Heat sink for 1GBT CM600HA-24H

[gary350]

FWIW, I don't see shortbus= as a 'complainer', he is more like one of a number of spectators, watching (unfortunately) someone flailing around in a pool. Life rings are being tossed in from every direction, but the swimmer appears to not want to grab any of them.

That's my view on this, and a number of other threads.

I have copies of all the things we talked about in the past. Its not time for me to build a REAL induction heater until I have all the pieces & parts made. It appears to me making chokes are trial and error and heat sinks are the same thing too. At the moment I have no heat sinks large enough. I have learned some very good information about making chokes that makes it much easier to make a certain value and get it much closer to the value I need.

 

[rjenkinsgb]

I think the reference to a "real induction heater" was just a bad choice of words.

You have definitely constructed real and very functional induction heaters - impressively so, from my point of view!

The debate is only about how the FETs are driven. You know the frequency it's running at & stated it earlier, about 60 KHz.

If you drove the FET gates with an external oscillator at that frequency and added proper gate drivers, as you would use for the lower half of an H-Bridge and with non-overlap timing so each FET is allowed time to turn off before the opposite one turns, the overall efficiency should be much higher and heat dissipation much lower.

That would also avoid all the problems with the choke setup, as that should then be far less critical, if it's needed at all.


As it is, each FET only turns off _after_ the opposite one starts to turn on, so they are fighting each other and producing high current spikes, which require the choke to keep under control. There is a lot of wasted power in that configuration.


In principle, you would be converting it to a push-pull switch mode PSU like this -

https://sound-au.com/articles/smps-f12.gif

But without the transformer secondary or any feedback, as the heating load itself becomes the secondary of the coil/transformer when that is put in the power coil..


Edit - oops.. wrong thread, I just saw the tail end about the "real" comments...

 

[gary350]

I think the reference to a "real induction heater" was just a bad choice of words.

You have definitely constructed real and very functional induction heaters - impressively so, from my point of view!

The debate is only about how the FETs are driven. You know the frequency it's running at & stated it earlier, about 60 KHz.

If you drove the FET gates with an external oscillator at that frequency and added proper gate drivers, as you would use for the lower half of an H-Bridge and with non-overlap timing so each FET is allowed time to turn off before the opposite one turns, the overall efficiency should be much higher and heat dissipation much lower.

That would also avoid all the problems with the choke setup, as that should then be far less critical, if it's needed at all.


As it is, each FET only turns off _after_ the opposite one starts to turn on, so they are fighting each other and producing high current spikes, which require the choke to keep under control. There is a lot of wasted power in that configuration.


In principle, you would be converting it to a push-pull switch mode PSU like this -

https://sound-au.com/articles/smps-f12.gif

But without the transformer secondary or any feedback, as the heating load itself becomes the secondary of the coil/transformer when that is put in the power coil..


Edit - oops.. wrong thread, I just saw the tail end about the "real" comments...

I under stand that. I was thinking about waiting and using the controller on a better circuit. Maybe I should build an H circuit with the controller for 4 mosfets it is an educational learning project too. I have not read all the information about the control yet all I can remember at the moment is I can set the KHz at any setting I want. I don't remember reading about over lap yet not sure I can set that at any setting I want. You explained that so well I think I need to build a small H circuit with controller before building a larger H circuit with controller. Once I get a feel for how this controller works the larger H circuit will be easier to build. Maybe by then I can find 4 large 1GBT heat sinks.

 

[rjenkinsgb]

You can get twice the overall voltage on the power coil with a centre tap and two power FETs, as you already have.

It's a lot simpler than the full H bridge and less to go wrong.

 

[gary350]

You can get twice the overall voltage on the power coil with a centre tap and two power FETs, as you already have.

It's a lot simpler than the full H bridge and less to go wrong.

My only power supply at the moment is a car battery. Bridge rectifiers in the power supply circuit shorted out. I need some good power diodes for a bridge rectifier or a power bridge rectifier. The 250a diodes i have will work it is over kill for this tiny circuit but I can use it later on the larger H circuit up to 1500W. My power transformer is 21 vdc at 1500w. My biggest problem is LARGE heat sinks I don't have any.

 

[ChrisP58]

Why are you changing to IGBTs for this design?

Be aware that generally IGBTs have much slower switching times than mosfets do. According to the the Features list on page one of the data sheet, the CM600HA-24H is spec'd for High Frequency operation of 20-25 kHz.

 

[shortbus=]

Why are you changing to IGBTs for this design?

My guess is that he isn't aware of Iso-top style mosfets, or he doesn't have any.

 

[DerStrom8]

He may be choosing IGBTs instead of MOSFETs simply due to power dissipation issues at high currents, just like DRSSTC builders choose IGBTs over MOSFETs. The power dissipation of MOSFETs increases exponentially as current increases due to its Rdson (P = I^2 * R), whereas it only increases linearly with current due to the forward voltage drop of the diode in the IGBTs (P = I * V). For this reason, above certain D-S or C-E currents, an IGBT is a more ideal solution than a MOSFET as it will waste less energy in the form of heat.

 

[gary350]

He may be choosing IGBTs instead of MOSFETs simply due to power dissipation issues at high currents, just like DRSSTC builders choose IGBTs over MOSFETs. The power dissipation of MOSFETs increases exponentially as current increases due to its Rdson (P = I^2 * R), whereas it only increases linearly with current due to the forward voltage drop of the diode in the IGBTs (P = I * V). For this reason, above certain D-S or C-E currents, an IGBT is a more ideal solution than a MOSFET as it will waste less energy in the form of heat.

I want to build a bigger better H circuit induction heater. Several years ago someone suggested using IGBTs because they are almost bullet proof. I ordered 4 IGBTs and controller. I built a water cooled work coil. I have a water pump too and capacitors. How many years has it taken to get this far, I forgot maybe 3 or 5 years. I have moved 4 times in 5 years, moved from Arizona to TN then lived in 3 different house now we are finally in a place to stay. I don't work on this project in summer, usually Jan to April. I have diodes & caps for 5kw power supply. I need power supply heat sinks and 1GBT heat sinks. Progress is moving slow next time I ask a question how many people remember this post, I don't have my shxt together yet but I am working on it.

 

[tvtech]

Soldier on Gary. Just soldier on.
At least we know where your coming from.

 

[gary350]

I need help. Here is the circuit from the IRF2453D datasheet.

What is a high impedance floating power supply?

I assume cap across pin 1 & 2 is the PS filter cap?

None of the caps or resistors are marked datasheet refers to VB1 cap and VB2 cap etc. I set the value of cap & resistor on pin 4 to run at 100KHz.

After reading all the datasheet info I have no clue how to set cap & resistor values for pin 14, 13, 10, 9, 7, 6 ?

Power supply will be 120vac wall outlet, bridge rectifier, filter cap. = 170VDC

What about the Load coil can it be any, size, shape, diameter, length, any number of turns?

**broken link removed**

 

[rjenkinsgb]

The device has an internal zener for the pin 1 supply, so the feed to that must be from a voltage above the lockout release (12V) and anything up to the HT supply, via an appropriate resistor to set the current to 5mA.

The "VB" caps are "bootstrap" capacitors, they work in a charge pump system, charging when the upper sources go low and discharging to feed the upper gate drive sections when they switch on.

The value needed depends on the switching frequency.

This is for a different driver, but the principles should be the same:

https://www.onsemi.cn/pub/Collateral/AND9674-D.PDF

 

[gary350]

What is a floating high impedance power supply?

Does this circuit look right?

 

[rjenkinsgb]

The bootstrap caps are probably too small; another data sheet I found says they must be greater than ten times the FET gate capacitance, which is 4200pF according to the IRFP460 data sheet.

That makes the minimum 42nF so probably 47nF (or bigger) using standard values.


I explained the "high impedance" supply already - it's supposed to be a 5mA current feed.

The simplest option is a 30K resistor from +170V to pin 1.
It will be dissipating around 1W; I'd use three 10K 0.5W or 0.6W in series.

 

[gary350]

The bootstrap caps are probably too small; another data sheet I found says they must be greater than ten times the FET gate capacitance, which is 4200pF according to the IRFP460 data sheet.

That makes the minimum 42nF so probably 47nF (or bigger) using standard values.


I explained the "high impedance" supply already - it's supposed to be a 5mA current feed.

The simplest option is a 30K resistor from +170V to pin 1.
It will be dissipating around 1W; I'd use three 10K 0.5W or 0.6W in series.

Is it possible for bootstrap capacitors value to be too big? I can try 47nf how will I know if this is the perfect value? If I go bigger how will I know if the value is too large?

New circuit drawing.

 

[shortbus=]

Is it possible for bootstrap capacitors value to be too large? I can try 47nf how will I know if this is the perfect value? If I go bigger how will I know it the value is too large?

New circuit drawing.

View attachment 120478

Why are you using such a high gate resistor, 1K ! Some where around 50 ohms is normal. The 1K(or higher) would be normal for the Gate to Source resistors that you don't have.

 

[gary350]

Why are you using such a high gate resistor, 1K ! Some where around 50 ohms is normal. The 1K(or higher) would be normal for the Gate to Source resistors that you don't have.

1K is a temporary guess. Is there a way to calculate the correct resistor? My other circuit no load idle current with 220 ohms = 5a on the meter. 240 ohms = 1.5a on the meter. I think gates are still not completely turning off. 250 ohms might be a better choice to get 0a on the meter. I have not compared datasheets until now. I should probably test this circuit on 15v before trying 170v. I don't understand this circuit what limits current there is no choke. I don't see Gate current on datasheets.

 

[rjenkinsgb]

There is no steady-state gate current, there is a high gate capacitance to charge as the drive voltage changes between high and low.

The resistor is a balance between not overloading the current drive capability of whatever is controlling the gate (if it's too low) and making the switching time unreasonably long if it is too high - causing a lot of power loss as the FET is part-on for some time.

If by the "other circuit" you mean the self-oscillating one, the resistors ion that have a different function and cannot be used as a direct comparison.

The IRS2453 sheet mentions 180mA output. Without going through the sheet to find if that is peak or average, 100 ohm gate resistors should be safe.
Lower is better but I don't know what the IC can stand, I don't have time to read the data in detail.

The driver IC only appears to have a set 1uS dead-time, that's all it allows for one FET to turn off before the other is turned on.


With a circuit like that & a driver without built-in current sense, the normal method I've seen is to use a low value resistor in the connection to the source of the bridge FETs (lower centre rail in your drawing).

Then have a comparator between that and a preset fed from a low reference voltage. If the current is too high the comparator is triggered and that feeds the shutdown input to the driver.

I do not know if the shutdown on this IC works like that...


ps. If you are considering using this to drive the same coil as in the previous design, I'd advise against it. It is not suitable to drive a tuned coil.
Also remember the change in power is proportional to change in voltage squared; ten times the voltage means 100 times more power and everything needs to be built based on that, plus plenty of excess capacity in case of miscalculations.

 

[shortbus=]

You using 170V on an induction heater? Are you trying to win the "Darwin Award'?

Why are you going from one type of circuit for this to another? Don't you have Google? Google "ZVS induction heater circuit" and you will get some real circuits that work. Somethings like -

ZVS Driver

http://adammunich.com/zvs-driver/

Both of them are well known and the circuits are proven and work.

 

[gary350]

If you are considering using this to drive the same coil as in the previous design, I'd advise against it. It is not suitable to drive a tuned coil.
Also remember the change in power is proportional to change in voltage squared; ten times the voltage means 100 times more power and everything needs to be built based on that, plus plenty of excess capacity in case of miscalculations.

This is the work coil I made for a larger project a year ago but never used it. I learned from the other project if the work metal piece touches the uninsulated coil the mosfets all go up in smoke that is why I only make insulated coils now. Will this uninsulated coil be a problem or does it need to be insulated? I was planning to have water cooling on this coil but maybe not at first it should work fine for a short 10 second test. 8 turns, 2.75" long, 1.25" center to center diameter.