On the AC modulation, I'm just speculating, but it is helpful to remember that a linear amplifier consumes more than just DC power. The way it works is pretty much like a linear voltage regulator. The greater the power it delivers to the load, the greater the power it consumes (which is why a Class B amplifier is not a paragon of efficiency). And it is the transistor that is doing that work. In other words, if the DC bias circuit can be thermally destabilized, then a thermal transient from turning up the volume is a possible trigger. The question, then, should shift to the stability of the biasing approach and/or thermal management of the transistors.
What you seem to be observing is a subtle effect involving what professionals in stability analysis like to call "the existence of two local minima." One of those is the one you want: the desired DC operating point, adjusted within limits by the potentiometers. The other is what transistor saturation leaves you after the DC bias circuit involving two transistors and multiple feedback paths slides off due to thermal runaway. I have seen this in my F2J and in simpler resistively loaded Class A test circuits, but usually when I'm pushing the envelope on power dissipation while biasing to the far side of the sweet spot. Those leaky gate rejects I mentioned can be more susceptible too if the bias circuit resistances are not lowered. That is why I speculate in your case that it most likely has something to do with a series resistance in the DC bias loop that is ballasting the forward leakage current of the device. One explanation is that one of your JFETs gets more leaky compared to the rest as the temperature increases. That doesn't necessarily make it a bad transistor, just different enough to destabilize the bias circuit. Lowering R5 or R6 might improve the stability margin, but since I have not analyzed the circuit I would not stray far from a proven design; those resistors in the schematic may have those values for more than one reason. Naturally, others on this forum that have built the amp will have better advice on that than me.
Also, moving the JFETs around to different positions in the F6 and seeing what happens can narrow down a leaky culprit if there is one. But don't be too quick to accuse a JFET. As has been suggested, less than ideal thermal bonding to the heat sink is another possibility to consider. The finger test can be useful but it has its limits. (I do it myself sometimes as a sanity check despite having the optical infrared measuring gadgets, although I rue those days when some pesky 2N3904 leaves a TO92 imprint on my index finger
!) I'll reply to my pal Generg about my thermal management practice next. I recommend to measure nevertheless the temperature of the SS cases or nearby on the heatsink. Possibly your DMM can do it.
I often had differences of over 10 degrees although I had the impression to have screwed them all the same. Doing it again often solved the problem.
But I never had the effect you describe neither with my three J2 clones nor with two different F6 and my friend building the same amps also never spoke of such a problem.
So I do not really know.....interesting that Mike knows this effect too....
Mike you should better screw your SSs.....
Well, let me begin by saying I completely agree with Generg that thermal management of the JFETs is something that requires attention to detail. But what may surprise some folks is that in all my testing I have not used a drop of thermal greese nor sacrificed even one mica insulator. I just bolt the transistors to a metal heat sink left over from projects long gone and wire the transistors up. I very often use clip leads as Nelson himself has admitted to (sinners we are). I wouldn't do this in an amp I expect to keep and show case unless I redesign it so that the drains of the transistors bonded to the heat sink are at ground potential. But otherwise, I view a thermal insulator as a terrible evil only justified by the need to electrically isolate the tab of the transistor.
Thermal grease is practically persona non grata in my prototypes as its practical application can easily be (and often is) thermally counter productive. Having been involved in research for many years whose goal is increasing the power density of power electronics, I have been shocked, from time to time, to discover that the thermal resistance of "thermal grease" is quite high. My earliest teachers in bench electronics (when I was merely an engineering student or less) insisted that only thin coatings of the stuff be used. It turns out its real purpose is to fill voids that might represent air pockets. In other words, go for a thin sheen, not a blob when applying the stuff in your amplifiers.
So I am heat sinking without thermal insulator and without thermal grease. Kinda like vacationing at a naturalist resort! But, wait, there is one more wonderful advantage. Going thermal grease free I can't ruin my wardrobe! Ever try to get that stuff off of anything, especially clothes? IMPOSSIBLE! The slightest drop seems to end up everywhere.
Thermal grease is practically persona non grata in my prototypes as its practical application can easily be (and often is) thermally counter productive. Having been involved in research for many years whose goal is increasing the power density of power electronics, I have been shocked, from time to time, to discover that the thermal resistance of "thermal grease" is quite high. My earliest teachers in bench electronics (when I was merely an engineering student or less) insisted that only thin coatings of the stuff be used. It turns out its real purpose is to fill voids that might represent air pockets. In other words, go for a thin sheen, not a blob when applying the stuff in your amplifiers.
So I am heat sinking without thermal insulator and without thermal grease. Kinda like vacationing at a naturalist resort! But, wait, there is one more wonderful advantage. Going thermal grease free I can't ruin my wardrobe! Ever try to get that stuff off of anything, especially clothes? IMPOSSIBLE! The slightest drop seems to end up everywhere.
Thats were my findings too! From my recent experiments with different insulators with and w/o grease. For now I use these guys: SPA2000-0.015-00-104 Bergquist Company | Mouser, they do not require any grease.
Going thermal grease free I can't ruin my wardrobe! Ever try to get that stuff off of anything, especially clothes? IMPOSSIBLE! The slightest drop seems to end up everywhere.
I guess you are unlucky, I never got it on my clothes =)
And it's removed easily with some rubbing alcohol.
Last edited:
So I am heat sinking without thermal insulator and without thermal grease. Kinda like vacationing at a naturalist resort! But, wait, there is one more wonderful advantage. Going thermal grease free I can't ruin my wardrobe! Ever try to get that stuff off of anything, especially clothes? IMPOSSIBLE! The slightest drop seems to end up everywhere.
I can always tell when a rookie has been playing around with the transistors by taking a quick look at the amount of grease smeared all over everything. The chances of things being really screwed up seem to be proportional to the amount of goop....
As to thermal barriers like grease and insulators the thinner is always the better.
He, he. If I had, I wouldn't admit it here. Because I know for sure my wife hasn't.
He, he. If I had, I wouldn't admit it here. Because I know for sure my wife hasn't.
Hello Doc. I am glad that you and your interesting posts are back. On a less serious note, what is the future of the JFETs made by SemiSouth?
Best regards
The reason I mentioned "spot welding" was that in 1964 I was involved in a university computer project that actually used spot welding of components rather than solder. No PCB traces, just point-to-point wiring with a type of wire (and teflon insulation) suitable for welding. It was a very expensive and time consuming construction technique. The reason this approach was taken was that many existing computer PCBs were very flaky, having cold solder joints and poor copper to insulator adherence.
In the limit, the semiconductor can be made intrinsic to or built in its custom-made heatsink. It goes from its manufacturing wafer to a prepared cave in the sink rather than to make a TO-3 packge etc.. The destiny of R100 in F6 is cast in stone; to sit on an adequate heatsink; might as well make it part and parcel or inextractable from it.@Semisouthfan
I see, no grease no mica, that is possible with F2, only one SS and two separate heat sinks with no comon case connection…..
but in case of F6… two SS per side… nevertheless…..i will think over it……
Thanks for the long and interesting explanation!
@Semisouthfan
I see, no grease no mica, that is possible with F2, only one SS and two separate heat sinks with no comon case connection…..
but in case of F6… two SS per side… nevertheless…..i will think over it……
Thanks for the long and interesting explanation!
Agreed, since an F2 or an F2J uses complementary parts: p-channel MOSFET in the high side is the current source with an n-channel MOSFET or n-channel JFET as the low side gain device. Thus, both can be conveniently bonded to the heat sink with good thermal and electrical communication, as long as nothing else touches the heat sink. It works fine on the bench for casual testing, but I wouldn't count on that in a finished amp that isn't fully contained in a separate enclosure.
Unless a bias design were used that ensures the common point is always at ground potential. Hmmm, I think I've got one of those on the bench now...