next topic »

[on:]

I am currently undertaking a task that will probably get the best of me - A vacuum tube tester to check various types for Plate Voltage, Current Draw at full operating voltage (300-500VDC) and Transconductance. I need a good accurate way to sense the voltage and current with the arduino. My method to read current will be to install a high tolerance 1ohm resistor across the cathode to ground to measure in mV to get current (1mV-1Ma with ohms law). Plate voltage can be measured with the same sensor by creating a scaled voltage divider.

Are there any sensors available that could read these low voltages with good accuracy?

My final task will be to vary grid bias and have the arduino calculate transcontuctance, but that's another post all together.

Thanks. First time poster.

[Reply #1 on:]

I see two choices:
 1) Use a 10 Ohm sense resistor to get a 2V-5V range, then connect to an Arduino analog input.
 2) Use an op-amp wired with a gain of 10 to amplify the 500mV to 5V, then connect to an Arduino analog input.

[Reply #2 on:]

SFE has a OpAmp breakout board with gain 100:
http://www.sparkfun.com/products/9816
For anode ( high voltage side ) current measurements:
http://www.sparkfun.com/products/8883
With maximum gain 47 you can cover 0 - 100 mA range

[Reply #3 on:]

Perhaps just changing the reference voltage at AREF and setting it to external in programming would be enough
If you can get a reliable 1v refernece then the arduinos analog read would be from 0/1024 == 0/1v
so u could read with millivolt accuracy

[Reply #4 on:]

SFE has a OpAmp breakout board with gain 100:
http://www.sparkfun.com/products/9816
For anode ( high voltage side ) current measurements:
http://www.sparkfun.com/products/8883
With maximum gain 47 you can cover 0 - 100 mA range

Unfortunately the latter device has ground plane too close to the current-sense traces to be safe at 500V - elementary error in PCB layout!  It might flash-over at that high a potential despite the actual chip being rated for 2.1kV...

[Reply #5 on:]

Using this method you will get you 10bit ADC from 0V to 3.3V.  You lose a little over 1/3 of the values, as it starts at 1.2V, so you're really look at about 650 ADC values that you will read.  This method requires no other parts but and opamp,  4 resistors (2 the same, another of value x and a fouth of value 6x), and hook up wire.  You will get as accurate of a result as you can without adding a voltage subtractor (which you could actually get a full 1023 value range on the ADC, but does add a little more complexity).

• run an opamp with a gain of 6.  You accomplish this by having a 6:1 resistor value ratio on the opamp (any opamp in radshack or in the kits will do).  they just need to be powered.  this will give you a 1.2V to 3V range (6*200mA to 6*500mA).  google opamps as amplifiers to see the two resistors' placement and proper wiring.  Its REALLY easy
• Power the opamp it with your 5V line.  Run the ground to GND
• Then, take your 3v3 line and connect it to aRef with two matching resistors going to ground.  a good AREF tutorial is http://tronixstuff.wordpress.com/2010/12/07/tutorial-arduino-and-the-aref-pin/

Enjoy!

EDIT: I do like johnwasser's post, but then you need an external power supply that is above 5V to power the opamp, or else clipping the sensor's voltage signal will occur too early (like 4.7V) since the range is 5V.  You can't amplify up to or beyond the voltage you have powering your opamp .  Both ways, even if properly externally powered, will give you an identical amount of ADC discernible values of the sensor.

winner10920, the lowest AREF voltage is 1.1 according to the tutorial above

Please look up the voltage subtractor though.  You will need to buy a dual or quad opamp.  they are like $3.  Thi also takes a bit of research and the ight resistor values.
If you want to do the voltage subtractor, you then will need to first subtract 200mV from the signal.  find out how to do that using google.
Your desired span is 3.3V.  your actual span is 300mV.  you need to have a gain of 11 on your opamp.  For a factor of safety though, as you may not get the resistor values 100% right with your resistors on hand with either the subtracting opamp or the amplification opamp, 10 is easiest and the best (like a 1kOhm and 10kOhm resistor) though it is simple to have 11 (a 1kOhm and a series like of a 1KOhm and a 10kOhm resistor, for 11kOhm).  this will give you a 90-100% of your 0-1023 span.  That is as accurate as you will get (period).  Problem solved!

Transconductance...that might take the Vsense resistor, saving that value of the voltage drop to measure the change in current, and then finding the change in voltage and taking a ratio of the two.  This one may get a bit complicated...of i'm just too tired.

gm = dI / dV

Here is a good debate.  i'm sure someone got an answer eventually

http://www.electro-tech-online.com/general-electronics-chat/19707-measuring-transconductance-gm.html

Please excuse typos...its 4AM...and i need to get up in an hour.

[Reply #6 on:]

Unfortunately the latter device has ground plane too close to the current-sense traces to be safe at 500V - elementary error in PCB layout!  It might flash-over at that high a potential despite the actual chip being rated for 2.1kV...

I'm impressed that they were able to get that much of a measurement range into such a tiny chip.... passing 25A through a SO8 is a truly impressive feat. But then I took a look at the heat sink requirements to do something like this on a continuous basis  and then I started to wonder just how much of my PCB I would have to set aside to prevent this chip from blowing up. So I went with a Tamura S22P instead - 3x the money, much bigger package but no heat and the same 2.5V centered output signal.

What's also nice about the S22P is that you can adjust the sensitivity of the unit on the basis of how you configure the current flow through the three conductors that pass through the center of it. So, a nominal 0-18A sensor can become a 0-6A sensor, for example. I also sleep a lot better with three comparatively fat lugs on either side of the hall effect sensor carrying a 18A load vs. two legs of a SO8. Not saying that it doesn't work, it obviously does, but the Allegro solution likely needs more finesse to remain robust at continuous high currents, a condition I cannot discount in my design.

For example, it wasn't unusual to find a high-powered residential microwave pulling over 1800W on startup. While the power draw decreases somewhat over time, it's still close to 16A the whole time it's running. Makes me wonder if the 712 shouldn't be brazed to the PCB instead of being soldered. 

[Reply #7 on:]

The current path is through a sheet of copper, 25A isn't that hard to handle (it will be getting hot though, but the same copper conducts the heat to the PCB traces.  These high current devices are just a sheet of copper next to a hall-sensor, simplicity itself...  Of course it will act as a fuse at higher currents (an expensive fuse!)

[Reply #8 on:]

The current path is through a sheet of copper, 25A isn't that hard to handle (it will be getting hot though, but the same copper conducts the heat to the PCB traces.  These high current devices are just a sheet of copper next to a hall-sensor, simplicity itself...  Of course it will act as a fuse at higher currents (an expensive fuse!)

My concern was around using it in a small enclosure. I was not convinced that there would be adequate thermal heat transfer from in there, no matter how big the PCB area dedicated to allow heat transfer. These Allegro chips strike me as a great solution in enclosures that are either ventilated or in free-air situations. The Tamura sensor will create almost no heat so I feel that it's a better solution for a totally enclosed application without active ventilation.