#qcx PTT modification and 1.03 firmware #qcx


Hans Summers
 

Hi all

Last week my QCX suffered an explosion or to be more accurate, the 50W PA attached to it, suffered an explosion. It was during some testing of the CAT firmware enhancement. I had switched the QCX off and on with the toggle switch on the my front panel, rather fast. There was a bright flash and a loud bang (remember, this thing is connected to a 20V 20A power supply). Eventually I had some time to investigate this and found that one of the IRF510 had blown up.

I developed a 1-resistor modification that prevents this kind of issue. I have documented this here:

It is unlikely that anyone keying an external PA would have this particular issue but I have included it in case it helps someone. It's only one quick resistor, too... and after all replacing IRF510s is a bit tedious and the bang is a bit scary... 

Note also that from now one we will supply new chips with low fuse setting 0xD7 not 0xF7. F7 has a 65ms start-up delay which is NOT needed when you are using brownout detection. 

73 Hans G0UPL



Alan G4ZFQ
 

I developed a 1-resistor modification that prevents this kind of issue. I have documented this here:
http://www.qrp-labs.com/qcx/qcxmods/ptt.html
Hans,

Strewth! Are you sure you did not engineer this fault just so you could prove it existed by using your new scope?

73 Alan G4ZFQ


Dave New, N8SBE
 

Hans,

One of my well-worn sayings:  "If I had a nickel for every bad power supply issue, I'd be rich."

Once again, I appreciate the thoroughness of root-causing the issue, and doing some testing to validate the fix.  In retrospect, designing everything to come up in a 'fail-safe' condition (floating pins properly terminated, for example) would go a long way to preventing this sort of thing in the field.

A programmable smart supply or at least prototype bench supply with a solid-state switch that you could set determinant delays on over a given range would be very useful in finding and fixing these kinds of things.  I can appreciate you flipping switches madly, but how do you know you've covered all the possible delays that might cause an issue?  There might still be a 100 msec window some place...

I know that you aren't flush for testing equipment.  Maybe someone reading this will have an old programmable power supply they might donate to the cause.

At my day job, a Keysight N6705B is part of my arsenal for testing electronic modules.  The price, of course, puts it outside the reach of mere mortals, I'm afraid.

73,

-- Dave, N8SBE

-------- Original Message --------
Subject: [QRPLabs] #qcx PTT modification and 1.03 firmware
From: "Hans Summers" <hans.summers@...>
Date: Tue, December 03, 2019 6:35 am
To: qrplabs@groups.io

Hi all

Last week my QCX suffered an explosion or to be more accurate, the 50W PA attached to it, suffered an explosion. It was during some testing of the CAT firmware enhancement. I had switched the QCX off and on with the toggle switch on the my front panel, rather fast. There was a bright flash and a loud bang (remember, this thing is connected to a 20V 20A power supply). Eventually I had some time to investigate this and found that one of the IRF510 had blown up.

I developed a 1-resistor modification that prevents this kind of issue. I have documented this here:

It is unlikely that anyone keying an external PA would have this particular issue but I have included it in case it helps someone. It's only one quick resistor, too... and after all replacing IRF510s is a bit tedious and the bang is a bit scary... 

Note also that from now one we will supply new chips with low fuse setting 0xD7 not 0xF7. F7 has a 65ms start-up delay which is NOT needed when you are using brownout detection. 

73 Hans G0UPL



Hans Summers
 

Hi Dave

One of my well-worn sayings:  "If I had a
nickel for every bad power supply issue, I'd be rich."

Yeah agreed! Cause of many problems!

Once again, I appreciate the thoroughness of root-causing the issue, and doing some testing to validate the fix.  In retrospect, designing everything to come up in a 'fail-safe' condition (floating pins properly terminated, for example) would go a long way to preventing this sort of thing in the field.

Yes... you are right. All the pins of the processor are floating during power-up. Anything that controls anything important ought to be pulled in a default condition. It would be over-pessimistic to pull EVERY pin but a sensible decision about which are important is very useful. There haven't been any problems with QCX in this regard but when it comes to QCX driving an external PA... apparently this obscure failure mode can arise. I don't think it would have been likely to have been widespread but it is easy and prudent to eliminate the possibility up front! 

A programmable smart supply or at least prototype bench supply with a solid-state switch that you could set determinant delays on over a given range would be very useful in finding and fixing these kinds of things.  I can appreciate you flipping switches madly, but how do you know you've covered all the possible delays that might cause an issue?  There might still be a 100 msec window some place...

The PTT signal is now low properly, until at least 1 second after power-up, when the processor takes it briefly high in order to mute a click while the Si5351A configuration occurs. I left that 8ms pulse in, because muting is nice and by then I believe the PA must be long in a stable well-defined state. Switching in the PA is micro-second scale. 1 second is ages. 

I doubt any reasonable power supply glitches or start-up glitches could cause this particular failure mode now with the PTT fix in place. Which isn't to say that there isn't some as yet undiscovered failure mode lying sneakily in wait. 

I know that you aren't flush for testing equipment.  Maybe someone reading this will have an old programmable power supply they might donate to the cause.

At my day job, a Keysight N6705B is part of my arsenal for testing electronic modules.  The price, of course, puts it outside the reach of mere mortals, I'm afraid.

I have a quite reasonable 'scope and a slightly aging but good spectrum analyzer. These two between them, do a great job on most things. But yes a programmable supply is not available. I manage to scrape by mostly! 

73 Hans G0UPL