Induction Heater (based on MK-II)
gmtandi, Fri Dec 25 2015, 04:19PMHi,
I'm trying to replicate ReactorForge MK-II induction heater.
My MK-II uses UCC37321, UCC37322 in place of TC4421, TC4422 and IRFZ44n in place of FQPF20N06L and in place of mmbt4401 i'm using mmbt2222, also in place of atmega168 i'm using atmega8-au. As ReactorForge project did not supply an .Hex file to the microcontroller, and the .basic code written is outdated (missing things, wrong pins, buttons, also uses a commercial compiler), i did not tried to compile that code also i had not started to write a C code for the MCU using AVR compiler. To run the oscilator(4046) i'm just disabling inhibit pin of 4046 with a wire from GND via to an Inhibit via in the PCB.
I'm with ringing problem in the Gate waveform. When i connect HV side to 30Vac the gates are perfect. When i connect to 110Vac, there is too much ringing. I'm using a 10ohm resistor in each gate, and my GDT consists in a ferrite core wound 17 turns of cat5e twisted pair wires. This ringing is not seen in the primary side of the GDT, only in the secondary side.
Any tip about what would be this ringing? Already checked GDT, wires, resistors, tried other values, but no luck :/
Attached are the waveforms.
I'm sorry for my english, sorry for typo errrors, i'm typing from my phone at now.
Thanks,
Happy X-mas and Happy new year!
--
(little hard to attach a photo here, my first post :P had to upload to my server and edit post. and sorry for taking photo from scope, my scope USB port is not working).
--
Just noticed that this ringing is only occuring in high-side of the igbts, would it be ripple in DC bus?
Re: Induction Heater (based on MK-II)
Steve Ward, Fri Dec 25 2015, 10:41PM
Ringing on the gate, like i see in your pictures, is usually a result of "hard switching" (or hard commutation) of the IGBT or MOSFET. In particular, the gate ringing is worse when the transistor switches late, or after the load current has crossed zero and reversed direction. The reason the gate rings is that when the transistor turns ON, the voltage across it drops suddenly, and there is some capacitance from the drain/collector back to the gate, and so it looks like negative feedback on the gate signal. When you increased the supply voltage, this negative feedback effect became greater, and the gate waveforms look worse. You can research the "miller effect" and "dv/dt induced gate ringing" to learn more.
Its better to have the transistors switch early, or just before the load current crosses zero. This will give slower voltage transitions at the drain/collector because now the voltage transition is due to the load current "back driving" into the inverter rather than the transition being caused by the gate command directly.
So if your PLL circuit can manage it, change the phase of the transistor switching vs the current and observe the gate waveforms, they should look cleaner when the switching is leading, and ugly when the switching is lagging the current zero crossings. You could also disable the PLL feedback control and simply drive above or below resonance of the tank circuit... the waveforms should be clean at frequencies just above resonance, and ugly at frequencies below resonance.
Steve Ward, Fri Dec 25 2015, 10:41PM
Ringing on the gate, like i see in your pictures, is usually a result of "hard switching" (or hard commutation) of the IGBT or MOSFET. In particular, the gate ringing is worse when the transistor switches late, or after the load current has crossed zero and reversed direction. The reason the gate rings is that when the transistor turns ON, the voltage across it drops suddenly, and there is some capacitance from the drain/collector back to the gate, and so it looks like negative feedback on the gate signal. When you increased the supply voltage, this negative feedback effect became greater, and the gate waveforms look worse. You can research the "miller effect" and "dv/dt induced gate ringing" to learn more.
Its better to have the transistors switch early, or just before the load current crosses zero. This will give slower voltage transitions at the drain/collector because now the voltage transition is due to the load current "back driving" into the inverter rather than the transition being caused by the gate command directly.
So if your PLL circuit can manage it, change the phase of the transistor switching vs the current and observe the gate waveforms, they should look cleaner when the switching is leading, and ugly when the switching is lagging the current zero crossings. You could also disable the PLL feedback control and simply drive above or below resonance of the tank circuit... the waveforms should be clean at frequencies just above resonance, and ugly at frequencies below resonance.
Re: Induction Heater (based on MK-II)
gmtandi, Sat Dec 26 2015, 10:42AM
Okay, thank you Steve, i got the point! the firmware on MCU is what makes the phase adjust, as i'm running without a MCU, no phase adjust is being done. I'll be writing a C code based on the basic code at reactorforge's github, let's try, probably next week we'll have a answer (will travel today) :)
Strange is that this behavior(gates ringing) is only on the high-side, why the ringing does'nt occurs at low-side of gates? Will take a better look in miller effect and dv/dt induced ringing later
thanks!
gmtandi, Sat Dec 26 2015, 10:42AM
Okay, thank you Steve, i got the point! the firmware on MCU is what makes the phase adjust, as i'm running without a MCU, no phase adjust is being done. I'll be writing a C code based on the basic code at reactorforge's github, let's try, probably next week we'll have a answer (will travel today) :)
Strange is that this behavior(gates ringing) is only on the high-side, why the ringing does'nt occurs at low-side of gates? Will take a better look in miller effect and dv/dt induced ringing later
thanks!
Re: Induction Heater (based on MK-II)
Justin, Sat Dec 26 2015, 02:09PM
Looks like you're using two different gate resistors. I don't suppose one is wirewound, or of a different value?
Justin, Sat Dec 26 2015, 02:09PM
Looks like you're using two different gate resistors. I don't suppose one is wirewound, or of a different value?
Re: Induction Heater (based on MK-II)
gmtandi, Sat Dec 26 2015, 02:58PM
Yes, that's correct, but their value are the same, 10r. Would that being the cause of the ringing? Also, even switching between high-side with low-side, the high-side keeps ringing, i don't think it's the resistors...
I had to use diferent kinds of resistors due to the x-mas holiday, nowhere to buy resistors on holidays :/
Today i'll gonna travel, when i back will buy new resistors and try again :}
gmtandi, Sat Dec 26 2015, 02:58PM
Yes, that's correct, but their value are the same, 10r. Would that being the cause of the ringing? Also, even switching between high-side with low-side, the high-side keeps ringing, i don't think it's the resistors...
I had to use diferent kinds of resistors due to the x-mas holiday, nowhere to buy resistors on holidays :/
Today i'll gonna travel, when i back will buy new resistors and try again :}
Re: Induction Heater (based on MK-II)
Steve Ward, Tue Dec 29 2015, 08:42AM
How is your power source isolated from the oscilloscope to allow high-side gate measurement?
Because the high-side transistor sees a large "common mode" voltage change, this can make the oscilloscope "ground" reference ring, which looks like signal ringing.
To help get the best view of the gate signal at the high side, i would recommend a few things:
(remember, for safety, the power source to the bridge should be isolated from the mains so that you can connect scope ground to whatever single point you want).
1) Set the probe for X1 mode when possible, this helps reject noise.
2) Connect the probe directly to gate and source pins using short spring contacts that solder to transistor pins (see
this
3) Wind the oscilloscope probe cable through a ferrite toroid to help isolate the oscilloscope from the big common mode voltage.
Steve Ward, Tue Dec 29 2015, 08:42AM
How is your power source isolated from the oscilloscope to allow high-side gate measurement?
Because the high-side transistor sees a large "common mode" voltage change, this can make the oscilloscope "ground" reference ring, which looks like signal ringing.
To help get the best view of the gate signal at the high side, i would recommend a few things:
(remember, for safety, the power source to the bridge should be isolated from the mains so that you can connect scope ground to whatever single point you want).
1) Set the probe for X1 mode when possible, this helps reject noise.
2) Connect the probe directly to gate and source pins using short spring contacts that solder to transistor pins (see
this
3) Wind the oscilloscope probe cable through a ferrite toroid to help isolate the oscilloscope from the big common mode voltage.
Print this page