[andy55]

Hi Group,

I'm designing an AC servodrive that needs current sensing on two phases.
It will be controlled by a microcontroller which interfaces to a pc or motioncontroller.

The high-voltage circuit obviously needs to be isolated from the pc/motioncontroller and I see two choices for where to do the isolation:
pc<->servodrive or microcontroller <-> powerstage

right now I'm favoring the latter, the digital pwm signals, enable, fault, etc. can be isolated easily with optocouplers. However the current sensing signals need to isolated too...
I've found two-three solutions so far:
1) IR2175 outputs a 130kHz square wave with differing duty cycle, could be easily isolated with optocoupler. but the 130kHz is a little high for my microcontroller which runs at max 30MHz giving me only around 7-bits accuracy...
2) ACS704 hall effect based sensor. isolated in itself, outputs an analog voltage centered around 2.5V. Availability seems to be really poor... Xerxes has had some problems with this sensor...
3) sensing resistor + diff amplifier + isolation amplifier. isolation amplifiers seem to be really expensive, need bipolar supplies etc...

any other ideas ??

Andy

PS. You might also convince me to isolate the pc<->servodrive interface but that involves 3 encoder wires, , a +/- 10V analog input, possibly 2 step/dir wires, and possibly 3-4 RS232 wires so that is not trivial either...

[pminmo]

Curious why you think your micro will only get 7 bits at 130khz? Assuming it has a decent counter/timer function built in?

[andy55]

the period of the 130kHz pwm wave from the current sensor is 1/130kHz = 7.7 us
the period of the uP is 1/30MHz = 33.3 ns

so for each 130kHz period there are 7.7us/33.3ns = 231 uP 'ticks' and the uP could thus theoretically resolve 231 different current levels. this is equivalent to ~7.8 bits

another way of doing it would be to transfer the pwm over to the uP side as a digital signal(isolated with optocouplers) and then low-pass filter it and feed it to the 10-bit analog input of the uP.
This might not be a bad idea...
the motors run at about 4500rpm maxiumum which is 75revs/s, I think there are 4 electrical cycles to one mechanical cycle so the maximum frequency for the sinewave output for one phase would be 75*4=300Hz. If we would like to sample the voltage say 10 times during this cycle then the cutoff for the lowpass filter could be set at 3 kHz ??

I wonder if my BW calculation is approximately right ? and what would 130kHz square wave lp filtered at 3 kHz look like ?

Andy

[pminmo]

Andy are you looking for more resolution or is the 7 bits ok?

[abasir]

Not sure whether this chip would work but here's the link anyway... (samples are free anyway )
http://www.maxim-ic.com/quick_view2.cfm/qv_pk/1654

[Mariss Freimanis]

OK. You mention 'low side current sensing' and then it gets confusing beyond that.

1) Switching frequency: Most loads are inductive in theory but have a significant parallel capacitance in reality. This capacitance leads to ringing (LC resonance) and whatever current servo you use must have a blanking period period after it. This blanking period must be on the order of 3 to 5uS to allow the resonance to die out or otherwise it will spoof your current servo. At 20kHz it's less than 10% of the switching cycle; at 130kHz it's well over half.

Pick the switching frequency above what anyone can hear; 20kHz, no more than that.

2) Hall sensors; forget that. The bandwith is simply not there. Use current-sensing shunt resistors and level-shifting circuits. Bottom sensing is not the way to go anyway; any short on the circuit to ground provides a sneak-path that subverts the circuitry and kills the drive due to overcurrent. Takes only microseconds.

A well designed drive uses top-sensing and a reasonable switching frequency; 20kHz instead od 130kHz. Think about switching losses; 6.5 times higher than at 20kHz. The switching losses are significant. 20A * 100V * 100ns; 400 uJ per cycle per phase per cycle.

Keep the switching freq as low as possible (ultra-sonic to human ears). Practical motor loads require no more than that. Anything above it incurrs pointless losses.

Mariss

[andy55]

thanks for your input mariss.

the IR2175 is a current sensor which uses a shunt resistor, measures and amplifies the voltage drop across the resistor and then converts this to a 130kHz square wave with variable duty cycle depending on the current. nothing to do with the pwm switching of the motor phase current itself. sorry for the confusion.

One more question: is there any difference between placing the shunts above the top transistor in the bridge ('in series with the + rail') compared to placing the shunt on the phase leg between driving IGBT stage and motor ?

[vacpress]

a side note, a distraction... this IR2175 Current sensor.. I looked at the datasheet, but am a bit confused. It says it can handle upto 400v, yes..

Also, I am not clear on the wiring. I am quite good with electronics, but do we really beleive they are passing the motor drive voltage throught that little IC?

could this be wired to see how much current a 1.75HP Single Phase AC Motor is using?

Thanks! Sorry for the side queston

[andy55]

vacpress:

The IR2175 does high-side current sensing, thus the need to sustain a high common mode voltage.

The current is measured as a voltage drop across an external shunt resistor. You need to size the external shunt resistor yourself. I was planning on 0.020 Ohm 5W resistors. the maximum input voltage to the IR2175 is 260mV so with a 20mOhm shunt that would give max 13 A current and dissipate around 3.3W of heat...

I think the IR2175 could do it, but there are probably easier ways of measuring your motor current.

[vacpress]

sure. a multimeter would probably work...

i am working on a closed loop AC router spindle speed control. (1cpr encoder, microcontroller, etc).

This is for my porter cable router, but i may offer the design to everyone if it works out. I think an affordable LCD based RPM speed control for our devices would be nice.

However, I cannot even figure out the best approach for the speed control. VFD, PWM, etc. Lots to learn. DC is much easier for me, this AC stuff is confusing, what with phases and all...
Robert.

[apollonono]

I have one opinion.
Because English cannot understand it, I use an interpreter.
I cannot express it definitely.
You should consider three systems.
A power driver and a microcontroller.
Meanwhile, with Class IR gate driver, I use common wiring.
It is the controller of the master with a microcontroller more.
Meanwhile, I use CAN or the serial communication.
I break with a photo element during this period and do digital communication.