Headache

[esaj]

Not quite there yet... actually, not by a long shot. But at least I got a good  excuse  reason to play with Legos...  

[Cranium]

Updates?  

[esaj]

4 hours ago, Cranium said:

Updates?  

Yes and no. Here's the current incarnation:

I rebuilt the chassis (again from Legos ), but it still isn't that good. Should build a more sturdy one from wood and/or aluminum, but I really can't use the garage much nowadays. The motors aren't actually attached by anything except slight friction between them and the legos . The potentiometers on top are used for tuning the PID-control values.

Another problem I have is the missing torque: if the PWM values is below around 170 (out of 255, so about 66%), the motors cannot produce enough torque to move (if using PWM-frequency below 20kHz, I can hear the resonating with lower values, but the motors won't turn).  It can sort-of balance for a while on it's own, but sooner or later ends up tipping too much to one direction and falling over as the chassis touches the floor and the tires get up in the air. Also the center of gravity isn't really in the middle, that's why there's that green "plate" sticking out.

I have 7.2V / 3600mAh NiMH-batteries in mail (haven't picked them up yet) that are capable of 10C continuous / 30C maximum output, so that should clear up the torque problem (if it's indeed up to the 9V / 200mAh batteries, that can barely push out 1A while dropping to around 5V). Also I really should build a better chassis for it, with the center of gravity in the middle & lower than now, that should help a lot too.

Haven't done that much with this project lately, mostly I've been reading on analog electronics in general and been slowly building the PSU-project. Currently I have a couple of overcurrent protection -circuits built into matrix board, waiting to be cut out, but if left the adjustment range only up to 4.5A, waiting for lower resistance power resistors (0.01ohm), the ones I have now are 0.1Ohm / 5W max, and with 4.5A I already get around 0.45V voltage drop over those, so 0.45 * 4.5A = 2W.

Haven't been that active in the forums lately, mostly because they keep pissing in my cereal. Most of the time when I go see a topic, this is what I see:

The page stops loading in the middle, with the err_incomplete_chunked_encoding. Hitting reload "enough" times I can finally get the reply-box in the end to load, but it makes it really slow to read or especially respond to anything. If I hit the "like"-buttons, it throws me back to the top of the page, apparently simultaneously reloading it & missing a bunch of posts again   Still no problems with any other site, so I still suspect it's due to something that's happened either during the last forum update or their server config.

Edited February 15, 2016 by esaj

[esaj]

Finally got around to build a more "proper" frame for the robot... it came out slightly too big maybe, but definitely improvement over the Lego-version   Still shaky:

I have limited control over the PID-values with the potentiometers, here I've slightly adjusted them to get the robot to stay up (ie. very rapid "shaking" ).

I can now start focusing more on the software instead of building/fixing/rebuilding the Lego-frame    Most likely I'll throwaway the current software version (which has been changed so many times, that there's commented out pieces of code everywhere & hastily copy-pasted examples glued together). The current version uses an example code that enables the "built-in" DMP of the MPU6050 for calculating values. I already had a version using raw data from the IMU, which was then further processed with filters, but threw it away as I wasn't sure whether my code was faulty or if the motors lacked torque (most likely the latter, as I had similar problems with the current version too before starting to use that battery-pack on top)

 

Edited February 23, 2016 by esaj

[esaj]

So close... basically it can almost stand still, but throws a bit of a "fit" every now and then (my better half said it looks like an Alzheimer-patient ). I hot-glued the MPU6050 in place (it was earlier attached only by two pieces of tape, so it moved a bit and needed to be corrected all the time) and added a button for calibrating the "zero-position". Holding the button down, that "target" angle is constantly stored from the IMU-reading and when it's released, the last value is used as "target".

 

Another problem I've had (when it still didn't stay totally in place) was that it didn't speed up fast enough when it started to lean more heavily in either direction, while it kept accelerating, it wasn't fast enough to get it back upright... I'll have to see what I can do about that (I've upped the range of the "I"-coefficient of the PID so it would react faster to "accumulating error"), but in the worst case, the motors just don't have enough torque or the max RPM is too low. These are pretty much the cheapest basic DC-motors with a gearbox I found

[esaj]

Problem solved, at least sort-of. I couldn't get the "jitter" away completely no matter how I tried to tune the PID-values (especially the P-value), it always ended up "overshooting" or doing those more "lazy" back and forth corrections when the P-value is too low. But I then tried lowering the PWM-frequency from around 31kHz to around 3.9kHz (the next lower option there is) and after a very little of tuning had it almost perfectly still (it does do some minor corrections, but not as "violently" as before, actually it looks almost graceful ):

I think the "whine" you hear (much fainter in the video than in real life) is the PWM-frequency resonating either in the motors or in the motor driver when they "stall". And the dog doesn't like it at all   If I pick it up and let the motors run "freely", the whine stops. Anyone know why is that? I think my lowest PWM value (50%) is still too low for the motors to turn, so it tries to drive them, but cannot, and something starts vibrating... that's my theory anyway.

I promise not to spam any more videos of this, unless I have something really new to show (like remote control or AI with ultrasonic/infrared obstacle sensors)

 

 

[Cranium]

Great job!  I'm not sure what mean by when they "stall".  I would assume it is the PWM frequency under load from the dual H-bidge.  With no load when you pick it up, the current is too low to resonate.

I had issues with my balancing skateboard being jittery too.  I had to adjust the Gyro rate way down without a rider on it to get it to balance smooth.  And then with a rider on it, I had to turn it back up to react fast enough to the higher tilt inertia from a person.

There are 2 12V 7AHr SLA batteries in series on top to power it.  And the wire being held is a dead man's switch.

Gain turned down:

 

Gain turned up:

 

Edited February 25, 2016 by Cranium

[esaj]

6 hours ago, Cranium said:

Great job! 

Thanks. I actually messed around more with the PID and PWM-frequencies, and couldn't get it to work as "good" as in the last video again last night  

Quote

I'm not sure what mean by when they "stall".  I would assume it is the PWM frequency under load from the dual H-bidge.  With no load when you pick it up, the current is too low to resonate.

If (and when) I adjust the PID-values to "wrong" again, and it starts to "swivel" more back and forth to stay up, the sound stops. Ie. even with more load, it seems that as long as the motors are turning, but when the motors are stationary, it starts again.

EDIT: Worth mentioning that I do map the PID-output to PWM-values of 128...255 when the output isn't exactly zero (and it rarely is ). So that's why I was suspecting that my lowest (around 50%) PWM-duty cycle is too low to get the motors turning, and that could be the cause of the sound.

 

Quote

I had issues with my balancing skateboard being jittery too.  I had to adjust the Gyro rate way down without a rider on it to get it to balance smooth.  And then with a rider on it, I had to turn it back up to react fast enough to the higher tilt inertia from a person.

Nice. Funny thing I noticed is that it seemed to be at least a bit easier to get it to balance with the heaviest part (the battery) on top. Probably also related to inertia, too light topside makes the corrections overshoot easier or something...

Next step I'm going to try is using a boost-converter to get the voltage higher (those are 12V motors, but were recommended to be used with 6-8V and the battery only gives out around 8.5V when fully loaded, 7.2V nominal) to see if having more voltage-range (and higher max) will help staying balanced while moving, I think it should.  I've also got two more of those gearboxes to which I've switched the motors (the originals had one of their "ears" torn off during a crash with the Lego-frame, luckily they're easy to replace), don't know if they're higher RPM (although that might mean less torque). Probably I should invest in some proper motors & motor clamps 

Edited February 25, 2016 by esaj

[esaj]

Ok, I tried the booster... first attempt was an epic failure. I measured the attachment pillars to get the holes for the screws in correct positions (under the lowest horizontal level of the 'bot) and even made recessions for the screw heads. Got it nicely in place, wired everything, added a voltage display and then tried it. Nope, the voltage doesn't go up at all, no matter which way I keep turning the trimmer. Then I remembered: this one only works with 10V or above of input voltage, and the battery only gives out about 7-8.5V   

So, I took another, much smaller booster (I've got several different types, "just in case" ), soldered some wires to it (no screw terminals), made new attachments and tried again. This time it worked. Trying it with the motors, there was maybe some improvement, but not much. It seemed the motors react a bit faster, but torque-wise there wasn't that much difference. There was some voltage sag all the time (when almost stand still, the voltage would stay at the set 12V, but more movement caused it to drop to around 10V or below),  and also the booster heated up considerably.

I had earlier replaced the motors on couple of those yellow gearboxes (as usual, I needed two, so I had bought four ), as one of the "ears" had ripped of during the Lego-frame crashes earlier on those two others. I don't know what the original motors on the gearboxes were, but the replacements I had were 6V / 7000rpm ... 12V / 14000rpm motors  (these: http://www.aliexpress.com/item/5pcs-6V-7000rpm-12v14000rpm-Standard-130-motor-Green-Micro-DC-motor-Free-shipping/1698373099.html ) and fit directly into the gearboxes, I just needed to pull off the gear that goes to the motor axis from the originals and attach them to the new ones. They were certainly at least louder, and seemed that the axes turned faster with the gearboxes having the replacement motors, no idea if the torque would actually then be much worse or not.

After putting on the "new" motors, I was pleasantly surprised: they definitely ran a lot faster and also seemed to have some more torque. The problem was that the voltage from the booster sagged so much that the voltage display would turn off at times (requires around 4...4.5V just to stay on). Removing the booster in-between, they seemed to still work just fine, if not better. At least for a little while. Then it seemed the whole robot went crazy, starting to suddenly either oscillate back and forth wildly (after standing still for a while) or the motors would start jerking back and forth seemingly randomly... Bug hunting the code I didn't notice anything which could cause that (and the same code worked just fine with the other motors no matter how long I ran it in one go).

After some tries and the robot going "crazy" again, I tried the heat sink of the motor driver and nearly burnt my finger. With an infrared surface temperature thermometer, I found out that the heat sink temperature would be about 95...100 Celsius (around 200...210 Fahrenheit) after the robot goes crazy. Investigating further I found a mention of "Overtemperature protection" in the L298 datasheets (but didn't find a notice of any actual temperatures where it triggers), and then looking for that temperature-information, I found this: http://www.rugged-circuits.com/the-motor-driver-myth/

 the L298 “steals” voltage from the power supply. More formally, there is a voltage drop between the input voltage and the L298’s motor outputs. The worst-case drop in voltage is 3.2V at 1A of motor phase current and 4.9V at 2A of motor phase current.

According to the datasheet, typical voltage drop (the center column on the right side of the datasheet) in each L298 channel is 2V+1.7V=3.7V (worst-case is 4.9V). If the motor current is 2A, the typical power lost in the L298 would be 7.4W (worst-case is 9.8W). Not only is this power that is wasted because it never makes it to your motor, it is a power loss that heats up the L298.

Can the L298 operate when it’s dissipating 14.8W of power (7.4W in each channel)? Not even close. The L298 will go into thermal shutdown long before it supplies 2A to two motor phases.

I had missed those voltage drop mentions entirely in the datasheets   Around 2 to 5V voltage drop? Up to around 15 watt power dissipation with two channels (remember, I'm driving two motors with it)? No wonder it gets hot   After having set PID-control to nearly "perfect" (robot stands pretty much standstill, just doing a few very small adjustments now and then and staying pretty much stationary), letting the driver cool down a bit between attempts, I let the driver cool down to room temperature, then set the robot standing on a floor. After about a minute of almost total stillness, it goes crazy, starting to oscillate back and forth with growing amplitude before falling over after some seconds. My best guess is that the motor driver goes to thermal shutdown, the heatsink drops the temperature fast below whatever the limit is, the driver turns back on, does a fast acceleration to regain balance, the motor driver again goes to thermal shutdown, the temperature drops, there's a fast correction, the motor driver goes to thermal shutdown... you get the picture.

Also, I probably could get more speed/power out of the motors without that voltage drop. Apparently the original motors must have a much larger internal resistance, as it never happened with them (and the heat sink didn't even warm up much when I tried it a couple of times earlier), whereas these replacement motors seem to draw much more power (even with same voltages). So, it now looks like I'll have to build my own H-bridges with low resistance power mosfets to drive these (I have some L293-drivers, but I doubt those would do any better, maybe worse as they're just dip-chips without any cooling). Luckily, this coincides with my other project, the 3-phase motor drivers. Probably the best course of action is to go ahead and build 10 half-bridges... why 10? 4 half-bridges for the two H-bridges for these single phase motors, and 6 to build two separate 3-phase motor drivers.

 

[Cranium]

I've released the magic smoke on a L298 driver before.  They do like to get hot.  I was driving a stepper motor and playing with various levels of current at the time.  Just to stay upright, it sounds like it is using too much current though. 

I look forward to seeing your custom driver.  It is giving me the itch to get back into my self balancing projects.  I went the easy way with the driver on mine: http://www.dimensionengineering.com/products/sabertooth2x25

I need to figure out how to do an encoder on the motors I have or to do one with the bigger wheelchair motors I have.  I want the bot to know how to stay in one spot.

[esaj]

3 hours ago, Cranium said:

I've released the magic smoke on a L298 driver before.  They do like to get hot.  I was driving a stepper motor and playing with various levels of current at the time.  Just to stay upright, it sounds like it is using too much current though.

It might be my stupid PWM-values (min. 50% duty cycle, if there's any error ), so it's (needlessly) running current through the motors?

3 hours ago, Cranium said:

I look forward to seeing your custom driver.  It is giving me the itch to get back into my self balancing projects.  I went the easy way with the driver on mine: http://www.dimensionengineering.com/products/sabertooth2x25

That looks like a nice driver, but waay out of my budget... unless I can find a chinese copy at the fraction of the price   This "budget" (read: cheapskate) robot has cost me maybe about 30€ so far (out of which the battery cost nearly 20€ ).

Here's my current (work-in-progress) H-bridge:

I decided to build it separately instead of using similar half-bridges as what I had with the earlier 3-phase motordriver. The high-sides are pretty much the same as there (taken from the e-bike schematic here:   

 

The rest (low-side drive, direction selection) was designed by me (which is not a brag, but a warning: don't trust it to work ), and probably overly complicated   But hey, if I can select the direction just with one input...   Still needs more simulation, I want to make sure the bridges won't shoot through when the direction changes (although to be doubly sure, I'll probably implement small dead time insertion in code too) or otherwise. After I get good results from simulation, I'll try to build it in a breadboard, and then we'll see how well it works out in real life. I'll try to get around to write up a proper explanation how it works (or at least is supposed to), if it works .

 

 

3 hours ago, Cranium said:

I need to figure out how to do an encoder on the motors I have or to do one with the bigger wheelchair motors I have.  I want the bot to know how to stay in one spot.

I've seen at least "slotted" optical encoders, maybe that + a disc with holes could work?

[esaj]

Ran some simulations, and while in general it seems to work, I'm not 100% happy. Here's the motor current, changing DIRECTION-signal every 25ms (which is pretty fast, 40 times per second, but as the simulation runs fairly slow at around 1ms per wall-clock second, I didn't want to wait too long ):

All is dandy, the current direction through the motor changes.

But, as the direction is changed, there IS a short shoot-through on some of the bridges (at least sometimes):

For whatever reason, it didn't happen on the right-side half-bridge (M3 & M4), but you can see those spikes in M1 and M2 -drain currents here and there when the direction change occurs. So they're both conducting for a short while at the same time, and instead of the current passing through the motor, it shoots through the bridge (and could potentially fry the mosfets). Also, there's some small current flowing "in wrong direction" through M3 & M4 (right-hand side half-bridge high- & low-side), not sure why that is, maybe because the transistors allowing/blocking the PWM-pulse are "the other way around" (NPN vs. PNP) than in the opposite side?

Here's a close up on the currents at around 225ms (just as the motor is changing direction):

Although it lasts only for a short while (a few microseconds?), you can see almost 10A shooting through M1 & M2 (and I've used mosfets with relatively high Rds(on), 75mOhm, as I was planning on using IRF540's for the bridges, which have max 77mOhm @ 10V gate-source -voltage & 17A drain current). 

I tried different sorts of capacitor-thingamabobs with the DIRECTION-signal, but they only made the whole thing go kablooie as the PWM-pulse started "leaking" to all the direction-controlling transistor bases    Might be that it's not an issue with software-based dead time insertion, but I was hoping I could somehow make it "sure" on hardware-level that it shouldn't ever occur. So probably safest bet is to turn off PWM (0% duty cycle), switch the DIRECTION-signal, wait a short while (some tens of microsends?) to make sure the mosfets are not conducting anymore, and then start the PWM again. Also, do note that the "motor" is only represented as a resistor here, I have very little idea of how it's going to affect things in reality. I'm really, really starting to lean on ordering that oscilloscope (Rigol DS1054Z) that I've been eyeing for over a month or so

Lastly, at least some good news: during "normal" operation (ie. when it's not changing direction), it seems that it works as it's supposed to. Here's M1 & M4 conducting, and M2 & M3 are supposed to be off:

 

The top-most value is the current running through the motor. Id(M1) and Id(M4) are the drain-currents of the mosfets that are supposed to be conducting with the PWM-pulse. All looks fine, except there is that slight "bump" before it fully turns on, and it can actually be seen in the motor current too. Maybe I should still take a closer look at that...

V(n027) is the voltage at the gate of M2 (low-side mosfet of the left-hand side bridge). There's some ripple there coming from, um... somewhere? But not enough to make it conducting (at least much). 

Id(M2) is the drain-current entering (and flowing through) M2, there's some leakage (about 20mA in both directions), but it's so little that I don't think it will matter?

V(n011) is the voltage at the gate of M3 (high-side mosfet of the right-hand side bridge). There's some voltage going up and down there, but as it's the high-side, the gate-voltage must be above the source-voltage, so not much current is being conducted, as Id(M3) shows (closer to 50mA in the "other" direction, but I don't think it's that dangerous?).

I probably have extraneous parts there, as the whole thing was built (more or less) by trial & error, and some values probably still need tweaking, but overall, I'm surprised I could get it to work (at least mostly) in the first place  

Edited February 27, 2016 by esaj

[lizardmech]

I have to find some new voltage regulators for my board, I need 5v - 15v boost for mosfet driver and 5v - 3v for logic. Any suggestions? I'm trying to reduce size and assembly complexity as much as possible.

[esaj]

9 hours ago, lizardmech said:

I have to find some new voltage regulators for my board, I need 5v - 15v boost for mosfet driver and 5v - 3v for logic. Any suggestions? I'm trying to reduce size and assembly complexity as much as possible.

The boost of the high-side mosfet gates in the above H-bridge is done with charge pumps (the high-side gates are boosted up to little over 13V according to the simulation, the battery voltage is 8.5V), similar setup should work, although it probably requires tweaking C1-C4 (in the half-bridge high-sides) depending on voltages and the PWM-frequency (I've 31.4kHz here in the simulation, as that's what I'm going to use in real world)? I still need to tweak those a bit myself to get higher voltage (it seems to mostly depend on the capacitor sizes and their capacitance ratios, unfortunately I cannot offer an equation to solve "perfect" values ).

As you say that you want to reduce size and complexity, the pumps are probably not a good solution. There are probably lots of mosfet gate-driver ICs, that should save up space & make it less complex to assemble, but I really don't know much about those.

For voltage regulation (not needed here, but in general) I have basic 78xx's (linear regulators), LM2596's (switching regulator, 3.3V, 5V, 12V & adjustable) and LM317's (adjustable), all in TO-220 -packages. None of those can go above 35...40V input voltage, and I've mainly picked them just because they were cheap

78xx: https://www.fairchildsemi.com/datasheets/LM/LM7805.pdf

LM2596:  http://www.ti.com/lit/ds/symlink/lm2596.pdf

LM317:  http://www.st.com/web/en/resource/technical/document/datasheet/CD00000455.pdf

Edited February 27, 2016 by esaj

[lizardmech]

Oh, my one is totally different, I have a 3 phase 600v mosfet driver, it needs 15v input while my other logic uses 3v. The board has a 5v input from another source, but converting it to 15v and 3v is annoying, they're taking up too much room and need too many other components to function. If I could find a monolithic IC for voltage I might be able to replace all my components with film versions which would save space and assembly time.

[esaj]

3 hours ago, lizardmech said:

Oh, my one is totally different, I have a 3 phase 600v mosfet driver, it needs 15v input while my other logic uses 3v. The board has a 5v input from another source, but converting it to 15v and 3v is annoying, they're taking up too much room and need too many other components to function. If I could find a monolithic IC for voltage I might be able to replace all my components with film versions which would save space and assembly time.

Oh, ok. Sounds like you know what you're doing, unlike me   Being just a hobbyist, my "electronics design" is more like best guesses and trial-and-error

Stupid question: is that 600V a typo (did you mean 60V?) or are you using something like 3-phase AC-mains for the power? Somehow I doubt you'd take that much voltage from batteries, unless you have a hell of a booster?

I think @Cranium had some evaluation board for some bigger input-voltage (100V? Don't remember, and didn't bother to check) regulator, but that still wouldn't be nearly enough... if you need only very small currents, you maybe could use a resistor-voltage divider, but if it's more the resistors are going to dissipate a lot of power. If it's AC-input, I'd hazard a guess you'll need a transformer + rectifier, filtering and then regulation?

If it's DC-input, there could be some rectifiers or buck-circuits to bring down the voltage, but I don't know of any (I've never looked for anything that can use such high input-voltage).

Maybe regulating and boosting the 5V input to 3V and 15V could be a better option? I have some cheap small buck/boost converters built around  LM2596's for such (not made/designed by myself, I bought them ready-made), although they probably cannot supply very large currents (< 1A or so?)...

Edited February 27, 2016 by esaj

[esaj]

Time for yet another stupid question. I've been planning on using IRF540 (or 75NF75, but that's probably overkill) for the H-bridge. The reason the low-side gate-circuits are as complex as they are (although there could be a way to simplify them and still work like they currently do) is that I use 5V PWM-signal (from an Arduino) and the battery voltage is 8.5V. To make sure that the low-side Mosfets are fully saturated (is it called "active mode"?), I designed those circuits to give as high as possible voltage to the gate (oscillating between around 0.1V and 8V currently in the simulation). But trying to decipher the voltages from the datasheets, I've noticed that it says 4.0V max gate-source threshold voltage.

So my stupid questions are: is the threshold voltage of the datasheet the voltage where the conduction begins or where it is already fully saturated? Looking at these two graphs:

It would seem that with 4.5V V_gs it would enter the saturated region with very low V_ds and then stay in the saturated region even as V_ds goes up. It looks like the max current is then around 6A, which still should be plenty for a small DC motor. I'm not sure what my Vds on the low-side will actually be, as the motor probably affects things there (I've only simulated it as a resistor, instead of resistor + inductor + voltage source, as I don't know the inductance values and simulating the changing voltage of the motor is a bit difficult). Using higher gate-voltage, it would actually seem that the saturated region cannot even be reached with low V_ds (But the currents are higher)? The thing is, if I could "get away" with 4..5V on the gate, I wouldn't need that complicated circuits on the low-side gates, still I'd like the mosfets to give "max output" and as little power dissipation as possible. Thoughts?

 

[Cranium]

Gate threshold voltage is when it starts to conduct.  In the spec sheet, this is a maximum of 4V with a minimum of 2V and typical of 3V   With these MOSFETs, they will saturate very quickly.  I just tested one and it started to conduct between 3.6V and 3.7V and would become fully saturated at 4.1V.  This was with a Drain voltage of both 8V and 30V.

These tests were at very low current limits (0.1A).  I believe what the graph is showing you is that for higher currents, you need more gate voltage to fully saturate at lower drain voltages.  For example, at 1V drain voltage, I didn't reach full saturation of 1A until the gate voltage was 4.6V (Gate had to be higher than Drain).  This corresponds to what the graph shows for these values.

[Cranium]

On ‎2‎/‎27‎/‎2016 at 10:37 AM, lizardmech said:

Oh, my one is totally different, I have a 3 phase 600v mosfet driver, it needs 15v input while my other logic uses 3v. The board has a 5v input from another source, but converting it to 15v and 3v is annoying, they're taking up too much room and need too many other components to function. If I could find a monolithic IC for voltage I might be able to replace all my components with film versions which would save space and assembly time.

How much current are you wanting to use at 15V?  A step up convertor won't be very good at higher currents so if it is >0.5A, I would look at stepping down from 15V to 5V instead (if this is possible) since your 5V won't likely be using as much power.  And then use something like a LD1117 or LM317 to step it down to 3V for logic level stuff since this will be very low current.  But I'm doing a lot of guessing here since I don't know where your 5V comes from now or how much power you are needing at each voltage.

[lizardmech]

7 hours ago, Cranium said:

How much current are you wanting to use at 15V?  A step up convertor won't be very good at higher currents so if it is >0.5A, I would look at stepping down from 15V to 5V instead (if this is possible) since your 5V won't likely be using as much power.  And then use something like a LD1117 or LM317 to step it down to 3V for logic level stuff since this will be very low current.  But I'm doing a lot of guessing here since I don't know where your 5V comes from now or how much power you are needing at each voltage.

I'm using this driver http://www.infineon.com/cms/en/product/power/motor-control-and-gate-driver-ics/isolated-gate-driver-ics/6EDL04I06PT/productType.html?productType=db3a3044300452850130640721652fc3

I think it only needs a few hundred milliamp? I have the h-bridge isolated from the logic powersupply, 5v is provided via canbus. eventually I will build a separate module for converting high voltage to 5v. For testing I will just use a 5v battery pack. At the moment it's annoying because the 15v and 3.3v regulators are using up maybe 25% of the space on my board design.

[esaj]

After lots of changes & fiddling & simulations to find "correct" values, I'm starting to feel (somewhat) confident that this could actually work in real life (maybe ), at least it doesn't anymore draw up to 400mA from the PWM- and DIRECTION-inputs

The low-side gate controls are now much more simpler (yeah, I had totally unnecessary components there ). I'll try to build it on breadboard soon(ish) to see if it actually works & to make some measurements with the cheapo oscilloscope, before trying to solder it together...

If anyone more knowledgeable on electronics / motor controls could take a closer look at it (I can also provide LTSpice .asc -file if someone needs/wants it) and tell me if there's something clearly wrong (or just comments / tips in general), that'd be very much appreciated.

[esaj]

One of these days... I actually learn to use the space in the breadboards and not cram everything so tightly together, that I cannot even get clips in between or see which wire goes where.

I have half-working H-bridge... Why half-working? Because it works running the motor in one direction (so at least the other high-side and other low-side is ok), but not in the other. So "fun" to try to find the mistakes from that mess, probably easier to take it apart & rebuild than trying to find the problem.  

Oh yeah, pulled the trigger on the Rigol. That scope there is a pain to work with (well, what could I expect for about 16€... ), the voltage values "drift" a lot, it's painfully s...l...o...w if you need to scroll the signal in vertical direction and when it shows the frequency (it doesn't all the time even when it triggers just fine etc), it jumps between values like 25-35kHz. I won't even go further with all the little quirks and "usability" of it. But still better than nothing, though, otherwise I wouldn't see anything and would be left totally guessing 

 

[dmethvin]

14 hours ago, esaj said:

One of these days... I actually learn to use the space in the breadboards and not cram everything so tightly together, that I cannot even get clips in between or see which wire goes where.

I have a jumper kit like this which makes it easier to trace things. You can always use the long loose jumpers as well when you need them.

http://www.aliexpress.com/item/140-Pcs-U-Shape-Shield-Solderless-Breadboard-Jumper-Cable-Wires-Kit-for-Arduino/1845027948.html

 

[esaj]

11 minutes ago, dmethvin said:

I have a jumper kit like this which makes it easier to trace things. You can always use the long loose jumpers as well when you need them.

http://www.aliexpress.com/item/140-Pcs-U-Shape-Shield-Solderless-Breadboard-Jumper-Cable-Wires-Kit-for-Arduino/1845027948.html

 

That might be useful, thanks.   

[Cranium]

You will be very pleased with the Rigol!  Did you get a 2 or 4 channel model?  Mine is two but for the motor controller, I could see where 4 would be very helpful.  At some point you may want to look into a logic analyzer as well.

 

I use those wire jumpers as well.  They are very helpful but the spaghetti mess can still occur.  lol  This is from a LED stairs project I did. I mocked up a prototype before making the circuit boards.