24 June 2014

584. Build your own pH meter (analogue)

A few years ago a friend of mine with a degree in engineering helped me build a simple pH meter. In terms of quality of the readings there is no difference between it and our $15,000 titrino.

Either way, I'm putting the description of it online here in case other people are interested. Note that 'we' designed it based on what we found available from various websites, so the design is hardly unique.

Also note that it requires a multimeter/voltmeter to get a reading -- there's no USB/serial output or anything fancy like that.

(in related news I'm planning on taking night classes in electronics at a local TAFE (which is like a community college) so that I can start building more instruments/toys myself. A potentiostat would be nice...)

Anyway, back when we built this we ordered the parts from Digikey (I lived in the US at the time), and here's a list of what we got as shown on the invoice (the parts numbers of the linked items differ in some cases -- I presume that they are the same, but cannot guarantee that they are.
* Note that you'll need more than one of some items.
* Note that the parts includes stuff for using a whiteboard/prototyping board i.e. we first made a non-soldiered version, and then made a permanent assembly. 
* The stuff for the final pH meter (i.e. the stuff in the pomona box) is shown in red
* The stuff for connection the pH meter to a voltmeter/multimeter is shown in blue
* The stuff that's needed for the power supply is in bold black
* Stuff that I ordered at the same time but can't remember what we used it for is in the default colour (i.e. black)

Price     Item number               Description       
(USD)    (Digikey)
---------     ---------------------                --------------------      
3.05       LMC6081IN-ND      IC OP AMP PREC CMOS SINGLE 8-DIP  
2.96       ACX1046-ND          Conn. enc. bulkhead female jack x 2
3.78       7-1437529-5-ND     Conn socket dip 8 pos gold T/h

7.71       J6212-ND               BNC cable        

8.07       501-1032-NB          BNC female-dbl banana
47.85     945-1081-ND          Converter AC/DC 15W +/-12V out DL T/H

In addition, looking at the board (see pictures below), you'll also need
3 x female banana sockets
2 x K5M104(?) capacitor (x 2)
2 x BC1020TR-ND BC102 capacitor (x 2)
3 x female banana sockets
and lots of wires.

You'll measure the potential (emf) using a voltmeter, and by calibrating  the potential against a set of pH standards you can calculate the pH. In theory 0.0 mV should be at pH 7, and the potential should increase ca 59 mV per pH unit, so that pH 6 is +59 mV and pH 8 is -59 mV.

Whether this is actually true (it won't be) depends on your pH electrode, temperature, ionic strength etc.

pH box:
Add caption

Annotated view

The BNC cable is connected to a double banana contact, which attaches to the voltmeter

Power supply:
You can get a proper one, or build one yourself. I did the latter. It ain't pretty -- in fact, the following pictures should horrify you. NOTE: unless you know what you are doing you MIGHT DIE! Playing with high voltage stuff may also be ILLEGAL for unlicensed people in some jurisdictions.

Power OUT -- +12 and -12 V, and ground.

Power IN

Green goes from Ground IN to the ground pin on the right on the power supply. The thin yellow wire then goes to the ground pin on the left on the power supply. The fat yellow wire then goes to the ground out banana plug. If your ground touches ANY of the other pins you may ELECTROCUTE yourself.

583. Very, very briefly: using a USB graphics card with a headless box

This is an 'It Works' post. No tips or tricks, simply a report of something that works out of the box.

I've got a tiny server which has the following ports:
2 x USB type A, 4 x RJ45/8P8C (eth), 1x DE-9 (RS-232). There's no VGA out.

While this isn't a real problem -- I normally install linux on it either via a chroot or via a virtual machine by hooking up the HDD via a HDD-> USB cable to another computer -- it would be nice to be able to hook up a monitor to the server, in particular as I'm thinking about turning it into a lab computer to use for measurements. That it's so portable (21 x 15 x 4 cm) is an added bonus as it allows us to move it between labs.

Anyway, I bought an FY-1650  USB-to-VGA card (see e.g. here and here) for ca AUD 37 on ebay:

lsusb gives 
Bus 001 Device 003: ID 17e9:019e DisplayLink

I currently run debian wheezy (i386) on the server. Plugging in the card and a monitor and booting up everything worked without needing to do anything.

NOTE: by working I mean that I get the terminal -- I haven't tried this with a full desktop environment such as gnome, xfce etc.

Very happy...

13 June 2014

582. Briefly: freecall.com using linphone

I've covered how to set up linphone in the past: http://verahill.blogspot.com.au/2011/08/linphone-with-iinet-iitalk-voip-service.html

This is thus going to be a very brief post with just the settings for freecall.com, which offers a SIP service.

The specific settings that you need to make are (this is from my ~/.linphonerc
[auth_info_0] username=myUserId userid=myUserId passwd=PasswordInPlainText realm="sip.freecall.com" [proxy_0] reg_proxy=sip:sip.freecall.com reg_identity=sip:myUserID@sip.freecall.com reg_expires=3600 reg_sendregister=1 publish=0 dial_escape_plus=0

Yes, that's it. So, there you go - you now have a reasonably priced alternative to Skype for international calling.

I've charged my account with 10 Euro, and I've made a few test phone calls. So far so good.

See http://verahill.blogspot.com.au/2014/06/581-obi200-and-debian-and-iinet.html for how to use the obihai 200 ATA with freecall.

12 June 2014

581. Obihai Obi200 ATA and iinet Australia (and freecall)

I've had iitalk sip service as part of my naked DSL via iinet for years, but haven't really done much with it (apart from writing a post: http://verahill.blogspot.com.au/2011/08/linphone-with-iinet-iitalk-voip-service.html). I decided that it might be time to get something akin to a landline, and if it involves buying a new toy, so much better. The toy in question is an ATA device, which allows you to connect your regular old landline phone to your fancy internet router.

I ordered an Obihai OBI200 from Amazon US and had it delivered to Australia.

Getting started was pretty straightforward:
I unpacked the device, plugged in the power cable, and hooked it up to my router via an ethernet cable. I then gave it a static IP address,

Next I registered on the obitalk.com website, confirmed my email address, and logged in. You're then offered to set up your device. Note that you'll need to have a phone plugged into your obi device at this point. I bought a $32 Panasonic KX-TG1611 from JB HiFi.

Anyway, go to http://www.obitalk.com

go to dashboard

click on SP1

click on Next in the bottom right

select Generic Service Provider
The settings are as follows:
(state=vic, nsw etc. i.e. three-letter code)
Service Provider Proxy Server: sip.state.iinet.net.au
Service Provider Proxy Server port: 5060
Outbound Proxy Server: sip.state.iinet.net.au
Outbound Provider Proxy Server port: 5060
User Name: your phone number (e.g. 03 xxxx xxxx; look it up on your iinet toolbox)
Password: your VOIP password which you set up at some point
URI: yourphonenumber@sip.state.iinet.net.au

Hit submit

If all went well you'll show up as being registered.

I'm thinking about replacing skype with sip, and one provider is freecall.com. I originally signed up with them many years ago because I got a free number, but my account has expired, and they won't let me reactivate it. Either way,  you can still call pretty cheaply using their service.

The settings for the OBI200 and freecall are:
Service Provider Proxy Server: sip.freecall.com
Outbound Proxy Server: sip.freecall.com
User Name: your user name
Password: your password
URI: yourusername@freecall.com

I will work on setting it up for international calls at a later stage.

To select a specific service when making a call, type **1 for SP1, **2 for SP2 etc. See http://www.obihai.com/faq/About-the-OBi/How-do-I-make-calls-with-OBi

10 June 2014

580. Python script: Interpolate between structures in a multi-xyz file

I'm doing a lot of NEB (nudged elastic band) calculations using nwchem at the moment, and while getting 'neb convergence' is simple enough, I often get an error along the lines of
3965547 @neb NEB calculation converged
3965548 @neb However NEB Gmax not converged...Try increasing the number of beads.

While that sounds simple enough it's nicer if you don't have to go back to the beginning and e.g. run a more fine-grained PES job to generate a reasonable trajectory (straight linear interpolation often doesn't work), then keep on running neb iterations. One way to cut down on time (presumably) is to simply pad the neb path xyz with intermediate structures, and that is what this python (2.7) script does.

Oh, and I really wish blogspot would support code inclusion better...

How to use:
python nebinterpolate.py -i neb_A_F.neb_final.xyz  -o test.xyz

Say we have the structure of methanol, and methanol in which the oxygen-carbon distance is 3.0 Ångström:

Here's the corresponding xyz file, which we'll call first.xyz:
6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 0.83960 0.81877 0.83960 H 1.38912 0.20156 1.38912 6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.76087 1.75017 1.76087 H 2.31039 1.13296 2.31039

Run nebinterpolate -i first.xyz -o second.xyz and you'll get a new xyz file with three structures -- the first one plus an intermediate structure:

Here's the new file, second.xyz:
6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 0.83960 0.81877 0.83960 H 1.38912 0.20156 1.38912 6 structure 2 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.30024 1.28447 1.30024 H 1.84976 0.66726 1.84976 6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.76087 1.75017 1.76087 H 2.31039 1.13296 2.31039

 Run it again, nebinterpolate -i second.xyz -o third.xyz, and you'll get:

Here's the new file, third.xyz:
6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 0.83960 0.81877 0.83960 H 1.38912 0.20156 1.38912 6 structure 2 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.06992 1.05162 1.06992 H 1.61944 0.43441 1.61944 6 structure 2 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.30024 1.28447 1.30024 H 1.84976 0.66726 1.84976 6 structure 4 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.53056 1.51732 1.53056 H 2.08007 0.90011 2.08007 6 C 0.03517 0.00549 0.03517 H -0.61778 -0.63407 0.66798 H 0.66798 -0.63407 -0.61778 H -0.60514 0.64647 -0.60514 O 1.76087 1.75017 1.76087 H 2.31039 1.13296 2.31039

Now, the real use for this is when you've been optimising a string of structures using NEB and want to increase the number of images because you're not getting gmax convergence, or if you want to do a quick and rough optimisation and then get a prettier looking set of coordinates.

You can load the multi-xyz file in nwchem by using

  XYZ_PATH path.xyz



import sys

def getvars(arguments):

  if "-i" in arguments:
   print 'Input: %s '% (switches['in_one'])
  if "-o" in arguments:
   print 'Output: %s'% switches['o']

  if "-w" in arguments:
   print 'Weighting: %i'% switches['w']
   print 'Assuming no weighting'

  if ("-h" in arguments) or ("--help" in arguments):
   print '\t\t bytes2words version %s' % version
   print 'Creates interpolated structures'
   print 'from an multixyz file'
   print '--input \t-i\t multi-xyz file to morph'
   print '--output\t-o\t output file'
   print '--weight\t-w\t weight first structure vs second one (1=average; 0=start; 2=end)'
   print 'Exiting'
  a=0 # do nothing
 if doexit==1:

 return switches

def getcmpds(switches):

 for line in g:

   if len(xyz)==cmpds['atoms_'+str(structure_id)]:

  if n==1:
  elif n==2:
   line=line.split(' ')

 return cmpds

def morph(cmpds,structure_id):
 for n in range(0,cmpds['atoms_'+str(structure_id)]):

 cmpds['title_'+str(structure_id+1)]='structure '+str(structure_id+1)

 return cmpds

def genxyzstring(coords,element):
 x_str='%10.5f'% coords[0]
 y_str='%10.5f'% coords[1]
 z_str='%10.5f'% coords[2]
 xyz_string=element+(3-len(element))*' '+10*' '+\
 (8-len(x_str))*' '+x_str+10*' '+(8-len(y_str))*' '+y_str+10*' '+(8-len(z_str))*' '+z_str+'\n'
 return xyz_string

def writemorph(cmpds,outfile):

 for m in range(1,cmpds['structures']+1):
  for n in range(0,cmpds['atoms_'+str(m)]):
   g.write(genxyzstring(coords, cmpds['elements_'+str(m)][n]))

 return 0

if __name__=="__main__":

#check that the structures are compatible
 for n in range(1,cmpds['structures'],2):

  if cmpds['atoms_'+str(n)]!=cmpds['atoms_'+str(n+2)]:
   print 'The number of atoms differ. Exiting'
  elif cmpds['elements_'+str(n)]!=cmpds['elements_'+str(n+2)]:
   print 'The types of atoms differ. Exiting'
 if success==0:
  print 'Conversion seems successful'