care because the guy isnt going to be able to trouble shoot much, 24 bits is
above what is needed. Surely you can see the problems associated with this
It isnt going to help him with two opposites working on one problem, you go ahead and help him i will stay out of it.
I would take some time and think it through first, you have made some fundamental errors and assumptions, but cary on. I am interested to see how you
get this onto a pc and filtered out to a signal you can use.
It would have helped alot more had you got its function right. It isnt a DAC!! Its the complete opposite to a DAC, looking at the data sheet its a
ADC.
They are complete opposites, but i guess you think this dosnt matter either?
I dont care if you call an arm a leg, but if your going to operate on someone, its probably better not to tell them its a leg amputation and then cut
the arm off.
[Edited on 19-2-2018 by NEMO-Chemistry]
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Wow I found a HX711 for 79 cents:
https://www.ebay.com/itm/1pc-Weighing-Pressure-Sensor-AD-Mod...
Perhaps a better chip is coming and they are trying to clear the old stock out.
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
I look forward to seeing what you come up with.
The code side dosnt look too bad.
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Wow^2 here is an other bridge amp and ADC the CJMCU-1232 (ADS1232) 24bit about £3.50 from ebay but its higher accuracy to 0.1%, about 20 usable bits
and down to 17nV rms noise Its probably a knock off of the Texas Instruments ADS1232 that has an even better spec.
https://www.ebay.co.uk/itm/CJMCU-1232-ADS1232-24Bit-Low-Nois...
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
LOL off you go then, should take you 5 mins to knock up a circuit and connect that to a micro .
You might want to read the pen recorder manual first, and re read the GC one..... just in case
I take it your assuming two things
1 it outputs a voltage
2 the voltage output is <100mV?
Looking at your other post i can see why you think this, i can also see why you might think from the manual. But i suggest before you go jumping on
this and designing something, check the pen manual and read the specs again.
Out of interest what is your take on what is output on the back of the GC? I think you focused on the 1mV range without looking at the pen manual.
[Edited on 20-2-2018 by NEMO-Chemistry]
[Edited on 20-2-2018 by NEMO-Chemistry]
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Quote: Originally posted by NEMO-Chemistry  | LOL off you go then, should take you 5 mins to knock up a circuit and connect that to a micro .
You might want to read the pen recorder manual first, and re read the GC one..... just in case
I take it your assuming two things
1 it outputs a voltage
2 the voltage output is <100mV?
Looking at your other post i can see why you think this, i can also see why you might think from the manual. But i suggest before you go jumping on
this and designing something, check the pen manual and read the specs again.
Out of interest what is your take on what is output on the back of the GC? I think you focused on the 1mV range without looking at the pen manual.
[Edited on 20-2-2018 by NEMO-Chemistry]
[Edited on 20-2-2018 by NEMO-Chemistry] |
Though my brain may be half dead the GC manual is far from rocket science so I am confident I can read it and comprehend it. I have no small
experience in this area. Since you asked I will try again to explain it.
The manual unambiguously states at least a sensitivity of 1mV span (0 to 1mV) is required for maximum sensitivity of the GC. That’s a different
ional voltage between the + and – terminals. Which I have already stated in one of my posts and apparently you already know??? (rhetorical question
marks suggesting complexion). For signals larger than that, the bridge current attenuator can be used to reduce the signal to fit in the that 0 to 1mV
span though a much better option is to use more bits and digitise the larger signals with the same resolution required of the 0 to 1mV span.
The manual does not give the largest differential signal on the output terminals as its not relevant due to the attenuator. However it is possible
to guestimate the largest signal from knowledge of how the resistance of tungsten changes with temperature and the temperatures given in the manual.
I guestimate it would be no more than 10% which would produce a change in voltage of approximately half that. Given the maximum bridge current of
100mA half of which flows through each 32ohm sensor filament (attenuator set to zero) the nominal voltage is 1.6V and 5% of that is 80mV. That’s
perfect for the example of a low cost bridge amp and ADC I gave you. That would allow digitisation of the 0 to 1mV span and up to the full output of
the bridge with a reasonable resolution.
Note: there may also be a common mode voltage on the + and – terminal of up to 1.6V wrt the third terminal earth/ground. I should also add the
number of bits is an open question. I gave you a common strategy based in the noise floor of the bridge. There may also be practical, cost and
processing considerations too. Usually more is considered better if other constraints allow. Testing them is insignificant as they are output in one
serial stream.
The manual for any randomly selected paper chart recorder can not possibly tell you anything about the output of the GC any more than reading the
manual of an oscilloscope or your fridge manual would. (just proof reading this now I think I know where your 600V comment came from the max voltage
on chart recorder input voltage range selctor. The chart recorder is usually a general instrument like a scope its has nothing to do with the GC
output)
My purpose in posting my comments was to help you in determining what was required in digitising the output from the GC. As, putting it plainly, from
your comments you appeared be confused about it. Considerer my comments if you want or ignore them but throwing a tantrum and try to goad me are
unproductive and is likely to make myself and others just think your talking out of the wrong end of your alimentary canal LOL.
Here is a Rolls-Royce 24bit ADC the LTC2411IMS. About £12 from Ebay so its expensive but not unaffordably so. It has almost a true 24 bits in terms
of accuracy linearity and noise. No bridge amp and only about 6sps so probably not suitable for a GC but it would make a nice 6 digit DVM if you like
lots of digit on your meter.
Shoot I almost lost this post but paging back and a resend and its back. It does help if I remember to hit Post Reply LOL
[Edited on 22-2-2018 by wg48]
Delete the chart recorder manual its confusing you
[Edited on 22-2-2018 by wg48]
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
No one is arguing with you, go ahead come up with a design for the guy. I was honest, i am looking forward to see how you solve this the way you out
lined.
I dont have your experience this is true, equally I have stuck to the same concept from the start, i havnt swung from get a 5mV plug in oscilloscope
or PC plugin DAC to what you suggest now. I am also genuinely concerned you insisted on calling it a DAC until i pointed out the difference, might not
seem much, but its the difference between an acid and a base.
So forgive me if I doubt your ability to build this and program it, nothing you have said so far has given me confidence. But I am like that, i
actually love to learn and if you prove me wrong then great, I can learn far more from that than if I am right about it.
One question you didnt answer and I havnt looked it up, what protocol you going to use with the chip? I am kind of going to follow along using a
different way.
Seriously stop getting hung up on what i say, I am positive your wrong but hope your right. I wont say any more I will watch how you solve this, i am
always up for new ways to do things.
So as I said...........Over to you
|
|
|
wg48
National Hazard
Posts: 821
Registered: 21-11-2015
Member Is Offline
Mood: No Mood
|
|
Now that is good advice.
|
|
|
SHADYCHASE54
Hazard to Others
Posts: 150
Registered: 16-12-2010
Location: CaNaDay!
Member Is Offline
Mood: No Mood
|
|
I have finally received most of what I was waiting for. I have bought an additional ADC 12 bit 1015/1115 however I am now wondering if I need it for
this project? Considering the arduino uno has a 10 bit ADC included in hardware I am wondering, if I could find out the likely maximum voltage output
of the Gow Mac GC and assuming it is low enough 1-1.5V max. Then 10 bit 1024 res. From 0 -1500Mv should be precise enough for identification of eluted
samples? Just wondering if I might be on the right path with this thinking? The thing is building the hardware is simple as cookbook chemistry in that
I have a schematic, understanding the why will take longer, programming the parameters however is throwing me through the proverbial hoop. As much
as I am excited to gain a real understanding of circuitry, I am impatient to discover the functionality of my new toy. I write this so you know the
project isn't dead for me and I would still appreciate any help and suggestions.
Best regards.
[Edited on 27-3-2018 by SHADYCHASE54]
|
|
|
NEMO-Chemistry
International Hazard
Posts: 1559
Registered: 29-5-2016
Location: UK
Member Is Offline
Mood: No Mood
|
|
Hmm looks like the ideas stopped at plug on crap.
Ok can you give me a list of what you actually have now, just to save me going through the thread and working out what you did and didnt get.
List the lot including breadboards. I am free on Sunday for a few hours, so will sit down then and go through what we got. I have the penchart in
bits, mostly its driver stuff for the pen servos and a damping circuit. The opamp in the one i got is a ........741!! Which seriously is not what i
expected, bit like expecting a BMW and finding a pushbike.
But does have some really neat buffered stuff to make up for the utter crap Opamp they used lol. I will try it on perf board, but mostly its going on
the arduino by the looks of it. Any way you can get access to a scope if needed? Would be handy for checking stuff.
Otherwise i will pm you my email, just send me the text files when we get that far, those will be from the 232 output and be the raw data i can graph.
Past that and maybe a little Matlab there isnt anything else i can do at a distance.
So be prepared to run alot of known samples of things and see how far out the graphs are, without a scope its hit and miss whatever you use. But its a
10bit machine despite all the blah blah blah, not super accurate but pointless measureing to god knows how many decimal points when the actual output
is fairly narrow.
|
|
|
| Pages:
1
2 |
![[*]](images/xpblue/default_icon.gif){kind=icon}
