Induction Heater project / Projects / Forums

Forums

4hv.org :: Forums :: Projects

« Previous topic | Next topic »

Induction Heater project

previous 1 2 3 4 next

Move Thread

LAN_403

Shrad

Thu Mar 05 2015, 01:14PM

Registered Member #3215 Joined: Sun Sept 19 2010, 08:42PM
Location:
Posts: 780

mister_rf wrote ...

Thanks for the tips. Over the past few years I have done some research and a lot of reading on this subject and now I have prepared elements for testing hybrid gate drive prototypes. By the way, you're one of the people whose hard work and detailed experiments in this area leads me to start such a project. Thank you again.

For example I have considered to test the HCPL316J gate drive optocoupler, M57962L / VLA542-01R and based on Wolframâ€™s project, the Silabs SI8234/SI8244 drivers.

(@Wolfram, Iâ€™m still sneak a peek on your experiments, see the cloned auto-range frequency meters prepared for

)

have you considered optically triggered switches? I read somewhere that they had pretty switching times...

Wolfram

Thu Mar 05 2015, 02:20PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Nice frequency counter build. Glad to see that my thread gave you some ideas.

I've found a bug in the rounding code of the frequency counter. It's not very critical, but the problem is that the rounding doesn't work. 39.99kHz will be displayed as 39k9 and not 40k0 as it should.

I fixed it by replacing the following part:

round     MOVLW   digs+D'11'    ; point to first (10MHz) digit
          MOVWF   FSR
roun0     MOVF    0,W           ; search for digit > 0
          BTFSS   STATUS,Z
          GOTO    roun1
          MOVF    FSR,W
          XORLW   digs+4        ; at digit 4
          BTFSC   STATUS,Z
          GOTO    roun1
          DECF    FSR,F
          GOTO    roun0
roun1     MOVLW   3
          GOTO    $+DSPLY       ; skip for DL1414T:
          MOVLW   4
          SUBWF   FSR,F         ; point to last digit - 1
          MOVLW   5             ; add 1/2 last digit
roun2     ADDWF   0,W
          MOVLW   D'10'         ; decimal carry?
          SUBWF   0,W
          BTFSS   STATUS,C
          RETLW   0             ; no: done
          MOVLW   D'10'
          SUBWF   0,F
          MOVLW   1             ; carry
          INCF    FSR,F
          GOTO    roun2

I replaced it with the following:

round     MOVLW   digs+D'11'    ; point to first (10MHz) digit
          MOVWF   FSR
roun0     MOVF    0,W           ; search for digit > 0
          BTFSS   STATUS,Z
          GOTO    roun1
          MOVF    FSR,W
          XORLW   digs+4        ; at digit 4
          BTFSC   STATUS,Z
          GOTO    roun1
          DECF    FSR,F
          GOTO    roun0
roun1     MOVLW   3
          GOTO    $+DSPLY       ; skip for DL1414T:
          MOVLW   4
          SUBWF   FSR,F         ; point to last digit - 1
roun2     MOVF    0,W		;check digit
          INCF    FSR,F
          addlw   .255 - .4	;If digit after last digit is 5 or more
          skpnc
roun3     incf    0,F    	;Increase last digit
          movf    0,W 
          addlw   .255 - .9
          skpc
          return
          clrf    0
          incf    FSR,F   
          goto    roun3

It's not very refined but it works fine.

I like the silabs drivers because they have simple programmable deadtime, as well as high side UVLO. They are also much faster than many of the alternatives I've seen, and very powerful. The price is also nice.

There are even more alternatives from Infineon and Rohm, and the TI ISO5500, all with desaturation detection.

mister_rf

Fri Mar 13 2015, 10:39PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

Thanks for the info, next time Iâ€™ll modify the code to improve the displayed frequency accuracy. I had a nice surprise to use some old stuff lying around, and now I see still they are still producing intelligent displays similar to the old ones. So, back to the drawing boardâ€¦.

Anyway, as the typical situation for me, I was stuck in the mechanical details implementation, this time the water cooling unit for the coil.

As related to the drivers, I was curios and I have bought the Silab Eval Board SI823X , and now Iâ€™m ready to do some tests.
For this setup Iâ€™m prepared to test the C2M0080120D SiC MOSFETs.

1426285887 4465 FT1630 Si8234 Driver Low Res

mister_rf

Tue May 12 2015, 07:37PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

I have decided to split the project and to use dedicated PCB boards for.
Small steps towards the final tests.

The new PCBs frequency counters ready.
This includes also a JFET buffer option.

And the PCB MOSFET driver version 1.0 ready for tests.

Wolfram

Wed May 13 2015, 01:34PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

Very nice! It looks like a solid design, and I particularly like the the SiC MOSFETs. I've always wanted to try those myself, and I'm really looking forward to seeing your results with them. The only thing I'd be slightly worried about is the clearance to the ground plane in the bridge board. It's probably fine, but if you're doing a second revision I would recommend increasing the clearance.

With that bridge, you should be able to do many kilowatts at 500 + kHz. You could really make use of those capacitors, which I'm sure can do 1000 A RMS each. One of those caps and a proper work coil, and I would guess you could deliver 10 kW to a workpiece at 300 kHz.

Nice progress, keep us updated.

mister_rf

Wed May 13 2015, 07:27PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

Yes, you are right; the actual PCB's clearance is a real problem, so I have decided to use those boards only for testing at low power supply. That was a rookie mistake I have made on leaving the polygon â€œisolateâ€ option on default (EAGLE Light Edition). I was in a hurry to send those boards to combine shipping with another low voltage project. Anyway, that was cheap mistake, and sooner I will put the new order for the version 4.0.

Wolfram

Fri May 15 2015, 04:27PM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

mister_rf wrote ...

Yes, you are right; the actual PCB's clearance is a real problem, so I have decided to use those boards only for testing at low power supply. That was a rookie mistake I have made on leaving the polygon â€œisolateâ€ option on default (EAGLE Light Edition). I was in a hurry to send those boards to combine shipping with another low voltage project. Anyway, that was cheap mistake, and sooner I will put the new order for the version 4.0.

Yeah, mistakes are very easy to do, you alredy found some in my design

. I sometimes find it useful to use the bRestrict and tRestrict layers to keep ground pours away from critical areas also. You can draw lines, polygons and circles to define the areas where the polygon pour is left out.

I forgot to ask earlier, what are the dimensions of your coupling transformer toroids, and what material are they?

mister_rf

Fri May 15 2015, 08:55PM

Registered Member #4465 Joined: Wed Apr 18 2012, 08:37AM
Location: Bucharest, Romania
Posts: 145

My initial idea was to use the EPCOS B64290L0699X087 63MM, N87 ferrite cores, but recently I have also available for the project some large ferrite toroid made by FERROXCUBE , model T102/66/25-3C90

Experimentonomen

Sat May 16 2015, 10:45AM

Registered Member #941 Joined: Sun Aug 05 2007, 10:09AM
Location: in a swedish junk pile
Posts: 497

Big mofos, but not as big as the ones kizmo gotz :P

Wolfram

Mon May 18 2015, 08:24AM

Registered Member #33 Joined: Sat Feb 04 2006, 01:31PM
Location: Norway
Posts: 971

mister_rf wrote ...

My initial idea was to use the EPCOS B64290L0699X087 63MM, N87 ferrite cores, but recently I have also available for the project some large ferrite toroid made by FERROXCUBE , model T102/66/25-3C90

Nice cores. With the setup in your first post with only two of the 63mm cores, you're probably a bit high on the flux density. With two, you have a total core cross section area of 5.35 cm^2 . With 300V peak on the full bridge and 20 primary turns, you have 15 V/turn, and at 70kHz a half period is 7.1 microseconds, so you have about 200 mT peak flux density. At 70l kHz, the core losses will be aroun 30 watts, so they might get a bit toasty. Adding a third core would make it significantly better.

previous 1 2 3 4 next

Moderator(s): Chris Russell, Noelle, Alex, Tesladownunder, Dave Marshall, Dave Billington, Bjørn, Steve Conner, Wolfram, Kizmo, Mads Barnkob