Help with power electronics

[MagicSmoker]

I had a feeling this was a completely irrational thread, but, well... sometimes irrational is fun.

One approach would be a buck current-fed full bridge with interleaved buck switches (and a bridge rectifier using 1.2kV SiC diodes on the secondary). Each leg of the full bridge needs to run at slightly more than 50% duty cycle but there are even a couple of chips out there that can do this (e.g. LM5041).

A more conventional approach would be to use a fixed isolation stage (using, e.g., several two-switch forward or one or more full bridge converters) that also steps up the voltage then use a buck to step down voltage as needed. Both the buck switch and freewheeling diode need to be rated for the max output current and voltage (e.g. - 1.2kV and 200A) which makes them costly, but this isn't out of the realm of feasibility using, e.g., IGBT or SiC MOSFET modules. Oh, and don't forget that the inductor needs to have enough inductance to not allow the ripple current ratio (i.e. - "r") to get too high at the high voltage/low current end of the range, yet also not saturate at the low voltage/high current end of the range. Also note that you will either have to employ slew rate control of the buck switch and or quasi-resonant switching to keep EMI and dV/dt related problems under control.

All that said, I don't see why there is a need for such a wide range of output voltage/current for the stated objective of supplying an induction heater in the first place; don't you have considerable freedom to set the characteristic impedance of the tank?



 

[MarkoAnte]

don't you have considerable freedom to set the characteristic impedance of the tank?

Well not really. At least I don't think. Here is what I observed with my current set up.

You have a given object to heat in a given frequency to achieve a set heat penetration depth. (I'm looking at 1Mhz and above - if the planets aline). So with that your locked into a specific inductor shape. Then to achieve your resonant frequency you need a calculated capacitance. Then you can use a impedance matching transformer but there are limitation on what you can do with it. If you over do it then after you insert some material with iron you get no power out of the thing as the oscillations get dampend. But if you make it so that you get eg 15 KW of power on the inductor when the iron material is cold , when on max power, and then you reach Curie temperature you have to reduce the dc voltage by a lot or else boom-smoky goes the transistor. Then after the big drop in load, if you want to continue heating you have to start increasing the DC voltage. This goes if you want to work at a constant (max) current.     

Tho I have not experienced it its, supposedly heating of crystals (1-2 Mhz) is even more erratic. Now only done with tube oscillators.

[T3sl4co1l]

@ T3sl4co1l
I don't understand what you are trying to say.

fixed tuned ... direct drive the inductor ....  tuning kind of out of the picture to start with....

All the heaters I was followed the resonant frequency some more real time then others. Then the power was adjusted with frequency or with pwm of the main frequancy.

Hey, voltage is voltage and current is current.

You have a transformer in your circuit, right?

Tim

[oldway]

@DanielS: Switching DC current is a poor solution...It seems that you don't have experience with DC power electronics. When you will see what happen when you try to open 40A 166V with relais or contactors, you will understand that it is not a good solution.

I am absolutely sure that the solution I proposed works well, that this solution is no "technical adventure" and that I am able to project this power supply with 100% confidence.

SCR solution will be far more reliable than all SMPS solutions you can find (if possible !)

Your problem is that you don't know nothing about SCR's technology !

[MarkoAnte]

@oldway Would it be possible, to use a "normal" transformer with taps and use full wave SCR controlled rectifiers to get the same 6 wave rectification. (Or maybe upgrade to 12 wave rectification with full wave rectifiers on a 6 phase transformer.) I know that means double the SCR and double the SCR loose. But I may make it back with a reduce size/weight/price of a normal transformer.

[MarkoAnte]

I made a quick cad drawing to clarify what we are talking about.

[DanielS]

@DanielS: Switching DC current is a poor solution...It seems that you don't have experience with DC power electronics. When you will see what happen when you try to open 40A 166V with relais or contactors, you will understand that it is not a good solution.

Looks like you missed or overlooked the memo about turning output power off before changing the module configuration. There is no current during the tap change if you turn off power by shutting down the PWM chips first.

Since OP's application is induction heating, the voltage and current range is determined by the coil he wants to use and does not change by much during operation. Unless you are advocating that OP changes coils in a live circuit, then he already needs to shut down power before making any significant changes anyway.

[MarkoAnte]

@DanielS:

The thing is that by adjusting the DC link voltage you adjust the output power of the heater. And as I described before the regulation must be able to handle the goings on at the Curie temperature, meaning that a quick DC link voltage change is needed. If the application was not a HF induction heater but a universal "smiting" heater than it would be feasible to make it the way you describe as other power management approaches could be used and the DC voltage level only used for fine tuning. But for universal low frequency heaters there are batter ways of going about things. With fewer components and less cost.


On a side note, if anybody is interested:
"quick" - depends on the type of work peace your heating as it can happen that the skin of the work peace has already surpass the Curie temperature but the bulk of the mass has not. So the transition is not that quick. But if the work peace is e.g. a pipe that has to be hard soldered then the change will be quick.

[MarkoAnte]

@blueskull
I was looking at something like that but I have a 3 concerns.
1) I will have to make some kind of drivers for all the transistors.
2) I will have to do some voodoo when I will start hard switching something and I will get elector-magnetic garbage spewing all over the place - I had this problem with my prototype buck converter (but it was a simple synchronous buck converter, so maybe I will not have so many problems with a more advanced design or maybe even more).
3) The cost. Its why I like the thyristor idea. With 3*6 thyristors + a transformer I already get a usable DC voltage + a capacitor. But with a boost buck converter I first require a 3 phase rectifier then for 1 phase 4 switches. And the price for 1 SCR <<<< 1 1.7KV SiCMOSFET. And the thyristors can be fired with a simple impulse transformer while the MOSFETs and IGBTs can't. I have yet to inquire for a transformer price but the power chokes required in a boost buck converter aren't free also.

Tho the buck boost option may have a smaller weight and space requirement. (But a 20KW HF induction heater is not something you pack with your phasenprufer in your toolbox nether.) The ripple is going to be significantly better with a buck boost option tho. I think.       

[oldway]

@oldway Would it be possible, to use a "normal" transformer with taps and use full wave SCR controlled rectifiers to get the same 6 wave rectification. (Or maybe upgrade to 12 wave rectification with full wave rectifiers on a 6 phase transformer.) I know that means double the SCR and double the SCR loose. But I may make it back with a reduce size/weight/price of a normal transformer.

Yes, you can use full bridge SCR's rectifiers and 3 phase transformer, but, as you said, power losses on SCR's are twice higher....this can be a problem of efficiency at low voltage.
The two drawings are allright...

@Daniels: with 40A V to 166V ajustable units, look for the complexity of the comutations you will have to do.
- from 100 to 166V, you will need 5 converters in //
- from 166V to 332V, you will need 2 units in serie of 3 converters in //
- from 332V to 498V, you will need 3 units in serie of 2 converters in //
- from 498V to 664V, you will need 4 units in serie of 1 converter
- from 664V to 830V, you will need 5 units in serie of 1 converter
- from 830V to 996V, you will need 6 units in serie of 1 converter
CRAZY !
And the converters must be able to work with output voltage up to 1000V higher than ground...
High isolation is required.

@blueskull:
Smaller size --> better EMI performance. Not sure at all...you will have to make an ajustable power supply from 100 to 1KV and able to feed 200A at 100V and 20A at 1KV.
You probably will have to use a lot of units in //.
Dimensions seems not to be a concern.

SCR --> poor THD and PF --> your utility company will not be happy.: not right because;
- you will need a bridge rectifier to feed your converters and you will also have poor THD and high peak currrent with high rms value...your utility company will also not be happy
- full voltage range is divided in 3 subranges so phase delay in converters is limited and THD and PF stays acceptable.
- SCR technology is a well know and  reliable technology, with few components and none of them are working in critical situation.

[moffy]

Have to agree with oldway, having built a 1500v/130A variable PSU and a 100v/1500A variable PSU(3 sets of taps and 3 sets of 6 SCRS), a transformer with taps and SCRs is the best way to go.