An extremely informative and thorough explanation of the function of the MOSFET in this circuit, and in general, as provided by my wonderful father who helped me select parts for this repair.
1. I noted the Driver chip, the E and F are significant. The E part is speced at 10v max, while the F is good to 16v. Looking deeper though at the spec sheet, there are the ‘suggested operating range’. The lower-voltage E part has a MAX voltage of 12v. This should be OK. (Plus, the E parts have overload protection while the F do not. So – ya, stick with the E part)
https://octopart.com/search?q=S-8520F33mc&start=0https://octopart.com/search?q=S-8520E33mc&start=0 2. FET: Attached is original datasheet. It is a not great part, old.
· Mostly notice the Rd-ON of around 1 ohm! Causes heat, which is always bad. But also provided some level of current limit into the inductor. I have selected a like part you can use that should work well.
· Not also the driver chip in the spec sheet on Page 9: “EXT pin output current”. FETs can be looked at like a capacitor, that is the Ciss value the other guy mentioned (Good advice overall from him, BTW). Do when we turn on and off, we need to charge and then discharge that capacity. The Output current spec is part of this. We need to make sure we can fully turn on the FET and OFF while ti is switching. This part runs at 300Khz, which is getting up there. And the driver chip has only a few mA current available, so we want to watch things and not get to large of what is called the ‘Gate Capacitance’ on the FET. Critical items to look at:
i. Vgd Max (60v -- Higher = better)
ii. Rdon (1 Ohm = Typicaly lower=better)
iii. Id Max (1A, 4A peek = Higher = better)
iv. Ciss (10pF – lower = better)
v. Vg-th (2v, need to make sure it matches the driver chip output).
So, when a FET is turned on, we apply a voltage separation between Gate and Source (Know as Vgs). As the voltage raises, it will start turning on. How hard it turns on is a function of current through the FET (Id), as well as Vgs (and temperature of course). What we want to make sure is we can get the FET turned on rather quickly (use it as a Switch, not a liner amp) to reduce the amount of heat generated during switching (Known as Switching Loss). To do that we have to look at the driver chip current capability vs. the input gate capacitance (Ciss). These two form in effect an RC time constantan and define largely how quickly the FET will turn on or off. Increate the driver capability, and we charge the Gate Capacitance faster – and we turn on faster. Lower the gate capacitance, and the same. But reduce the drive current, and/or increase gate capacitance, and switching time will increase: Meaning we spend more time in the eval ‘analog’ region where the FET is sort of turned on, and as a result we end up dissipating heat in the device.
Once the device is turned on, Rdon comes into play. It will direct the amount of head loss vs. current through the device. Notice Rdon is impacted by the final Vgs we end up getting. Use a higher Vgs results in the lower Rdon, to a limit.
So that is some of the things to consider. The charge on the lower-right hand side of the 2nd page of the FET datasheet is a good one. Notice how it is showing in effect Vgs-th (Threshold) of around 2.2v And that is really does not get full on until Vgs reaches around 4 or 6v. Now, if we look back at the driver chip datasheet table 5 on page 9, we can see the EXT pin seems to run around 0.4v above Ground when on. If we have a 9V battery(which is more likely around 7v in the real world) we can expect the FET to be around 0.85 Ohms Rd-On. The final critical think to confirm is the Threshold voltage. Vgs-th, this indicated when the FET starts turning on. The original FET seems to have a Vgs-th of around 2.2v per the charts. We can select a part which is lower Vgs-th, but we need to make sure the driver ship will allow Vgs to get below the threshold and let the FET turn off. Again, going back to the driver chip datasheet I see Vext being specced at Vin-0.4v (again in table 5). So, that is telling me the driver chip will typical switch Vext from Vin-0.4v to 0.4v, resuting in a Vgs of 0.4v while off, and Vbat-0.4v (Approach 6.6v) when ‘on’. OK, Vgs-off of 0.4v should be simple to work with, but we need to take care as there are some special ‘logic level’ FETs which have a Vgs-th of 1v or so, and they might not fully turn off with that 0.4v.
The final issue the the packaging. Sanyo calls this a “2062A” package, which refers not to any standard format, but a specific SANYO drawing. Here I can only say experience can solve this. This package is more commonly known as a TO-89-3
OK enough of all that. I would say try this part – it would work just fine:
https://www.mouser.com/ProductDetail/ON-Semiconductor/PCP1302-TD-H?qs=xGcJQ%252bnsJwtl9GRuovbZ6g%3d%3d Ciss is notable higher at 262pF, however looking at the driver table 13 on page 34 I can see a few example transistors being used in their demos with even higher Ciss, so I expect we will be OK. Doing some VERY rough check cals: Charge on a cap à V=Q/c (Q = charge on the cap in coulombs. ) Q = I*t (Current * time) -à V = (I * t) / c
So, if we cals the time it takes a 160pF cap to reach say Vgs = 4v (make sure the FET is well turned on) when being supplied say 2mA (Small value from driver table, and a little lower --- the really CORRECT way to do this is calculate the resistance of the driver chip + the gate resistance, but some of those values are missing – so we will just use this as a quick check to see if we are totally in trouble). OK, 160pF for the original FET takes 320nS, while the 262pF part takes 525nS. On and off, the old part takes around 600nS, the new one 1uS. 300Khz switching (the frequency the driver chip is running at) takes 3.3uS -- hum, so we are taking a bit more time to turn on and off and it is a not-insignificant part of the total switching time. But here is my assessment: The extra loss fo the switching time might well be offset by the better Rdon value while not switching.
Bottom line: By feeling is you will be OK with this replacement part. Give it a try and if you really want try scoping the Vgs as it switched on, looking at Raise Time and Fall Time vs. the overall switching time period. (Might be hard, you need a FAST scope, likely a storage one, and also need to have really low capacitance probes… But give it a try.
OK I know this was a LOT – but though would take the time to pass on some of the thinking behind it vs just say: Buy this:
https://www.mouser.com/ProductDetail/ON-Semiconductor/PCP1302-TD-H?qs=xGcJQ%252bnsJwtl9GRuovbZ6g%3d%3d
