They discharge C35 (100nF) to 0 volts by configuring RA0 as an output and pulling it down for the right amount of time. Then, they configure RA0 as an input and leave the capacitor to charge through R48 (220k but acting mostly as a current source).
IC2B and IC2D act as comparators and allow the PIC to know how much time C35 takes to charge from 0V up to the voltage corresponding to each temperature level. That's all.
IC2B and IC2D act as comparators and allow the PIC to know how much time C35 takes to charge from 0V up to the voltage corresponding to each temperature level. That's all.
MOSFET driver buffer
I have revised the description of the MOSFET driver buffer by moving the resistor from the gate of the N-channel MOSFET to its source.
ImageShack
I have revised the description of the MOSFET driver buffer by moving the resistor from the gate of the N-channel MOSFET to its source.
An externally hosted image should be here but it was not working when we last tested it.
ImageShack
Re: MOSFET driver buffer
BTW. Magnetic EMI-probe can be very handy when tracking this kind of like problems. Linear website has somewhere in appnotes exellent EMI-probe constrution shown, but I cant find it right now as Linears website is sloooow once again.
High-voltage mosfets(esp. older ones like IRF450) are pain in **** to drive nicely, one reason why I recommend forward instead of half-bridge... Or go for bipolars as Eva does
Low impedance low-state drive is also good for preventing capacitive coupling shoot-trough at turn-on of opposite switch. Gate-drain capacitance kicks gate voltage over threshold and very short duration shoot-trough occurs between switches. These can be hard to see on scope, occuring easily on 100Mhz region.subwo1 said:
BTW. Magnetic EMI-probe can be very handy when tracking this kind of like problems. Linear website has somewhere in appnotes exellent EMI-probe constrution shown, but I cant find it right now as Linears website is sloooow once again.
High-voltage mosfets(esp. older ones like IRF450) are pain in **** to drive nicely, one reason why I recommend forward instead of half-bridge... Or go for bipolars as Eva does
Indeed my first SMPS project had 5 IRFP460 of the good old ones in it, those with 4200pf input capacitance and 350pF reverse transfer capacitance for a bare 0.27ohm Rds-on It also had a BYV29-400 buck diode with late 1980's manufacturing code, over which I was hard switching 
That project got me very disappointed about high voltage MOSFETs. I have never used them again. Also, bipolars produce nice low conduction losses when properly driven (0.1 ohm "rds-on" at 10A in a 400V TO-220 device) and 50ns fall times may be achieved for that current level, but that's not easy to reach either.
I think that the easiest to drive devices currently available are by far IGBTs. They feature low input capacitances below 1nF and reverse transfer capacitances as low as 10pF. That's what I strongly recommend for everyone into DIY SMPS.
BTW: I'm considering paralelling a fast IGBT of the ones featuring fall times around 80ns at the expense of an apparent "Rds-on" in the 0.25 ohm range with a slow one of those featuring less than 0.08ohm apparent Rds-on at the expense of fall times around 800ns (both in TO-220 case). The trick will be to turn on both at the same time but turn off the slow IGBT 2us before the fast one, effectively forcing the fast device to do the switching and the slow one to do the conduction. What do you think about that, mzzj? The "fast" device may be also something like SPP20N50S5.
It also had a BYV29-400 buck diode with late 1980's manufacturing code, over which I was hard switching That project got me very disappointed about high voltage MOSFETs. I have never used them again. Also, bipolars produce nice low conduction losses when properly driven (0.1 ohm "rds-on" at 10A in a 400V TO-220 device) and 50ns fall times may be achieved for that current level, but that's not easy to reach either.
I think that the easiest to drive devices currently available are by far IGBTs. They feature low input capacitances below 1nF and reverse transfer capacitances as low as 10pF. That's what I strongly recommend for everyone into DIY SMPS.
BTW: I'm considering paralelling a fast IGBT of the ones featuring fall times around 80ns at the expense of an apparent "Rds-on" in the 0.25 ohm range with a slow one of those featuring less than 0.08ohm apparent Rds-on at the expense of fall times around 800ns (both in TO-220 case). The trick will be to turn on both at the same time but turn off the slow IGBT 2us before the fast one, effectively forcing the fast device to do the switching and the slow one to do the conduction. What do you think about that, mzzj? The "fast" device may be also something like SPP20N50S5.
Yeah, I have wasted my time with half-bridge IRFP450's once. Couldnt turn on that damn thing without triggering opposite mosfet also on. Modern mosfets are a lot better luckily. 600v 70mohm device with Crss=50pF sounds way better than IRFP450.Eva said:Indeed my first SMPS project had 5 IRFP460 of the good old ones in it, those with 4200pf input capacitance and 350pF reverse transfer capacitance for a bare 0.27ohm Rds-on
That project got me very disappointed about high voltage MOSFETs. I have never used them again. Also, bipolars produce nice low conduction losses when properly driven (0.1 ohm "rds-on" at 10A in a 400V TO-220 device) and 50ns fall times may be achieved for that current level, but that's not easy to reach either.
I think that the easiest to drive devices currently available are by far IGBTs. They feature low input capacitances below 1nF and reverse transfer capacitances as low as 10pF. That's what I strongly recommend for everyone into DIY SMPS.
BTW: I'm considering paralelling a fast IGBT of the ones featuring fall times around 80ns at the expense of an apparent "Rds-on" in the 0.25 ohm range with a slow one of those featuring less than 0.08ohm apparent Rds-on at the expense of fall times around 800ns (both in TO-220 case). The trick will be to turn on both at the same time but turn off the slow IGBT 2us before the fast one, effectively forcing the fast device to do the switching and the slow one to do the conduction. What do you think about that, mzzj? The "fast" device may be also something like SPP20N50S5.
And APT60N60B specs sheet claim amazing Crss=5pF when VDS>75v
Parallei IGBT's? sounds good for effiency but I really dont like extra complexity. Maybe ok somewhere like pfc-boost switch but full-bridge..nooo way
I would go for big $$$-mosfet if last percents in effiency count.
Re: Re: MOSFET driver buffer
There would be lots of EMI indeed, with 100mhz power spikes.
It would be interesting to look into the magnetic probe. Alternately, a short piece of wire connected to a probe combined with viewing a TV set on a low VHF channel could to do well.
I like ZVS because the switching times can be much slower and still get low or no switching losses. The intrinsic capacitances become much less relevant
mzzj said:
There would be lots of EMI indeed, with 100mhz power spikes.
It would be interesting to look into the magnetic probe. Alternately, a short piece of wire connected to a probe combined with viewing a TV set on a low VHF channel could to do well.
High-voltage mosfets(esp. older ones like IRF450) are pain in **** to drive nicely, one reason why I recommend forward instead of half-bridge... Or go for bipolars as Eva does
I like ZVS because the switching times can be much slower and still get low or no switching losses. The intrinsic capacitances become much less relevant
Hi Folks,
Here is a link to Quasi-Resonant half Bridge used in QSC amps for offline purposes...
http://www.qscaudio.com/support/library/schems/plx3402.zip
regards,
K a n w a r
Here is a link to Quasi-Resonant half Bridge used in QSC amps for offline purposes...
http://www.qscaudio.com/support/library/schems/plx3402.zip
regards,
K a n w a r
Re: Re: Re: MOSFET driver buffer
In case link outdates, try searching "an70" and look for "A Monolithic Switching Regulator with 100mV Output Noise"
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1130,P1535,D4159
Starting somewhere page n. 55 is small EMI-probe. Small size makes it possible to track down even invidual high-dI/dt traces on board, crappy recovery behaviour etc.
Btw toroidal transformer construction shown on page 52...what a hell they were thinking?
subwo1 said:
It would be interesting to look into the magnetic probe. Alternately, a short piece of wire connected to a probe combined with viewing a TV set on a low VHF channel could to do well.
I like ZVS because the switching times can be much slower and still get low or no switching losses. The intrinsic capacitances become much less relevant
In case link outdates, try searching "an70" and look for "A Monolithic Switching Regulator with 100mV Output Noise"
http://www.linear.com/pc/downloadDocument.do?navId=H0,C1,C1003,C1040,C1130,P1535,D4159
Starting somewhere page n. 55 is small EMI-probe. Small size makes it possible to track down even invidual high-dI/dt traces on board, crappy recovery behaviour etc.
Btw toroidal transformer construction shown on page 52...what a hell they were thinking?
Re: Re: Re: Re: MOSFET driver buffer
I am glad it finally came through since PDF causes my computer trouble under even good circumstances. The document has some really useful information in it. I think I will find it interesting to simulate the LT1533 in LTspice. Gotta start working on one of those probes.
It doesn't look like the lowest leakage inductance was the objective.
mzzj said:
I am glad it finally came through since PDF causes my computer trouble under even good circumstances. The document has some really useful information in it. I think I will find it interesting to simulate the LT1533 in LTspice. Gotta start working on one of those probes.
Btw toroidal transformer construction shown on page 52...what a hell they were thinking?
It doesn't look like the lowest leakage inductance was the objective.
Stone-Age
Eva, Sobwo1 and mzzj,
You three are making me feel as if I'm stuck in the stoneage
with my rails of old Motorola MTW20N50Es (1998 date codes) with their 0.24W Rds(on). I had planned on using some of them in my revived half-bridge SMPS. When I get the time, and $$, I will order some really good IGBTs, perhaps from IR. But I will stick with my batch of old 1990 date coded IR2110s.
Steve
Eva, Sobwo1 and mzzj,
You three are making me feel as if I'm stuck in the stoneage
with my rails of old Motorola MTW20N50Es (1998 date codes) with their 0.24W Rds(on). I had planned on using some of them in my revived half-bridge SMPS. When I get the time, and $$, I will order some really good IGBTs, perhaps from IR. But I will stick with my batch of old 1990 date coded IR2110s. Steve
Go on, MTW20N50E are somewhat better than my old IRFP460 (probably from early 1990s), MTWs have three times lower reverse transfer capacitance and lower Rds-on. Also, currently I only have 4 of these devices, but if I had rails full of them I would consider using them as a challenge. You can paralell banks of two or three and use high current buffers to drive the gates (see ZTX851 and ZTX951, I was using these as emitter followers driven from a CD4050 with all 6 gates paralelled).
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.