Author Topic: Buck converter for small wind turbine project (Read 51296 times)

Janne · « **Reply #33 on:** January 25, 2012, 02:03:50 PM »

Hi,

My MPPT converter is also based on the TL494 mentioned before. Principle of the switcher is about the same as you're working with.. A buck converter with a high side FET switch, and freewheeling diodes on the lower side. I'm driving the FET's with a full-bridge driver built from TC4420 and TC4429 (complementary gate drive IC's), and level shifting for the high side fet is done with a small toroidial transformer. Your schematic is missing the level shifting for the FET, I assume you're also going to be forced to use a transformer with it as well.. High & low side gate drivers like IR2110 need also a low side switch to work. I'm attaching a pdf version of the schmatic, it may be helpful if you decide to go with the TL494 route.

It's by no means perfect, but it's been serving me well for over a year now. I've since made some modifications, at least I remember changing the RC feedback dampener of the TL494 to something like 100ohm / 1ÂµF. It could still use some more dampening(on current measurement mayby), or someone with more experience on control loops, to identify the causes of the instability. As it happens, the pulse width output of the TL494 wants to modulate some (+/- ~5%)at a frequency of about 2kHz.

On my diary is more of the practical problems about making it work; http://board.fieldlines.com/index.php/topic,130334.0.html

If I would start from null again, I would do quite a few things differently now..

1; I'd forget about 100kHz, and go with something like 20-30kHz.. Just buy some bigger capacitors and inductor cores, and have easier life making it work reliable.
2; Use synchronous buck topology. It would be more efficient (diodes dissipate quite a lot of heat at big currents..), and it would allow me to ditch the problematic gate drive transformer and just use a floated capacitor driver like IR2110.
3; Forget about the analog control. There's a micro controller in it anyway, so it would be wise to use a right tool for the job, in other words a part with high enought clock speed to be able to generate the PWM waveform on it's own. Much easier to build any interlocks into it that way too.
4; Use something else than a low side shunt current measuring, It's just a pain.. High side shunt monitors, or ready-made hall effect current sensors are not too expensive.

note to moderators; on my diary some of the images are missing, could you investigate? I think they were uploaded to the FL server, what has happened to them??

oztules · « **Reply #34 on:** January 25, 2012, 05:26:09 PM »

Nice project there Janne...... makes me want to have a go .... but I don't want a mppt in my windmill mix.. but may do it just to have a play with bucks... pwm is sort of addictive too.

I'd always use an analogue chip to drive the output..... I like the simple idiot proof control in real time they give...

Enjoyed it thoroughly

.............oztules

jeffbirkle · « **Reply #35 on:** January 25, 2012, 11:59:21 PM »

Hi All,
So does the duty cycle decrease as alternator voltage (RPM) increases?
Say the alternator at a certain RPM is happy at 24volts, and the battery bank is 12volts, is the duty cycle is 50%?
Same thing at higher RPM with 120volt, 12volt bank, and duty cycle at 10%?

joestue · « **Reply #36 on:** January 26, 2012, 02:29:14 AM »

Quote from: jeffbirkle on January 25, 2012, 11:59:21 PM

Hi All,
So does the duty cycle decrease as alternator voltage (RPM) increases?
Say the alternator at a certain RPM is happy at 24volts, and the battery bank is 12volts, is the duty cycle is 50%?
Same thing at higher RPM with 120volt, 12volt bank, and duty cycle at 10%?

the duty cycle is the voltage conversion ratio (ideally) so, yes.

GoVertical · « **Reply #37 on:** January 28, 2012, 09:10:35 PM »

Hi, update of basic buck converter circuit , aka â€œthe crazy circuitâ€œ.

When voltage at point C is greater than the battery voltage
point A is enabled
point X is 24 volts
The top nand gate oscillator and inverter toggle the transistor switches
turning the MOSFET on and off.

I am ordering gate driver IC's.

I will post test results when available. Comments welcome.

GoVertical · « **Reply #38 on:** January 29, 2012, 10:06:13 PM »

Hi, waveform verifying function of nand oscillator and npn transistors switching circuit.
circuit update: all the parts of the control circuit are working, switching frequency @ 4 k.
Next I will try to piece them together and try to trigger the MOSFET.

Buck converter for small wind turbine project

updated drawing

Janne · « **Reply #39 on:** January 30, 2012, 05:07:34 AM »

The fet driving part you have in your schematic seems odd. Remember, to turn on a fet, you need to have the gate voltage higher than the source voltage.. Ok, you have transistors in there to drive it from 24V, but NPN transistors also of course need base voltage pumped for the base current to flow in correct direction. edit. ah you have a voltage pump to generate the base voltage
But, after the FET is turned on, this circuit runs into trouble. When on, the voltage from the generator is passed to the FET source, turning the FET off again, as gate-source voltage dimishes.

The only feasible way to drive the FET gate in this case would seem to be a signal transformer.. http://omapalvelin.homedns.org/tesla/SSTC/general-sstc-notes-gatedrv.htm or google for more information.. Plenty of application notes around in that subject.

GoVertical · « **Reply #40 on:** February 01, 2012, 08:00:45 PM »

Hi, the circuit is from the .PDF file. Searching for circuit to create Vdr, maybe a charge pump can be used.
Comments welcome.

http://www.usna.edu/EE/ee320/Supplements/dcdc5_driver.pdf

rossw · « **Reply #41 on:** February 01, 2012, 08:06:46 PM »

Quote from: GoVertical on February 01, 2012, 08:00:45 PM

Hi, the circuit is from the .PDF file. Searching for circuit to create Vdr, maybe a charge pump can be used.
Comments welcome.

Already answered you here: http://www.anotherpower.com/board/index.php/topic,48.msg1487.html#msg1487

GoVertical · « **Reply #42 on:** February 18, 2012, 10:20:37 PM »

Hi, photo shows input timing signal from 555 timer and output (HO) of IR2117. The MOSFET is not connected. I had to connect Vs to ground. Duty cycle is a little greater then 50%.

The application notes says that the logic ground and power ground should not be connected but their diagram shows they are connected, I find this very confusing. I am not sure where to generate Vcc? Comments welcome.

GoVertical · « **Reply #43 on:** February 18, 2012, 11:31:13 PM »

http://www.fairchildsemi.com/an/AN/AN-6076.pdf

A more complete application data sheet covering the boot strap circuit.

joestue · « **Reply #44 on:** February 19, 2012, 01:00:19 AM »

"The application notes says that the logic ground and power ground should not be connected but their diagram shows they are connected, I find this very confusing. I am not sure where to generate Vcc?"

the issues with connecting the two are covered in the application note you just posted.

think "stray inductance"

GoVertical · « **Reply #45 on:** February 19, 2012, 08:32:41 AM »

Hi, thanks, I will have to read it a few more times. I may have found a solution using a Zener and cap configuration. It will allow the voltage for Vcc to come from Vin at point C and provide a isolated ground. I may not sure if it will work, comments are welcome. I will post results when available. Thanks for the help.

joestue · « **Reply #46 on:** February 19, 2012, 11:58:09 PM »

i must have been confused earlier.

in this simple circuit, your analog ground is also your digital ground, they have to be connected.

however you have parasitic inductance everywhere, and that can cause problems so what you have to do is connect both grounds together at a single point, that way you don't end up with transients showing up in strange places.
Additionally, do not connect the vcc of your 555 timer and the gate driver chip to the smps positive rail directly. you can however connect them through either a small inductor or a resistor.
the reason you can't connect the two is because the voltage on the drain of the mosfet at turn off will rise above the nominal vcc for as long as it takes for the energy stored in the inductance of the capacitor to find some place to go. if the ESL of the capacitor is too high, you could see 100 volt spikes on the capacitor.

also there is one more problem, when the mosfet turns off its source will dip below ground, by at least one diode drop. however you might see it fall much farther below ground for a few hundred nanoseconds, and that topic is covered in that application note.

GoVertical · « **Reply #47 on:** February 20, 2012, 05:01:56 PM »

Hi, I reviewed the operation requirements for the IR2117 and I found two main mistakes that I have made. Vcc of the 555 timer and VDR of the IR2117 have to be different values. The other mistake I found is all the grounds have to be referenced to the negative side of the load. This is a first attempt and I apologize for posting misleading information.
I am making the corrections and I will post results when available.

http://www.usna.edu/EE/ee320/Supplements/dcdc5_driver.pdf

http://web.mit.edu/6.131/www/datasheets/float_drive.pdf

Thanks for the help. The app note suggests placing a small zener in parallel with the Cbs to prevent the cap from discharging below 0 volts.

GoVertical · « **Reply #48 on:** February 25, 2012, 11:24:54 PM »

Hi, The photo displays the Vs wave form on top and the IR2117 trigger input on the bottom. I am using switching diodes because I do not have any fast recovery diodes.

The only way I could get the MosFet to turn off is to connect Vs from the IR2117 to ground. Control signal is 10 kHz and I am using bench power supplies and a 12 volt light for a load.

I am making progress.

GoVertical · « **Reply #49 on:** March 01, 2012, 09:25:00 PM »

Hi, I managed to resolve most of the problems with the proto circuit when using the bench power supplies. They were mainly caused from the power supplies and I had the new oscilloscope probes set to x10.

I still have to resolve one problem. The output from the 555 oscillator has a severe ringing that is transferred to Vs of the mosfet. But the boot strap appears to be working with the Vs from the IR2117 connected to Vs of the mosfet and the mosfet is turning on and off.

joestue · « **Reply #50 on:** March 01, 2012, 10:27:25 PM »

I don't know what waveform you're looking at, but it should be square, there should not be a slope to it.

In the second photo, the trouble starts about 10 microseconds after the rising edge, and in reply #68 on anotherpower dot com you have the same 10 microsecond half step in the waveform.
It would appear that 10 microseconds is the time it takes for the current in the inductor to fall to zero, (it would appear the current is discontinuous).
That ringing you observe would then be the inductor resonating with the capacitance of the mosfet and the diode after the current stops flowing and this is normal.

if you have any resistors from the gate to the source, remove them, change your gate resistor to 5-10 ohms instead of 60.
10uF is a more reasonable bootstrap capacitor.
also, you didn't happen to invert the signal did you? this is making less sense the more i look a it...

GoVertical · « **Reply #51 on:** March 03, 2012, 06:05:35 AM »

Hi, photo of wave form at load.

I am currently searching for away to supply the IC's with Vcc. I am not sure if I should use a 12 volt linear regulator from the 12 volt battery bank or if there is a method to use the large input capacitor at Vin. The voltage at Vin can range form 0 to over 200 volts open circuit and I do not know how fast the buck converter will drain the cap at Vin when the mosfet is oscillating. Comments are welcome.

GoVertical · « **Reply #52 on:** March 03, 2012, 10:28:43 PM »

http://www.youtube.com/watch?v=ATZ7Tfr_JQ0&feature=youtu.be

http://www.youtube.com/watch?v=omrqepe4rCo&feature=youtu.be

Hi, video shows manual operation of PMA and buck converter to a resistive load.
The mosfet is turning on during the off cycle of the IR2117. Is there a solution to this problem??? Comments welcome.

joestue · « **Reply #53 on:** March 06, 2012, 02:44:39 AM »

what is coming out of your 555?

GoVertical · « **Reply #54 on:** March 06, 2012, 06:46:34 AM »

Hi, the top waveform is the output of the 555 timer.

At this time I am replacing the 555 with a different oscillator circuit to determine if it will remove the ringing during the off state of the MosFet. With the new oscillator I should be able to vary the duty cycle. I will post results in a few days.

http://pcbheaven.com/circuitpages/Triangle_Wave_Generator/
http://www.pcbheaven.com/circuitpages/Voltage_Controlled_PWM_Generator/

joestue · « **Reply #55 on:** March 06, 2012, 06:25:53 PM »

http://www.youtube.com/watch?v=ATZ7Tfr_JQ0&feature=youtu.be
how did you connect your oscope probes to get this waveform?

GoVertical · « **Reply #56 on:** March 09, 2012, 08:01:17 AM »

Hi,
I found a reference to the problem on page 8 and 9.

http://www.nalanda.nitc.ac.in/industry/appnotes/IR/ASSETS/TECHDOCS/APPNOTES/An978.pdf

It recommends increasing the boot strap capacitor to a valve greater than 0.47 uf. I will post results when available.

joestue · « **Reply #57 on:** March 09, 2012, 08:51:30 PM »

Increasing the bootstap capacitor cannot fix the problem* because the under voltage lockout resets the flipflop.
this means that the IR 2117 will not self oscillate, if the under voltage lockout resets the flipflop, it will stay reset until the next trigger, which occurs on the next cycle.

*btw, you don't actually have a problem here. what you see is perfectly normal. keep the duty cycle fixed at 50%, and increase the load.
you'll see that the oscillation goes away when the current through the inductor crosses the continuous/discontinuous boundary.
also, if you disconnect the load you will find that the output becomes the input voltage regardless of duty cycle.

GoVertical · « **Reply #58 on:** March 09, 2012, 10:27:03 PM »

Greetings, I added the larger caps to the boot strap and the ringing is still there but only noticeable at lower RPM's.

I am still using the 555 timer. I increased the duty cycle to 80% and inverted the output so the duty cycle at the input of the IR2117 is 20% high. I connect a 12 volt 50 watt light and spinning the PMA manually I was able to illuminate the 50 watt light with relative ease. Wow, a shorter duty cycle made all the different. The PMA was much easier to spin and even maintained momentum for a short period when I stopped spinning the PMA.

Without the buck converter I was unable to illuminate the 50 watt load under manual operation.

I am still using the bench power supply to power the IC's.

Performance has definitely improved.

I still need away to control the duty cycle at different PMA RPM's and find away to power the IC's from the battery or PMA. Next step is to try charging the battery. It maybe time to start thinking about adding a micro controller. Thank you for all the suggestions.

http://www.youtube.com/watch?v=B3dLzZZ5jKQ&feature=youtu.be

GoVertical · « **Reply #59 on:** March 16, 2012, 11:51:06 PM »

Hi, I exchanged the IRF510 with a IRF540 and I am unable replicate the results shown in the last video. It is a puzzling situation. It may require a different bootstrap capacitor or some other solution. I am unable to continue until a solution is found. The circuit will function with a lower load but does not work with the 50 watt load. The oscilloscope shows the IRF540 starts to turn on then very quickly turns off, then the Vin gains magnitude and the 555 timing signal shuts down. Comments welcome.

joestue · « **Reply #60 on:** March 17, 2012, 04:18:35 AM »

can you post a photo of the whole arrangement, and a complete schematic?

I'm concerned that the reason you're running into problems is because there's too much inductance between everything. You have to build this stuff in a similar manner to the photo i posted on page 1 of this thread or you're going to run into problems.

GoVertical · « **Reply #61 on:** March 18, 2012, 07:29:36 AM »

http://www.irf.com/technical-info/designtp/dt94-1.pdf

Hi, thank you for all the suggestions I have been receiving. It looks like I purchased the wrong gate driver. I will post results when available.

Using a small 12 watt light and two bench power supply the circuit will work using the IRF540. It looks like the IRF540 requires more gate current to operate, but I am not sure. I am going to focus most of my efforts getting the circuit to turn on and off with a battery load using the IRF510. Is a circuit drawing available of your project?

joestue · « **Reply #62 on:** March 18, 2012, 08:45:43 AM »

yep, that's exactly what i was afraid of.
your schematic doesn't show the two circuits connected together.
is that the case?

GoVertical · « **Reply #63 on:** March 18, 2012, 09:19:12 AM »

I forgot to draw in the common ground. I tried it both ways with the same results. I am going to order some IR2125 and try the recommended configuration. Another problem I am have to solve is when the capacitor at Vin starts to charge the 555 timing signal cuts out. A solution maybe to try a isolated power supply but I am not sure. comments welcome

joestue · « **Reply #64 on:** March 18, 2012, 10:26:43 PM »

There's enough inductance in all those wire loops to make all manner of weird stuff possible.

As to why the 555 cuts out, the only reason it might cut out is either the power supply cuts out, or there's some voltage induced in the loops of wire that trigger the 555. i'd say that's unlikely but not impossible.
A 1 uf capacitor connected directly across the 555 will probably fix it.

If you want the weird stuff to go away, the mosfet, diode, and that big electrolytic capacitor need to be within 2 cm of each other.
additionally the bootstrap capacitor gets charged when the source of the mosfet is pulled all the way below ground. however if you have 15 cm of wire between the diode and the capacitor, then it won't get charged.

also, there's no reason to switch from one highside driver to another at this point.
take a look at this circuit board for an example of what a low inductance layout looks like.
http://diymania.hv4all.com/newihstuff/KarimLadhainspiredih/simpleihbrd2.png

GoVertical · « **Reply #65 on:** March 18, 2012, 11:20:07 PM »

Hi, If I use a voltage charge pump to increase the voltage of the bootstrap capacitor to 20 volts or greater would that turn on the MosFet? Is the bootstrap referenced to ground or Vs of the MosFet? Comments welcomes

News:

Author Topic: Buck converter for small wind turbine project (Read 51296 times)