throbscottle
Well-Known Member
The device in question takes a DC input and provides a 3 phase output to drive a motor, in this case the one in my daughter's washing machine. One of the internal IGBT's had developed a short through to the substrate I suspect.
I present the modifications I made in order to insert a more modern device, in the hope that it will be useful to somebody.
The STK621-015 is obsolete and expensive especially when you add on the postage. Whilst searching for a replacement International Rectifier's site showed up with the device listed as an alternative package (obsolete) for the IRAMS06UP60A. I emailed to ask them if they could confirm with was a drop in replacement, got email back saying please use the 10A version, the IRAMS10UP60A. Since I was able to obtain one of these locally much cheaper than the obsolete chip, I got one (just over £6 with my brother-in-law's staff discount at the supplier!).
This is where it got really interesting. Believing I had got a drop-in replacement, I attempted to fit the new chip in place of the old one, only to find the pins didn't line up with the holes. It took a lot of head scratching and careful examination of the datasheets of the two devices before I realised the differences - especially considering the datasheet for the old one leaves a lot to be desired (and a good example of "Engrish")
1) pin positions different. Both are 23 pin devices, with some pins not present. The old device has pin 1 missing, so pin 2 is the first actual pin. The new device starts at pin 1. Functionally the pins are identical, so I just had to bend them to the side by 1 place. At the other end of the chip, the pins needed to be bent the other way, and the last 2 pins needed the opposite joggle, and one pin on the old chip doesn't exist on the new one.
2) different low emitter connection & shunt. The old chip has the 3 low emitters connected together internally, and to V- via a shunt. The emitters themselves are brought out to another pin. On the new chip the 3 emitters each have their own pin, and the user is expected to provide an external shunt.
3) fault protection different. The old chip has a MOSFET which pulls the fault pin low in the event of a fault. The new chip has a thermistor which causes the same pin to gradually go high as temperature increases.
4) bootstrap different. The old chip requires 3 diodes to be connected to the high side bases, (do you call them bases or gates on IGBT's?) from a resistor connected to Vdd. The new chip has these provided internally.
Modifications necessary.
1) bend all of the pins so they would line up with the old holes, since they are functionally identical.
2) remove the 3 bootstrap diodes from the PCB
4) I made a 3 way shunt using some 0.47 ohm resistors, to connect to the low side emitters. This was then connected to the original emitter hole in the PCB. I provided another connection to what would be the sense pin on the PCB, only to discover this was not connected, thus making the shunts redundant. So I ditched those and just connected the emitters together and into the original hole. I did this by bending the 3 pins up and out from the chip and inserting into a small bit of copperclad with 4 holes in it - hole 4 being the single wire into the board.
5) the pin on the old chip which connected the emitter end of the old chip's internal shunt doesn't exist on the new chip, and (I discovered after hacking out more rubber sealant) not connected on the PCB anyway, so I left this hole vacant.
6) I wanted the built in thermistor to provide a function similar to the original fault pin, so I made a little board with 2 resistors and an NPN transistor on it. I connected both end pins (ie 22 and 23) of the chip into this so it could provide the correct joggle for them. The transistor switches on when the input is about 3 volts - corresponding to about 100 degrees C at the thermistor. Can't actually test the behaviour under fault conditions, but at least it's functionally similar.
Relying on the chip itself to provide overcurrent protection as advertised. Put it all back together, let the machine spin for 10 minutes, felt the chip - hardly warm at all. Must be Good.
I present the modifications I made in order to insert a more modern device, in the hope that it will be useful to somebody.
The STK621-015 is obsolete and expensive especially when you add on the postage. Whilst searching for a replacement International Rectifier's site showed up with the device listed as an alternative package (obsolete) for the IRAMS06UP60A. I emailed to ask them if they could confirm with was a drop in replacement, got email back saying please use the 10A version, the IRAMS10UP60A. Since I was able to obtain one of these locally much cheaper than the obsolete chip, I got one (just over £6 with my brother-in-law's staff discount at the supplier!).
This is where it got really interesting. Believing I had got a drop-in replacement, I attempted to fit the new chip in place of the old one, only to find the pins didn't line up with the holes. It took a lot of head scratching and careful examination of the datasheets of the two devices before I realised the differences - especially considering the datasheet for the old one leaves a lot to be desired (and a good example of "Engrish")
1) pin positions different. Both are 23 pin devices, with some pins not present. The old device has pin 1 missing, so pin 2 is the first actual pin. The new device starts at pin 1. Functionally the pins are identical, so I just had to bend them to the side by 1 place. At the other end of the chip, the pins needed to be bent the other way, and the last 2 pins needed the opposite joggle, and one pin on the old chip doesn't exist on the new one.
2) different low emitter connection & shunt. The old chip has the 3 low emitters connected together internally, and to V- via a shunt. The emitters themselves are brought out to another pin. On the new chip the 3 emitters each have their own pin, and the user is expected to provide an external shunt.
3) fault protection different. The old chip has a MOSFET which pulls the fault pin low in the event of a fault. The new chip has a thermistor which causes the same pin to gradually go high as temperature increases.
4) bootstrap different. The old chip requires 3 diodes to be connected to the high side bases, (do you call them bases or gates on IGBT's?) from a resistor connected to Vdd. The new chip has these provided internally.
Modifications necessary.
1) bend all of the pins so they would line up with the old holes, since they are functionally identical.
2) remove the 3 bootstrap diodes from the PCB
4) I made a 3 way shunt using some 0.47 ohm resistors, to connect to the low side emitters. This was then connected to the original emitter hole in the PCB. I provided another connection to what would be the sense pin on the PCB, only to discover this was not connected, thus making the shunts redundant. So I ditched those and just connected the emitters together and into the original hole. I did this by bending the 3 pins up and out from the chip and inserting into a small bit of copperclad with 4 holes in it - hole 4 being the single wire into the board.
5) the pin on the old chip which connected the emitter end of the old chip's internal shunt doesn't exist on the new chip, and (I discovered after hacking out more rubber sealant) not connected on the PCB anyway, so I left this hole vacant.
6) I wanted the built in thermistor to provide a function similar to the original fault pin, so I made a little board with 2 resistors and an NPN transistor on it. I connected both end pins (ie 22 and 23) of the chip into this so it could provide the correct joggle for them. The transistor switches on when the input is about 3 volts - corresponding to about 100 degrees C at the thermistor. Can't actually test the behaviour under fault conditions, but at least it's functionally similar.
Relying on the chip itself to provide overcurrent protection as advertised. Put it all back together, let the machine spin for 10 minutes, felt the chip - hardly warm at all. Must be Good.
