One thing that comes to mind when comparing the inductive spike at the drains between the TN/VN*06 and the BS170 - you saw a 120V spike with the former compared to 60V with the BS170 - is that the BS170 is not turned on all the way with a 5V drive while the *06 devices are more strongly turned on due to their lower gate thresholds. This was the primary reason I suggested the *06/*10 FETs. The lower gate threshold of the *06 FETs results in a higher drain current which builds a larger magnetic field in the inductor and therefore a larger inductive kick when the FET turns off. Regardless of what physical limitations in the BS170 are clamping the spikes at 60V, the *06 devices will produce a larger spike and will also produce more power output because they’re simply better switching transistors than the BS170 for the existing PA design. They also dissipate less power than the BS170 because Vds is smaller when they’re on. In any event, I certainly would not recommend allowing a spike over 60V so the clamp diode is a good addition. Long story short and with all other things being equal, the TN/VN*06 transistors are better suited to the existing PA design than the BS170. Others on this forum have found this swap to be very beneficial and I’ve yet to hear a complaint of any *06 devices blowing.
Tony - AC9QY
The diode is a good idea I think. Thisis based not only on simulation but on measurements I have made with my scope.
In my case, the loss of two or three sets of BS170 in my two QDX units led me to try different transistors. I ran the other transistors without any failures we for months. But I was curious why the BS170s would fail and the TN0106 and VN0606 would not. Yesterday
I got a new scope that allows me to see transients and I decided to look at the QDX with VN0606 transistors in it. To my surprise, the glitch at the end of a transmission was over 120 volts measured at the center tap. The transistor drains were similar. The
spike was 100ns wide at the 60 volt level. And this is into a dummy load. The transistors must just be capable of dealing with this.
I then replaced the transistors with BS170s and discovered another surprise. The spike was below 60 volts and had a broad flat (but ragged looking) top.?
Then I added the commutating diode and the spikes were gone.?
Right now my theory is that the BS170s are operating in a "make before break" mode which minimizes the di/dt. The VN0606 (and TN0106) might be in the "break before make" mode. The specific operating conditions might possibly allow the circuit to move between
these two cases and the BS170s are just not capable of handling the over-voltage spike while the other transistors can. Just a theory mind you. I have not tried yet to make measurements to verify this. Prior to these measurements, I was thinking that the opposite
would be the case, thus explaining why the VN0606/TN0106 did not fail.?
In any case, I am now running with BS170s again but with the diodes installed. We shall see how long this lasts.
Tony
AD0VC
Simulations are here:
/g/QRPLabs/message/105254
JZ
On Mon, Aug 7, 2023 at 8:07?AM John Zbrozek <
jdzbrozek@...> wrote:
>
> Hi Pierre,
>
> The diode simply is in parallel with the 47uH inductor L14, with the
> orientation shown in the photo provided by Paul W9AC. Cathode towards
> the power supply connection, anode to the transformer center tap.
> Any of the diodes he mentioned? (1N4448, 1N4148, 1N914) will do a fine
> job of absorbing the end-of-transmission voltage spike it produces.
> I have used a 1N914.
>
>
/g/QRPLabs/message/105256
>
> Here you can see simulations with and without the diode present.
>
> On another thread, Evan, Hans and I explore the ability of QDX to
> withstand SWR. It seems to have a reputation for SWR intolerance but I
> think there is a case that the problem may actually be as described
> above.
>
>
/g/QRPLabs/message/106759
>
> Regards, JZ KJ4A
>
> On Mon, Aug 7, 2023 at 7:46?AM Pierre FK8IH <
jb.gallauziaux@...> wrote:
> >
> > Could you please elaborate: what diode, how did you connect it? A handmade schematic and a photo would be nice.
> > 73
> > Pierre
> > FK8IH
> >
