International Rectifier IRMCK201

[Xerxes]

Originally Posted by andy55 iramx: yes, that is what I have drawn there right now. need to find some 3.3V compliant optoisolators to put inbetween irmck and iramx.

Most optocouplers do have diode forward voltage of 1.2 to 1.7V. So 3.3V is no problem. It just needs smaller resistor value compared to 5V input.

BTW, it's really nice to see multiple AC drive projects going on. It will be interesting to see how MCU based design compares to IRMCK based one.

[lolailando]

I just got a CNC engraver and I was thinking to mill the PCBs. I have never done it, but I have seen examples on the net. I need to install some software and try milling a couple of them. I am not sure if it will be able to mill that precise, but I am optimistic.
Send me some files with a qfp100 to mill some boards just for testing the engraver at least.
And we will see how they come out
If it works I will mill you some boards
sorry I didn't look good enough to your schematic before posting, I missunderstood your message.

[johan_tr2000]

It is nice to have someone start the big step to realize the project. Great job Andy55.
BTW, I think we do not need to isolate the IRMCK from IRMX, becouse IRMX have the insulation it's selft (like IR2137). I think we should insulate the RS-232 out from PC. Correct me if I am wrong.

Regards,

TR2000

[Xerxes]

Optoisolation might not be necessary at all if a transformer isolated power supply is being used. I think Rutex boards don't have any kind of isolation (correct if I'm wrong). But of course isolation provides protection if something goes terribly wrong.

Hmm. How about those isolated RS-232 adapters sold in many places?

[andy55]

worked a little bit on the schematics today also, see attachment.

I really like the idea of isolating all the HV stuff. It's not nice to blow up computer + motioncontrol card etc. the first time testing the servodrive...
so I have put HCPL optoisolators between IRMCK and IRAMX + IR2175
I also re-named the high voltage ground to GNDHV, it's now separate from logic ground.

a few of my earlier questions were answered, but a lot remains:
- at24 eeprom: what to do with the pins that are not connected
- emi suppression caps for HVDC bus
- as it is now the HVDC-bus sensing (for braking) is not optoisolated in any way. The voltage is coupled thru 2MOhm resistors to the op-amp. maybe that's enough ??
- what to do with the Itrip/Temp output from the IRAMX ?
some trivial stuff that I have not had time to do:
- size of braking resistor + braking IGBT type ?
- RS232 pinout
- regulators: 5V input for logic side, need 3.3V regulator
15V input for HV side, need 5V regulator
- filtering caps for all ICs
- choose OC-sense FET
- size of series resistors for 3.3V driven LEDs (REDLED, GREENLED, HCPL2531 optos) ?

looks like it will fit on a double sided 100x220mm eurocard.
I threw the components in preliminarily... the autorouter finds a solution for double sided pcb (with a lot of vias ofcourse...some erroneously under irmck etc...) see attached JPG.
please don't go off and build something like this, it's just a very preliminary test.

maybe we should change all components to SMD and make it fit on the smaller standard eurocard 100x160mm which all boardhouses seem to quote cheap prices for ?

you will see I have used a bnc connector for the torque command, db9 for rs232, db15 for encoder+hall. 2/3 screw-terminals for DC power and motor.

now I'm thinking maybe I want two db15 connectors for enc+hall, one input for the drive and one output for passing on the signals to motioncontroller.
the rs232 could be a temporary header somewhere in the middle of the card since it is only used for tuning the system.

EDIT: attachments removed, see end of this thread for current version

[lolailando]

I think the three pins A1 A2 A3 should go to ground. That's the hardwire identifier for position zero, which I guess should be the default position when using only one eeprom IC. I have looked at the registers and I haven't seen any where any bit that adresses eeprom IC position.
Nevertheless I have emailed irf to check on this. I will post it when I get an answer.
I think all smd is good if it makes the board cheaper. We can allways rerun a new design with through hole components later.
good work andy!!

[Xerxes]

Originally Posted by andy55 - as it is now the HVDC-bus sensing (for braking) is not optoisolated in any way. The voltage is coupled thru 2MOhm resistors to the op-amp. maybe that's enough ?? - size of series resistors for 3.3V driven LEDs (REDLED, GREENLED, HCPL2531 optos) ?

Andy, you should notice that optocouplers are quite slow devices and IR2175 uses 130 kHz PWM frequency. Isolation might cause inaccuracy in current sensing.

And you can not use this kind of voltage sensing if you want isolated power stage. Grounds should be connected in your schematics. You could use an isolated opamp to do the trick instead. But if you feel that large resistor provides enough protection then why not just put large resistors between IRCMK and IRAMS? Maybe 50 kOhm with zener clamped voltages? 200 volts over 50 kOhm equals 4 mA current which can be easily absorbed by zener in case of fatal power stage destruction. Of course grounds would not be isolated so it might be dangerous to touch when powered.

I think it would be simpler isolate all I/O and let IRCMK be in the power stage side (like it's meant to be?).

[andy55]

Here's the result of todays work...

isolation: I am copying IR's example schematics where they always include isolation between IRMCK and powerstage so I assume isolation is required. Need to find optos that can cope with the 130 kHz of the IR2175. Motor phase PWMs should be lower, maybe 20 kHz.
The IRMCK measures both the duty cycle and the period of the current feedback pwm so if the optos introduce some pulse shaping then that should effect both period and duty cycle ? (=so no effect on irmck)

questions/comments compared to last version

- at24 eeprom

still have one pin not connected 'WP' ?

- emi suppression caps for DC bus:

there's room for about 4-5, how many are needed and what values ?

there's also room for more bootstrap cap's if they are needed. from reading the datasheets it's still unclear to me if IR recommends more of them or not...

do we put one braking resistor + IGBT per drive or just one braking resistor/IGBT in the psu and have the drive signalling the psu ?

'solved' this one by removing DC-bus sensing and braking altogether. maybe better to have it on the psu.
likewise (removed) for 'ITrip' and 'OC sense' inputs for IRAMX. Don't know where these would be connected so I have removed them.

- use mainly SMD or thru-hole ?

Now it's SMD for IRAMCK, 3.3V regulator, XTAL. the rest is thru-hole and it seems to fit OK onto a duoble sided 100x160mm eurocard.

- where to put the current sense resistors (irmck201 datasheet has them low-side, ir2175 datasheet has them high-side)

included them right on the motor wires like on the ir2175 datasheet. 1W SMD resistors should be big enough ? probably 20 mOhm for my 400W motors.

- connectors: logic DC in, encoder, hall, analog in, HV DC, motor, analog out

encoder+hall = one db15 connector
encoder+torque command to motioncontroller = one db15 connector (coax/bnc for torque command was probably overkill)
hv power (+HV, +15V, GND) = one 3-pin screw terminal
motor = one 3-pin screw terminal
do we need protective earth for motor also ??

RS232 is included as a 10-pin ribbon cable header in the middle of the board.
IFBCAL and FLTCLR_IN pins might be needed during setup/debugging so I have included small pushbuttons for them.

Also included 'enable' and 'fault' from motioncontroller. Maybe enable should also go to 'gatekill' on the irmck, but that pin is not 5V compliant so would need 5V->3.3V conversion (some chip?)

the autorouter finds a 2-sided solution quite quickly but it's not very pretty, 70 vias or so...

comments please ! (thermal design/requirements anyone?)

edit: if you're curious about the colors in the pcb layout they are as follows:
yellow: logic GND
light green: logic +5 V
modified green: logic +3.3 V
and on the high voltage side,
orange: HV GND
light blue: +HV
green: +15 V HV
dark yellow: +5 V HV

edit: attachment removed, see end of thread for newer version

[lolailando]

eeprom at24:
Pin WP should probably get a switch or a jumper
To ground= normal read/write operations
To Vcc= write protect the full 1k memory.
It is probably a neat option to load the parameters into the eeprom and being able to write protect them in hardware, otherwise that pin could just go to ground.
Great work!!

[andy55]

I converted most components to SMD so now there is no boardspace problem.
added protective earth connection between psu and motor connectors

still unresolved:
- do we need faster optoisolators than hcpl0531 ?
- what to do with iramx Itrip/Temp output ?
- what to do with suggested iramx shutdown circuit ?
- how to convert 5V TTL enable signal for irmck201 gatekill (3.3V)
- how to sense the DC bus voltage in an isolated way
- lots of pins not connected on the irmck201, tie to ground or leave NC ?

if and when these issues are resolved it's time to throw together a prototype. I might be able to mill a 2-sided proto pcb... otherwise I will have to order from olimex or similar. the current board is with smallest traces of 12mil and smallest clearance 6mil so it is not childs play to mill it...

and, some work on the software side for rs232 communications is probably needed...

[johan_tr2000]

I still thinking about the insulation. Why we need that much insulation. I think we can let the IRMCK201, IRAMX, ADS7818, IR2175 with the same ground reference.
What we need is to insulate the RS-232 from the PC. If we need to use the osciloscope for throubleshooting, just use the Insulation Transformer (input 110V/220V, out 110/220V, depend on your local power line) to power the Osciloscope or other device needed for throubeshooting.

This will simplified the total design & cost. Once we have get the job done we do not need any more the Insulation Transformer.

Please give some comment, and correct me if Iam wrong.

Thanks

TR2000

[lolailando]

I spoke with IRF tecnical support and they recommend no optocoupler for the current feedback.
About the unused pins they recommend to only leave unconnected the pins that have internal pull up/down.