03-27-2009, 09:46 PM
Well scratch that whole thing!
While looking at the data sheet on the FAN-73832 to verify the bootstrap connections I realised I over looked the fact it has a MIN supply voltage of 15V. I guess I will go with the IR2184 as well and start making changes. Good news is it gets rid of the 2 MOSFETS.
P.S. it's LWILL
03-28-2009, 12:41 PM
03-30-2009, 06:45 PM
so far so good.. The power supply is the limit at the moment. 20a is the max. Next will be real testing with the real power supply. (150v 7kw 3 phase iirc) has been a while since I have looked.
I am using igbt - the mosfets I initially used had too high Rds. These seem to hardly get warm at 20a 20khz. It the igbt are rated 900v 51a - the mosfets where 44a at 500v (but rds was .12 and you just couldn't get the heat away) Time will tell though.
I am excited.
I will post the schematic with my values and such. the blanking circuit is set to about 6us.
04-02-2009, 02:38 PM
Here are the latest eagle files - I think all of the componants are correct. :0
The schematic isn't really pretty yet - but I think the board is
I added a testpoint near the trimpot to be able to test the reference voltage. (oops)
The only real downside of this layout is you need to drill clearence holes in the heatsink to mount the cap. (but everything is really tight as far as traces go - I like)
Current limit seems to be as close to symetrical as I can measure.
04-03-2009, 04:08 PM
Interesting thread! I started to draw an servo amp today! Think this would work?
The screw terminals is still missing, and perhaps some diodes and caps in the H-bridge! I'm trying to keep it as simple as possible and only want to need 15V external psu with no regulation on the pcb.
I'm not sure if the SD would need som additional logic!?
04-06-2009, 06:38 AM
I decided to go with PWM-DIR inputs instead, feels a bit safer as you can't drive both sides at the same time! Also i put the shutdown in the logic so i can use cheaper gate drivers!
04-06-2009, 10:11 AM
Made some changes! Now the current limiter waits for the current cycle to end and therefore only limit complete cycles. Not sure if this would be a good or a bad thing But it looks nicer on the oscilloscope Comments?
04-06-2009, 12:32 PM
I don't know if your opamps common mode operation includes ground. (the lm393s Input common-mode voltage range includes ground)
Yes that is how my circuit works - cycle by cycle current limit. The flipflop doesn't get reset until the rising edge of the next pwm cycle.
As far as bot inputs going high - no difference than both inputs being low. (the motor would be either shorted thru the bottom 2 mosfets or the top 2 mosfets.) (unless the drivers chip sd is activated - then all mosfets are off)
04-06-2009, 12:38 PM
04-07-2009, 12:17 AM
Well - it passed the 160v 20a smoke test. So far so good.
That is 20 ft-lbs with the servos we have.
Also hooked up the 6kva 3phase tranformer we plan to use. - it outputs a good 125vac.
04-22-2009, 06:00 AM
Well I got a board made and after a few fixes tested it "dry" on the bench and it seems to be functioning, but I don't have a Pluto yet.
Does any one know (or can point me to) how to set up the HAL file to get it to run by it self? I know it will not be able to run at full speed, but I would like to know it works before spending any more money. (you know how times are)
I have looked at the HAL doc's (an old copy I had since the link at linuxcnc seems to be dead) and at the examples, and I am sure it can be done, I just can't quite get my head around it. It had been brought up earlier in the tread (some thing about PWM-PWM??) and it was pointed out that there would be more logic needed. Couldn't this be done in the hal file to combine (AND) the up and down signals with the pwm signal to generate them? If I knew what signal names I was looking for, I might be able to do it, but since most of the servo examples (except etch) use hardware to generate the signals, I can't find the ones I need., and etch of course does it quite differently.
Otherwise I will have to break down and buy the pluto now.
04-22-2009, 10:14 AM
I hacked the etch setup to test my first few designs. I have not looked at it recently but It is pretty close to what you want. It reads the encoders in the printer port and outputs pwm+dir.. iirc.
If you look at
you can see loadrt pwmgen output_type=1,1
Changing this to pwmgen output_type=2,2
(changes it to up/down) see..
you would have to goof around with it - good way to learn hal.