another small breakthrough on our NERD technology.

[TinselKoala]

@picowatt
The max Id of the IRFPG50 drops to 3.9 Amps at 100 degrees C, a normal operating temp _when properly heatsunk_.

Below, in the single mosfet version, one may view the heatsink of the lone mosfet. It's still there in the 5 transistor version but can't be seen as well.

(ETA: I see that Fuzzy has posted an even better picture of this. Isn't that a cute little heatsink for a mosfet that can dissipate 190 Watts? They got much more realistic on the Gang of Four.... I wonder why.)

I have both failure modes in the 830a, one each. One is shorted all around, and the other is open drain-source and shorted gate-source and of course won't switch. The open came when I had the thermal runaway before I was well-heatsunk, IIRC, and the other came from a short ... I think I applied full battery voltage to the gate or something, I'm not really sure. I've thrown them away already so I can't do any further checking.

You are "preaching to the choir" about failures in mosfets... believe me, I understand how mosfets can fail, having literally blown apart a double handful while developing the TinselKoil. Usually I could find most of the pieces ..... but I'll never know what the landlord must have thought when those "gunshot" like sounds were coming from the basement....."BANG dammit......"

[picowatt]

Rosemary,

I see that in your text describing Test 1 that you do state that you set the FG output to its full negative offset in order to prevent Q1 from turning on.  However, the FIG 3 for Test 1 does not indicate this.

Do you agree that the FIG 3 gate drive signal indicates +12.5 volts being applied to Q1 during the FG's positive portion of the duty cycle and that at that level, Q1 must turn on?

Assuming that the FG output did indeed not go positive during Test 1, is it possible that the FIG 3 is not the correct scope shot for that test?  Possiby the wrong save was pulled from the scope's memory...?

I'll look at the other tests, just trying to get past this discrepancy...

PW

[TinselKoala]

@picowatt:

Here's Fig 6 from Test 3.

It looks to me like there is some current flowing during the non-oscillating portions. It's really too bad we don't have the common drain trace, which would give a lot more information. The CVR trace is set to 2 v/div and is showing about half a minor division up from zero during the non-osc phase, so that would be around 0.2 volts positive drop.

Recalling that the CVR is 0.25 ohms, that means around 800 mA flowing, I suppose.

Certainly the gate drive signal does not correspond to the description in the paper or Rosemary's present description, going as it does from zero (with oscillations) to around 10 Volts.

Figure 7 again appears to show no voltage drop in the CVR trace, yet the gate signal, once again, is from zero to about 10 volts positive. And of course the claim is that the load heated strongly in this mode.

But that's not what was shown LIVE in the video. These are stored traces obtained at different times with cryptic filenames. What is the chance that the descriptions that we are given do not match the traces we are shown? Considering the history, I mean.....

[picowatt]

TK,

Did not mean to "preach"

Hey, were you not dealing with an H bridge on that coil?  I would think the most comon cause of failure would have been simultaneous turn on of the "upper and lower", and then BANG indeed...

Typically, in electronics, were to need goggles for flying wire lead ends when clipped with diagonals, not pieces of package material flying about!

It's all good...

PW

PW

[Rosemary Ainslie]

Rosemary,

I see that in your text describing Test 1 that you do state that you set the FG output to its full negative offset in order to prevent Q1 from turning on.  However, the FIG 3 for Test 1 does not indicate this.

Do you agree that the FIG 3 gate drive signal indicates +12.5 volts being applied to Q1 during the FG's positive portion of the duty cycle and that at that level, Q1 must turn on?

Assuming that the FG output did indeed not go positive during Test 1, is it possible that the FIG 3 is not the correct scope shot for that test?  Possiby the wrong save was pulled from the scope's memory...?

I'll look at the other tests, just trying to get past this discrepancy...

PW

Picowatt - I am well aware of the level of the applied voltage from the signal.  It does not, however, correspond to the actual applied voltage which is somehow overridden by the offset function. 

We could apply this same 'offset' to our 555 tests but not at the same scale of proficiency.  And we only ever applied it to restrict the flow from the battery and to enhance the over all efficiencies.

And I can assure you that it is NOT related to the degradation of any part of our MOSFETS.

Regards,
Rosemary