Electronics > Projects, Designs, and Technical Stuff
I need a method to control an AC load
Ian.M:
Fast MCUs with high performance ADC and DMA are readily available and cheaper than doing RMS in the analog domain for more than a couple of channels. e.g. Microchip's dsPIC33EP256MU806 can be got for about seven bucks one-off, and has 24 analog input channels and two 10 bit ADCs, good for up to 1.1 Msp/s (megasamples per second), with DMA and auto channel scanning.
Using both ADCs, sampling 16 channels, that maxes out at over 130 Ksp/s, per channel, which is quite a bit more than you need for reasonably accurate RMS measurement, even with a 1KHz PWM chopping the waveform.
Its 16 bit CPU core runs at up to 70 MIPS and has a single cycle multiply instruction so keeping up with squaring and summing in excess of 100 Ksp/s on 16 channels in realtime is practical. You only need to take the square root once per channel per block of samples, so it isn't a significant CPU load.
However if you *MUST* do RMS in the analog domain, all you need is an OPAMP squarer circuit per channel and a low pass filter to do the bulk of the averaging. Acquisition, longer period averaging and calculating the square root can be done even on a crappy slow 8 bit MCU.
If you don't need better than a few percent accuracy, that shouldn't cost you more than a couple of bucks per channel so it should be under $5 USD per RMS current sensor. What are you paying for your off-the-shelf RMS sensors?
MarkL:
Another possible solution is to ditch the four-output transformer and instead feed each of the four load banks with an audio amplifier(s) capable of driving around 1.4ohms per channel (100W @ 12VAC) either directly or through an impedance matching transformer.
You can generate 4x 60Hz sine waves from an audio card, providing computer controlled frequency and voltage for each channel, and the resulting 12VAC would be low in distortion.
There are lots of cheap, used, high wattage boat-anchor audio amps on ebay. The ones from Crown are particularly indestructible.
jackbob:
--- Quote from: Zero999 on October 23, 2019, 08:03:55 am ---You might not need to sample as fast as you think. It's possible to measure and compute the RMS voltage or current by sampling at a lower frequency than the waveform. It depends on how often you need to read the RMS value. For example if you sampled the mains voltage at 1Hz, over a few minutes you could quite accurately calculate the RMS voltage, as long as the sampling occurs at different parts of the mains cycle, which can be ensured by introducing some jitter into the sampling.
--- Quote from: jackbob on October 23, 2019, 06:10:41 am ---I used a simple constant resistance circuit with an op amp and feedback from the source going to the inverting input and the non inverting input from a potentiometer with one side ground and the other side the rectified sineusoid. The outcome is exactly what I was looking for, it regulates the load quite linearly (with the exception of some small zero cross distortion). The FET is acting as a resistor so it does dissipate a bit of power BUT its a very simple solution.
--- End quote ---
Being picky but that sounds like a constant current source circuit, rather than constant resistance. Please post the schematic.
I'm glad you got it to do what you want.
--- End quote ---
I understand what you mean but we need the response time to be rather fast we can't wait 10's or hundreds of cycles to get an accurate reading, it would work though but I dont see it doing well with this project.
I will post a crude hand drawing of the schematic below.
--- Quote from: MarkL on October 23, 2019, 04:33:22 pm ---Another possible solution is to ditch the four-output transformer and instead feed each of the four load banks with an audio amplifier(s) capable of driving around 1.4ohms per channel (100W @ 12VAC) either directly or through an impedance matching transformer.
You can generate 4x 60Hz sine waves from an audio card, providing computer controlled frequency and voltage for each channel, and the resulting 12VAC would be low in distortion.
There are lots of cheap, used, high wattage boat-anchor audio amps on ebay. The ones from Crown are particularly indestructible.
--- End quote ---
I thought of just designing an amplifier at some point but its too much for the project (all of this is a small portion of the project). Also I would prefer the project to be standalone in the end and not rely on external equipment. The project will be be designed and demonstrated by myself and a group of engineering students I am working with, however, I anticipate it will be used and demonstrated by lots of other non technical people lets just say "joe average" should be able to plug it in and demonstrate the project.
--- Quote from: Ian.M on October 23, 2019, 01:03:56 pm ---What are you paying for your off-the-shelf RMS sensors?
--- End quote ---
These are the exact sensors we are using for the project
https://bravocontrols.com/shop/current-sensor-single-phase/
We have purchased 20 of them so I would prefer not to let them go to waste. (Originally I was not experienced with these sensors and thought they would work with a triac controlled load without issues)
Now for my quick hand drawn schematic below, I built what I think is a MOSFET being used as a constant resistor. This circuit DOES work successfully and it appears to operate as a resistor. The amplitude of the AC current sinusoid can be adjusted without affecting the quality of the sinusoid (with the exception of some zero cross distortion due to the bridge). However, doing some research, I do have some concerns using the FET's as resistors. Turns out most FET's are designed to operate as switches (no surprise) but with this optimization, most cannot operate in the linear region for more than a few 10's of ms at most. It did work fine with an IRFP250N MOSFET when I tested it but that doesn't mean it will continue to work that way. The fets tend to form hotspots in the die and suffer from thermal runaway. There are FET's designed to work for this application, I'm looking at IXYS L2 line of FETs. I havent bought them yet as they are $10 each transistor.
So I have a few questions regarding this application of FET's.
Will using them in their linear region be any different with pulsed (full wave rectified) DC? They are still acting as a resistor but will be subject to a pulsed rectified sinusoid.
Doing some quick math using maximum power transfer theorem, the FET will dissipate the most power when it is equal to the load resistance (1.5 ohms) in this case the RMS current will be 4 amps so I expect to dissipate a maximum of 24W per FET worst case.
Realistically, could the IRFP250N's handle this reliably? How does Rigol use these transistors in their electronic loads for constant resistance mode if they are not designed to do so?
The last thing I want is a FET to blow, especially while demonstrating the project haha! So if I have to I will go for the IXYS L2's made for the application, but I do have a handful of IRFP250N's and IRFP150N's and wouldn't mind using them if I felt they were up to the job.
djacobow:
Use a triac in phase angle mode and use your existing sensors.
Of course they wrong, but you can calibrate them at a handful of phase settings and interpolate between those to generate a correction factor for the phase angle. You know the phase angle because you're the one generating it.
To do the calibration you'll just need one proper current meter. You can use one on your multimeter if it is a true RMS model with reasonable bandwidth.
Since you are a student, you might also consider calculating the RMS value of a phase cut off sinusoid using basic calculus and see how well that matches your measurements.
Do not try to pwm the sinusoid. Such a thing is obviously possible, but it sounds like you'd be in over your head to do it yourself, especially given your long list of constraints driven by decisions you've already made and refuse to revisit.
jackbob:
These sensors are so poor with triac controlled waveforms. When I increase the firing angle (reducing the conduction time of the triac) the sensors actually begin to report higher currents than full conduction without the triac. I was able to make some small modifications to my last schematic of the FET as a variable resistor and it is working great (before the potentiometer control was not linear at all only about 20% of the pot range controlled the current). Now its working perfect, I'm just not sure the FET's will continue to operate reliably in this mode of operation. This design does not use any PWM only a simple op amp to control the gate voltage and keep the FET in the linear region.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version