Society of Robots - Robot Forum
Electronics => Electronics => Topic started by: TrickyNekro on March 07, 2013, 09:52:01 AM
-
Hallo guys,
it´s seriously been sometime, since I last posted here, but my time with my job and studies is rather limited.
I ordered some days ago some free samples from Analog Devices and along with them, a leaflet came for
iCouplers. I was trying to build almost since forever a very good H-bridge for my project so I guessed it would
be nice to revive the idea of the project, with these new product that I found to be quite "matching" for what
I want to achieve so...
Here is my final board, unfortunately yet to be tested but that will come very soon ;-)
(https://fbcdn-sphotos-h-a.akamaihd.net/hphotos-ak-frc1/576508_10200324713072687_1199174257_n.jpg)
I will try to post schematics and stuff as soon as I have some more spare time!
For now I would love to have your feedback!
Best Regards,
Lefteris, Greece
-
Home-made surface mount board; pretty interesting!
Which iCoupler are you using? AD1410? And, as they are magnetic, how do they work close to a big inductor like a motor?
-
I just add the schematics here, remember the board is NOT tested yet, might work perfectly might not.
Most of the circuit should work like a trip but I am not sure about the two TS12A44513 part. They are
responsible for switching the polarity at the output by choosing which half bridge is going to receive the
two PWM signals. Normally it should be ok, I double checked it, triple checked it and so on, before
materializing the PCB, but we don´t live in a perfect world... :-/
Note that the design is implementing the boot-strap method to drive the high side mosfet, allowing for
all the mosfets to be the same, but on the other hand means that the high side mosfet can´t be 100%
of the duty cycle on. It must be off and the low side mosfet on to charge the capacitors which are used
for providing power to the high side driver.
This bottleneck can be countered by isolated high side DC power supplies, but they are one hell on their
own and usually expensive so preferably the boot-strap method is used.
(http://imageshack.us/a/img585/7648/schematicsphotosheet1.png)
(http://imageshack.us/a/img24/5333/schematicsphotosheet2.png)
I´ve recorded the process of building the PCB (or at least most of it, haven´t shot soldering time,
but then again SMD soldering is not a big deal, you will find many tutorials out there).
Kinda shot it like it´s a tutorial, but I need to compose the final video now, this will take time,
it´s a lot of footage.
-
Home-made surface mount board; pretty interesting!
Which iCoupler are you using? AD1410? And, as they are magnetic, how do they work close to a big inductor like a motor?
I used the ADUM4223A (six versions of this IC exist 3223 and 4223, they have 3kV and 5kV of input output isolation respectively,
then there are 3 combinations of each A, B and C, having high side low voltage drop-out of 4.1V, 7V and 11.(something if I am correct) V respectively. I am not sure about these values check them in the site, just wanna give you the general idea).
I haven´t tested the device yet, that´s the only truth and a motor is supposed to be close enough ( less than 10cm ), but they claim that they are quite noise immune both because they are shielded and cause of the technique the use to send the signal through. I shielded them in my design with some ground plane as good as I could, normally the type of environment I am going to have this is enough, but you can never know... can you? ;-)
Best Regards, Lefteris
Greece
-
Why did you connect the A and B grounds on the ACUM isolator?
-
Why did you connect the A and B grounds on the ACUM isolator?
You kinda freaked me out for a moment, that´s not really A and B. It´s input circuit ground and one isolated driver ground.
The project is battery powered, the isolation is needed for the high side, but you get the same isolator - mosfet driver for
the low side also because you want to kinda match the propagation delay ;-)
The low side isolator - driver, is only used as a driver for the mosfet.
And to be sure that I help you with your question, note that the drivers are isolated from each other also.
You have 3 grounds, 1 ground for each driver (resulting in 2 grounds) and 1 ground for the input circuit ;-)
-
Opps... I meant the connect between GND1 (pin 4) and GND_B (pin 9).
Is there only one Battery?
It does make sense to connect these GND pins in that case.
If you are using two batteries, one for the controller logic and one for the motors
then not connecting the GNDs could be better to keep the motor switching noise isolated from the logic.
Do you have the option of cutting this trace?
-
Do you have the option of cutting this trace?
Hmmm, yeeeaahhh eeehhh... yes you need to make two cuts and it´s possible but well,
there is the idea of placing common mode chokes rather than more batteries...
Having two batteries isn´t really what I expect to have in my project, get´s things messy with charging.
I really prefer the "one battery to rule them all" method, which by the way is a problem now. I am
thinking of 3s lipos @ 5Ah (at least 5Ah) but yeah that´s another subject so...
So, yes, you can separate the grounds no problem, but generally, I don´t like the idea. I am
using high noise reject ration regulators, chokes and many other stuff.
But even with the previous designs, I really hadn´t any problem with noise problems so...
In short I really want to avoid, even the idea of a second battery. Still I exclude nothing.
But generally that´s the idea. :-P
Best Regards,
Lefteris, Greece
BTW, I am testing the s*cker, so I should will have results soon :-)
-
Isn't the whole point of isolator components that you isolate the grounds? Else you might as well just use a MOSFET or something.
-
Isn't the whole point of isolator components that you isolate the grounds? Else you might as well just use a MOSFET or something.
That was what I was thinking when I saw the grounds connected.
I now see that the ADUM4223 was used to have a 'floating' high side gate driver so that the OP could use four N channel MOSFETs.
This should work and only need one battery source.
But, even with one battery the trace between the grounds could be (and may need to be) cut. The power and ground wires for the
H-bridges should be directly connected to the battery and like wise, the power and grounds for the logic should also go directly to
the battery. This is called a 'star' connection which controls the current flow path for each load.
Then the ADUM4223's input side (GND1) only connects to the logic and the GND_B to the H-bridge. The two ground are common
only at the battery.