Distorted signal on HO output of IR2110 driver

Hi,

D.A.(Tony)Stewart said:

EDA created the PNG. I just used the EDA process from a screenshot copy paste. And it was less than 1MB

Click to expand...

For a screenshot usually PNG is the right choice.

But here we have a photo and thus it´s better to use JPG.
It depends on how many different colors are used.
Also for a scope picture a the pixel count can be reduced .

D.A.(Tony)Stewart said:

Ic= C dV/dt= 330nF * 3V / 5ms ~ 20 uA not 1mA.

Click to expand...

Thank you for finding my mistake. My mind calculation did not do a good job .. it was in the middle of the night. I´d better sleep at that time.

Now I used a calculator and it says 200uA. Again a different result.
200uA is correct? Could you please confirm?

In either way, 200uA (and even 20uA) is too much for a MOSFET gate ... as continous current. The datasheet say 100nA @ 20V.

So I read the IR2110 datasheet for IB current. And it says 125uA typ | 230uA max. So now we have the culprit!
I guess I was not aware of this, because I always used the IR2110 at much higher frequencies.

****
@ OP:
Is this 5ms ON time realistic for your application?
* yes: you have to use a bigger capacitor
* no: use realistic and additionally worst case timings for the tests in future. Then you can see whether 100nF is sufficient.

Klaus

Unless the bottom devices turn on first at start up, and turn on enough during normal operation - then the bootstrap charging of the cap for the upper gate drive will not charge properly and the UVLO will stop the high side working

Also you need the de-coupling caps across the 6 switch bridge - else you will have issues related to the mid point going too low when an upper device turns off.

Hello,

It took me a while to respond, sorry for that. During this time I was working a little bit on my circuit and I would like to finally summarize what I learned since creating this post and come up to some solutions.

Thank you all who responded to this topic. Special thanks to Klaus and Tony for providing so much information for me, that helped a lot.

To sum things up:

My issue here was wrong bootstrap capacitor value for frequency I was using and that caused the UVLO to turn on. Increasing capacitor value or frequency helped to achieve nicer output signal.

Signal still wasn't perfect. The wave form looked like something had to drain the bootstrap capacitor (something like gate - source resistor but I don't have anything like this in my circuit)

That case still needs to be investigated but I think I should make another post for that because right now I have access to digital oscilloscope so I can present more precise measurments also what I need to do before going further is:

1) Solder the board because I can't really make proper diagnosis while it still on wired on breadboard. I am currently soldering the board.
Here is the current state of board

Picture 1 Current status of the board, not everything has been soldered yet.

I have some questions to design:

• There are some very tight places but I checked with my meter and nothing is shorted. Will these tight places cause a problem above certain voltage?
• Can I add resistor + diode parallel to HIN and LIN?

2) Check my circuit diagram once again because I still don't know why the ouput HO square wave isn't square. Klaus mentioned that there might be some gate - source resistor in my circuit, I need to check that.

D.A.(Tony)Stewart said:

Cb=Ic dt/dV so if you allow dV=1V for dt=4ms
Cb= 0.92 uF max 0.5 uF typ.

Click to expand...

In order to calculate Bootstrap capacitor value I have to consider, the Idle current of High side gate Iqbs and also frequency. Could you tell me why dt = 4 ms and
dV = 1V?

Once again thank you all for help

Hi,

note: Capitals are used to emphasize the importance. It´s not meant shouting.

Janecheck said:

My issue here was wrong bootstrap capacitor

Click to expand...

I´m still not sure about this.
Maybe not the capacitor was wrong but your test frequency.
In post#1 you wrote you want to do BLDC control ... I guess BLDC control uses much higher frequency. "I guess" means "I don´t know, because you didn´t tell us".
We asked several times for timing informations... to be able to help .. we need to know. Every application is different.
Maybe you work with 20Hz, then you need a bigger capacitor.
Maybe you work with 20kHz then you don´t need a bigger capacitor.

--> My recommendation: Do all the tests with the same parameters your real application uses.

This "guessing" makes it rerally difficult for us, it wastes a lot of time and causes misunderstanding.

Example:
Tony wrote: "if you allow dV=1V for dt=4ms"
And your reply is: "
Could you tell me why dt = 4 ms and dV = 1V?"

********
--> I want to emphasize two words in Tony´s line: "IF" and "YOU".
* YOU have to decide these values, not we. We don´t know about your application´s requirements.
* IF expresses that we don´t know. I´m sure Tony did NOT mean that you HAVE to use 1V and 4ms. The values were meant as example.
(Tony, please correct me if I´m wrong)

Janecheck said:

Signal still wasn't perfect.

Click to expand...

--> So tell us what "perfect" means for you. YOU need to give us the limit. How much voltage drop do you allow during which time?
Current is drawn from the bootstrap capacitor. And as soon as current is drawn .. the voltage will drop. The capacitor voltage will NEVER be a straight horizontal line in this application.

****
Now you show your new board.
I see you still think in "resistance" .. but for fast switching you need to think in "impedance".
"Impedance" depends on frequency ... and NOT on how thick the wire is.
And here we are again: We don´t know what is the expected switching frequency for your application.
--> I can say: Your board may work for 20Hz .. but I doubt it will work satisfactory for 100.000Hz.

Impedance increases with:
* increasing frequency --> mind that sometimes we need to focus on the edges (and NOT the 20Hz or what ever PWM frequency). Even at 20Hz, when a signal goes form LOW to HIGH within 1us you have to calculate with frequencies up to 1MHz.
* increasing length --> thus keep all traces that carry switching signals and wires as short as possible. As a number: If it was my design the traces between IR2110 and MOSFET would be shorter than 20mm.
* distance to return path --> thus keep the return path as close as possible to the switching signal. Like opposite side on a PCB, or twisted pairs on wires.

***

Janecheck said:

That case still needs to be investigated

Click to expand...

No, found the culprit in post#21.
****

Janecheck said:

Will these tight places cause a problem above certain voltage?

Click to expand...

Yes. Above a certain voltage. Do we know the voltage of operation?
The schematic you posted in post#1 shows 500..600V. Somewhere else you stated 12V. A huge difference in given informations.
--> I can say: your board may work for 12V but I doubt it will work for 500V.

Janecheck said:

Can I add resistor + diode parallel to HIN and LIN?

Click to expand...

For sure you can add them. Did you tell us the reason? For me it´s a riddle why you want to do so.
--> give us informations we can work with. Then we can support you the best.

I understand there is a lot that is new to you. Maybe you did not know, but designing a BLDC motor control is rather complicated. Many experienced professionals have problems with this task. It involves sensible signals, fast edges, high voltage, high current, ground bounce, noise ... and so on. All in one circuit.
Pitfalls not discussed yet are voltage spikes. They may kill your electronics ... unfortunately not immediately, it may take months...
Also: dead time between HS and LS, recovery time of the MOSFET internal body diodes.

So take your time. Don´t rush yourself. Better read a couple of application notes than focussing on a fast solution.
My advice is: go step by step, keep it simple. Start with the LOW side only (but use a fast diode instead of the HIGH side MOSFET to keep spikes low).
Do measurements on low side only. See what happens with low current then with higher current. Optimize your circuit .. then step over to the HIGH side ... with the experience you´ve got from the low side.

Klaus

Hello,

I learned so much from your answers but I still have much to work on.

I wanted to create a 3 phase inverter. I planned to generate 20 kHz SPWM from Arduino.
I wanted to test how IR2110 works by using low frequency programme but I wasn't aware of the effect it has on bootstrap circuit.
Earlier I used BLDC motor for testing with higher frequency but I didn't get best result so I thought that decreasing frequency would be a good idea...

My soldered board form previous post was a failure. HO outputs on 3 drivers showed totally different unpredictable results.

In order to find the root cause of my problem I have to get rid off the big impedance in my cirucit. In order to decrease the impedance I have designed new PCB according to your recommendations

By showing both schematic and PCB traces I can show how I really wired elements on my board.


Picture 1 Schematic

Picture 2 3D model of PCB

Picture 3 PCB traces

I tried to make HO and LO lines as short as possible and this is what I was able to achieve. Not as short as 20 mm but I hope it will work.
I also created a solid ground plane on the bottom side of the board.
I added bigger Vcc - COM reservoir and G-S resistors.

I am still using the over 300 nF bootstrap capacitor because too much in this case shouldn't be a problem? Also I noticed that people are using it with combination of electrolytic cap so did I.

Also I think that it would be good idea to pull up SD to 5 volts and control it from arduino digital output to keep MOSFETS off while arduino is restarting.

I wonder if I should add Vdd - Vss capacitor for addidtional filtering?

Also I have a lot more questions but I believe it is for another post. Right now I have to carefully analyze app notes you provided for me and think about my PCB desing.

Once again thanks for your help.

Janecheck said:

In order to calculate Bootstrap capacitor value I have to consider, the Idle current of High side gate Iqbs and also frequency. Could you tell me why dt = 4 ms and dV = 1V?

Click to expand...

I just estimated from your scope LO duration and estimated decay voltage

Hi,

wow, good job!
You really learned a lot. And it will have hige impact on your application´s quality regadring EMI/EMC, reliability, even efficiency...


I guess it`s the first time we see the transistors connected to your drivers.
The transistors - and also HO and HS - carrry high voltage. Thus it needs a certain distance to other signals and GND not to cause a spark in the tiny gaps. Be sure to have enough isolation.


The power transistors carry rather high speed swtiching signal, but also high current. High current means wide traces.
I´m not sure how much load current you expect .. thus can´t say whether the traces are wide enough.

The power (high voltage) bus carries a lot of high frequency current. Thus the DC bus needs to be stabilized by a capacitor. to suppress voltage spikes (that will kill the power transistors sooner or later) the capacitors don´t need to be huge in value, but fast = foil. I recommend one at each leg of the high side transistors source.
These transistors have nothing to do with big bulk capacitors for carrying the big motor current. For sure you need them additionally, Short distance is not that critical. They will be electrolytics, be sure they are rated for your swithcing frequency (ESR rating).

The fuse at the 5V regulator output is counter productive. You don´t need it. The problem here is that often they include some wire wound spring inside for fast swicht OFF in case they trip. These windings cause high impedance.
The IC input needs low impedance supply, thus I recommend to add fast capacitors at the 5V leg of each driver IC.

Als the LI and HI signals are swirching signals. These are not signals from A to B, they also need the return path for the current. Thus I strongly recommend to use 3 way connectors including GND. Then you are free to wire them directly or using Cs (to avoud GND loops) and Rs (to suppress ringing).

I also recommend to use 4 way connectors at the motor output... just in case you want to use shieded wirs or you want/need to install EMI filters.

Again: you really did a good job.
Some of the benefits will not be that obvious. You need a scope to compare a good design with a bad design to see the difference.
And some benefits can not be seen immediately: the bad design will fail by exploded (smoke, fire?) transistors maybe every half a year ... the other design will work for many years...

Klaus

Nice layout with adequate room for Mica insulated bus heatsinks

Now with all your Design Specs you can prepare Design Verification Test (DVT) Plan with spec limits on deadtime. Add 10:1 Res test points so 10:1 probe does not cause ringing on critrical HV nodes with gnd ref vias for coil spring 10:1 probe adapter. List all the TP's you want to add so your DVT report has ease of test references. (part of DFT)

There may be more issues with EMI filtering and sine PWM harmonics, so ferrite filters and output caps may be needed. (RLC filter)

Thesis attached : note future work
Soft switching reduce DM noise by 10 to 20 dBm at frequency higher than 3MHz
because it eliminates main switch diode reverse recovery problem.

You might also consider a 3 phase CM choke on the output. https://www.coilws.com/index.php?main_page=index&cPath=208_211_249

Hello,

I still forgot to mention important detail - load. I am sorry for providing too little information and making it harder for anyone to help me.

This project is only learning exercsie for me. I wanted to create a 3 phase inverter for typical high voltage asynchronous motor. But I don't have any motor of that kind so I am going to use my little 11,1 V BLDC motor.

Also what I want to mention is that I am going to etch the PCB board by myself at home and if this works then I want to order it from JLCPCB.

KlausST said:

The transistors - and also HO and HS - carrry high voltage. Thus it needs a certain distance to other signals and GND not to cause a spark in the tiny gaps. Be sure to have enough isolation.

Click to expand...

The distance between other lines is set to 1 mm and I think it is enough for 12V and would be even enough for 300V? Also I know that you can cover board in rosin and this should improve its insulation. Also GND is on the other side of the board so I think it is insulated enough.

KlausST said:

The power transistors carry rather high speed swtiching signal, but also high current. High current means wide traces.
I´m not sure how much load current you expect .. thus can´t say whether the traces are wide enough.

Click to expand...

DC Bus, GND, L1, L2, L3 traces are 2,54 mm thick and as I checked it should allow current around 4.5 A. I think that it is not very much but I didn't know at that moment how to make taces wider than 2,54 mm. But as I am etching PCB myself I want to put additional solder on those traces to allow more current to flow what I believe will be enough for my little BLDC motor.

KlausST said:

The power (high voltage) bus carries a lot of high frequency current. Thus the DC bus needs to be stabilized by a capacitor. to suppress voltage spikes (that will kill the power transistors sooner or later) the capacitors don´t need to be huge in value, but fast = foil. I recommend one at each leg of the high side transistors source.
These transistors have nothing to do with big bulk capacitors for carrying the big motor current. For sure you need them additionally, Short distance is not that critical. They will be electrolytics, be sure they are rated for your swithcing frequency (ESR rating).

Click to expand...

I am not sure if I understood you correctly. Because MOSFETs switch at high frequency it causes voltage spikes? (or the voltage spikes you mentioned are from indcutive load like motor and it is because indcution increases with frequency?) And to prevent this I should add a fast capacitor (rated for my PWM frequency) between source and DC bus like in a picture.


Picture 1 DC BUS + to MOSFET source capacitor

KlausST said:

The fuse at the 5V regulator output is counter productive. You don´t need it. The problem here is that often they include some wire wound spring inside for fast swicht OFF in case they trip. These windings cause high impedance.
The IC input needs low impedance supply, thus I recommend to add fast capacitors at the 5V leg of each driver IC.

Click to expand...

Yes I have to throw away this fuse. In fact today I've recently how important it is to add capacitors filter on the power supply inputs of ICs. I will add 100nF capacitors Vss - Vdd on each driver.

KlausST said:

Als the LI and HI signals are swirching signals. These are not signals from A to B, they also need the return path for the current. Thus I strongly recommend to use 3 way connectors including GND. Then you are free to wire them directly or using Cs (to avoud GND loops) and Rs (to suppress ringing).

Click to expand...

Here I am not sure what do you mean. A return path for a PWM signal? On the 12V input of my PCB I added 3 pole connector so I can connect +12V, GND from power supply to pin 1 and 2 and Arduino's COM to pin 3.

Picture 2 Arduino Ground input

For now I deleted the fuse in 5V circuit and also I added 100nF on logic supply voltage of each driver and also i switched to a 2x2 pole connector on the output.


Picture 3 Corrected PCB design

D.A.(Tony)Stewart said:

Now with all your Design Specs you can prepare Design Verification Test (DVT) Plan with spec limits on deadtime. Add 10:1 Res test points so 10:1 probe does not cause ringing on critrical HV nodes with gnd ref vias for coil spring 10:1 probe adapter. List all the TP's you want to add so your DVT report has ease of test references. (part of DFT)

Click to expand...

I haven't heard about DVT before but I think it would be a good thing to do.

D.A.(Tony)Stewart said:

Add 10:1 Res test points so 10:1 probe does not cause ringing on critrical HV nodes with gnd ref vias for coil spring 10:1 probe adapter.

Click to expand...

So I need to use a proper probe for measuring my signal but I don't understand what is a 10:1 Res test point? Should I add a resistor? I am planning to add wholes where I can solder a goldpin so I can catch it with my probe.

Here I marked places where I would want to add test points. I added them on HO,LO and Vss for refference. Test points are very close to their source signals so I guess it shouldn't distrub the main signal.


Picture 5 Marked Test Points

I also didn't add a low - pass filter to obtain a sine wave on the output but this is because I haven't studied about them yet. But I believe it will not be a problem to add this filter outside of my PCB.

Hi,

you did a good job.

Janecheck said:

I am not sure if I understood you correctly. Because MOSFETs switch at high frequency it causes voltage spikes? (or the voltage spikes you mentioned are from indcutive load like moto
and it is because indcution increases with frequency?) And to prevent this I should add a fast capacitor (rated for my PWM frequency) between source and DC bus like in a picture.

Click to expand...

Switching OFF current causes high voltage peaks generated by any inductor.
It may destroy the MOSFETs. To suppress them add a capacitor between DRAIN and GND (not SOURCE as you did)

Janecheck said:

Here I am not sure what do you mean. A return path for a PWM signal? On the 12V input of my PCB I added 3 pole connector so I can connect +12V, GND from power supply to pin 1 and 2 and Arduino's COM to pin 3.

Click to expand...

With A to B ... in your circuit it means: from microcontroller to IR2110.

So
microcontroller_LIN - cable - connector - IR2110_LIN
needs a return path close to this signal. --> in the same cable (as shield or extra wire):
IR2110_GND - connector - cable - microcontroller_GND

this automatically works for the HIN path.

Klaus

added:
I noticed you put a 1k resistor at each MOSFET G-S. I pesonally find 1k too low ohmic since the current drains out the bootstrap capacitor. I recommend to use 10k. (If I`m not mistaken the IR2110 application notes also use 10k.)

Hello,

Right now I am working on my PCB.
So I read that the capacitor betweend GND and DRAIN of high side MOSFET you mentioned is called a snubber capacitor.

For my purpose I calculated needed capacitance by using formula

Cs = 1/((U^2)*f) = 1/((400V^2) * 20 000 Hz) = 312,5 pF

For my snubber capacitor I want to use 47nF 400V polypropylene capacitor. I want to use it because it has small size so I can fit in my PCB. The capacitance is much bigger than needed and I am wondering if this will be an issue in my circuit? Also as I said earlier I want to design a PCB for a high voltage but I only have my 11V BLDC motor but I think it shouldn't be a problem?

I saw that also for snubber capacitor there is a snubber resistor added in series to limit the current but as I will be testing it on 11V motor then I think it is not needed. I will add it in future.


This is how the board looks like for now


Picture 1 Current look of PCB

I added 3 pole connector for HIN, LIN and GND so I can attach arduino's COM.
I added 47nF polypropylene snubber capacitors between drain and gnd, close to the mosfets.

I think now my board is ready to be made.

Janecheck said:

So I read that the capacitor betweend GND and DRAIN of high side MOSFET you mentioned is called a snubber capacitor.

Click to expand...

I doubt this. Give a link to the source of this information.
A snubber is always connected to the output_node = load_node.
But here the capacitor is just to stabilize the supply voltage.

We can see you added 6 cpacitors. But you only have 3 HIGH side MOSFETs.
--> Don´t use the capacitors at the LOW side MOSFETs.
Don´t use series resistors. They are no snubbers.

BTW: we only see the TOP side of the PCB. Please show the BOTTOM side, too.

Klaus

KlausST said:

I doubt this. Give a link to the source of this information.

Click to expand...

Maybe I misunderstood something about snubbers then. I was reading about it here:


"Snubber Circuits
A snubber is an essential part of a power conversion circuit. Snubbers are used in power circuits for a broad array of applications including reducing or eliminating voltage or current spikes, limiting dV/dt, or dI/dt, reducing electromagnetic interference (EMI), reducing losses caused by switching operations, shaping load lines, and transferring power dissipation to resistors or useful loads."

I assumed that they were same capacitors you mentioned because they are also used to eliminate voltage spikes.


Picture 1 Snubber capacitor and resistor. Source:

I thought about adding something like this parallel to my MOSFETs to suppress the voltage spikes.

KlausST said:

But here the capacitor is just to stabilize the supply voltage.

Click to expand...

Then maybe for that purpose I should you bigger capacitance?

KlausST said:

We can see you added 6 cpacitors. But you only have 3 HIGH side MOSFETs.

Click to expand...

Could you tell which one do you mean? I added 3 red polypropylene 47 nF capacitors connected to drain and ground of the high side MOSFETS. Maybe it will be clearer if I add an updated schematic


Picture 2 Updated schematic

And also I would like to show the bottom side of the board


Picture 3 bottom side of the board

Janecheck said:

I was reading about it here:

Click to expand...

I did not read the entire text.
A lot of text indeed, but no math, no formulas ..
At least the schemtic is misleading.

I strongly recommend not to use random internet source informations. From me the most reliable sources are:
* datasheets - directly from semiconductor manufacturer
* application notes - directly from semiconductor manufacturer
* design notes - directly from semiconductor manufacturer
* information from universities
* information from reputable electronics designers

***

Janecheck said:

Then maybe for that purpose I should you bigger capacitance?

Click to expand...

The key point is to suppress fast spikes, thus the capacitor needs to be fast and the PCB layout needs to be fast (which means, short traces, low impedance)
So: better use ceramics or foils, they are better than electrolytics.
For sure you need a bulk capacitor to store enough energy.
PCB layout. While you use copper pour ... even if used on both sides, it´s by far not that good as one really solid plane without traces or cuts.

Imagine this:
you have a solid ice block with 5m x 5m on water. It´s easy to move on this block. Now cut this block into 10 pieces and again try to move from one side to the other.
As soon as the block is cut ... it is getting unstable.

Janecheck said:

Could you tell which one do you mean?

Click to expand...

Sorry. It´s much clearer on the Schematic, but I should have seen it on the PCB, too.
You correctly added 3 capacitors.

3x 47nF is the minimum I´d say. But no need for more than 1uF. Again: for spike suppression, not as enrgy reservoir.

Klaus

Hello,

I added testpoints to my PCB at HO and LO of each driver.

Picture 1 Updated PCB layout

I wanted to to design PCB in a way that I could use it for a typical high voltage motor in the future that's why I used 400V 47nF red capacitors. Right now as I want to test my circuit on 11V BLDC motor I don't know if using a capacitor rated 400V won't affect the circuit performance. For testing purposes I planned to replace them with
50V 100nF ceramic capacitors.

I am still wondering if the snubber is needed (specially for 12V Vcc) in my circuit as long as I added those 3 capacitors you adviced. They are already suppresing the spike voltages

The FET's can switch-off the motor inductance from low voltage to extreme spikes that can damage your FET's

Choose a series R for max current such that I*R=V is the protected voltage limit well below (-25%) the FET max Vds then a series cap that decays >99% in less than 1 cycle

Hi,

Generally about electronics design around IR2110:

From all your questions it´s clear that you did not read a single reliable application note how do design such circuits.
I ask myself why. I see hobbyists rely on hobbyists circuits and designs. Making mistakes, known mistakes, spreading these mistakes.
I guess the IR2110 is more than 30 years old and since this time the problems are known, and the solutions are known. And documented in countless really good documents. Some are listed here: https://www.irf.com/technical-info/apphandbook.pdf
or directly at the IR21110 site of infineon: https://www.infineon.com/cms/en/product/power/gate-driver-ics/ir2110/#documents
In this forum, one of the most discussed IC topics are related to IR2110 (or similar dirivers). Repeatedly the same questions, the same mistakes and the same answers.

Not reading these documents makes that your circuit does not work well, that your parts explode, that your PCBs burn, that your circuit kills itself after months, that it kills your microcontrollers, makes microcontrollers to reset randomly, causes EMI problems on your devices nearby. Additional cost, additional delay additional stress.

*******

Janecheck said:

that I could use it for a typical high voltage motor

Click to expand...

currently it´s not suitable for high voltage because of too small isolation distance.

Janecheck said:

if using a capacitor rated 400V won't affect the circuit performance

Click to expand...

47nF are 47nF independent of voltage rating.

Janecheck said:

I am still wondering if the snubber is needed

Click to expand...

In opposite to BJTs, IGBTs, SCRs, Triacs ... the MOSFETs have internal body diodes. In half bridg configurations voltages on the swithcing (output) node beyond supply rails are suppressed by these internal diodes. (known and documented, btw). These body diodes have known issues.
I´m a friend of fast switching, because it reduces power dissipation. But careful design and maybe filters are necessary.
Other designers more tend to soft switching.
You are free to add capacitors and resistors.

Klaus