FacebookTwitterDiggStumbleuponGoogle BookmarksRedditTechnoratiLinkedin

Bob Higgins open sources his programmable back pressure regulator design

Written by Robert Greenyer on .

Bob Higgins has been building a programmable back pressure regulator in order to try to precisely keep a maximum pressure (above gauge) in a reactor core so that, for instance, pressure profiles claimed to have worked can be closely emulated - like those of Parkhomov.

The USB driven prototype controlled back pressure regulator circuit is now working but, as of this blog post, it is still missing a 10 micron orifice to slow the out-flow of the small volume of gas that would be in the reactor tube when the solenoid valve is activated.

Bob Higgins' USB backpressure regulator prototype


Figure 1: Bob Higgins' USB backpressure regulator prototype


The valve is supposed to be 12V 6W.  What Bob found though was that the coil is 18.9 ohms and it will activate at 3.5V and then will not close until the voltage falls to 1.0V.  At 5V, this coil will draw 265mA, so the possibility exists for it to be a fully bus powered device.  Bob will program the USB chip to tell the computer that the USB system could take up to 300mA.  Bus powered devices can be supported up to 500mA.  Once the valve is opened, it only takes about 60mA to keep it open.

The pressure is set to max. for the sensor's operation (in this case 250 PSI) when it is powered.  The device appears as a virtual serial port.  You send it bytes to program what pressure it will regulate at.  Each bit in the programmed byte you send changes the regulation by the (sensor range)/255.  The device echo's the byte you have programmed it to.  This should be easy to control from almost any control program, including Labview.


Figure2: Bob Higgins' USB backpressure regulator schematic

It is possible for this design to be modified to be bus powered, but the 5V LDO regulator would have to be replaced with a boost or buck-boost switching regulator.  For now, Bob will simply make this a <100mA USB device with an external 7-10V isolated supply to power the solenoid (mostly).  The drivers come from the USB chip manufacturer (FTDIchip.com).  FTDI has good driver support for: Windows from Win98 - Win10 (16b/32b/64b), Linux, Mac, and even Android.  The device will appear as a Virtual Com Port.  To change the back pressure regulation value, you send 0x00-0xFF to the port with 0x00 being max sensor pressure and 0xFF being 0 pressure.  The device will respond with 0xFF and then the value you programmed.  The pressure will fall until the programmed value is reached.  Use of a 10 micron flow orifice is recommended for the small volume of H2 in the Parkhomov-like experiments.  This is calculated to cause the gas pressure to fall at a sensible rate for the solenoid to control to the programmed pressure.  The flow orifice has not been tested, as of this blog post, to compare its flow resistance to what is calculated.

Most of the parts are easy to find - most are from Digikey, including the pressure sensor.  The hardest part to find was the M12 cable for the pressure sensor, which Bob found at Allied Electronics.  Bob bought the Gems solenoid valve from Amazon (believe it or not!).

Bob's prototype is constructed on a solder-less breadboard.  I made small adapters for the SOT-23 parts using small pieces of square pad vector board (the software engineer he used to work with called these "spiders").  The prototype is fully operational in evaluation testing so far.  Drivers used were the FTDI VCP default x64 bit driver for my computer (Win7).  I used RealTerm to communicate with the board over the virtual com port for testing.

Bob will hand wire a permanent version for his experiments.

If others are interested in building this programmable back pressure regulator, Bob - or perhaps another community member - may consider laying out a small printed circuit board. Please let us know in the comments below.

Comments   

 
0 #3 Robert Ellefson 2016-01-31 20:48
Great work Bob! This looks like a really useful device, and is remarkably simple to boot. I admire simple solutions like this, because I gravitate towards the complex end of the design spectrum way too quickly, and only arrive at simpler solutions after going full-circle first and ultimately discarding much of what I initially design.

I do have a bit of concern about the pulse generation function of U7 in your circuit, however. I realize that the prototype is working, but you are exceeding the specified maximum rise time of that gate (10ns/V is max per datasheet). Perhaps it would be best to change U7 to a schmitt-trigger ed buffer like the NC7SP17, which has input hysteresis and no maximum rise-time. This is the type of issue that can work fine in a prototype, but then exhibit unexpected behavior with a new layout, environmental conditions, process variations, etc.

Also, the simultaneous pulse at the FTDI chip's WR and !RD signals appears to violate the data sheet timing specs, which states a delay from !RD assertion to data valid of 20-50ns, and a data setup time prior to WR assertion (high-to-low edge) of 20ns. Again, although this is working in the prototype, it might not work under other conditions, die revisions, etc. Perhaps the use of a dual monostable multivibrator (I love that name) like a 74hc123 (or variant) would be good choice, to create a more robust timing circuit for the FIFO read, FIFO write, and ADC write signalling events.

For the 12V circuit, given the risk of hazardous ground interactions from an external supply, I would recommend adding a simple 5V->12V boost circuit like the LT1935 or similar high-switching- frequency, integrated-swit ch DC/DC controllers. This part costs $2, and uses minimal external components, including very small magnetics, and comes in a SOT23 package. It's a simple and sweet solution for what looks to be about $5 in components. There are also many monolithic micro-modules that can achieve this with less effort, for a little more cost.

Also, it might save some folks future headaches if you included series resistors on the +5V and pressure sensor signal lines that are leading to the off-board DAQ system, in case of wiring or configuration mistakes. 50 ohms should be sufficient for current-limitin g, and would also help reduce any ringing that might arise during rapid signal swings reflecting off the cabling impedance mismatch.

I hope you accept this design review feedback in the constructive manner it was intended, rather than as criticism. I think you're doing a great job here, and hope you keep it up. You're setting a wonderful example of open science and engineering for public interest.
Quote
 
 
0 #2 jeff morriss 2016-01-30 03:11
Quoting Alan Smith:
Nice work guys. If you want to try to make a 10 micron hole yourself, without resorting to lasermicrodrilling - which is commercially available but not cheap - I can only suggest you try using 6mm. soda-lime glass capillary tubing. Heat the centre of a piece to softening point and pull/stretch the ends apart.

Depending on how hot you get it and how fast to stretch it you can certainly make some microscopic tubes. Many years ago we used this technique in the lab for making hypodermic needles to puncture the nuclei of mouse white blood cells so we could suck out the contents and transfer them to another cell. All done by hand under a microscope, too!.


If you want a 10 um orifice consider using Swagelok VCR fittings. They make snubber washers with very small laser drilled orifices. I'm not sure if they go all the way down to 10 um, but a web search should give you an answer.
Quote
 
 
0 #1 Alan Smith 2016-01-29 10:40
Nice work guys. If you want to try to make a 10 micron hole yourself, without resorting to lasermicrodrill ing - which is commercially available but not cheap - I can only suggest you try using 6mm. soda-lime glass capillary tubing. Heat the centre of a piece to softening point and pull/stretch the ends apart.

Depending on how hot you get it and how fast to stretch it you can certainly make some microscopic tubes. Many years ago we used this technique in the lab for making hypodermic needles to puncture the nuclei of mouse white blood cells so we could suck out the contents and transfer them to another cell. All done by hand under a microscope, too!.
Quote
 

Add comment


Here is your generous contributions so far towards our $500,000 target, thanks everyone! : $45,020   Please Donate
See the current state of our booked costs here