Saturday, February 25, 2012

Prop drift uncovered

Yesterday I wrote that I was shelving the Propeller beacon project after discovering that the frequency stability is unacceptabe. Eldon WA0UMH persuaded me to try some tests to identify the cause of the drift. My conclusion is that it is a combination of factors.

The major factor causing the drift is the power supply voltage. Whilst developing the beacon using just the Gadget Gangster board I was powering it with a 6V supply - the minimum needed. After adding a PA I needed to increase the supply voltage to 9V. The drift is greater with 9V, regardless of whether the PA is actually connected.

Looking down on the Propeller board
 If you look down at the Gadget Gangster Propeller USB board there are two SMT chips to the left of the Propeller and its clock crystal which appear to be voltage regulators. There is a 5V regulator a few millimetres to the left of the 10MHz crystal and a 3.3V regulator below it. The greater the supply voltage, the more heat the 5V regulator has to dissipate. Plugging the LCD UI board on top of the Propeller board traps in more of that heat making the drift even worse. This is not the first time I have discovered that voltage regulators and crystal oscillators don't go well together.

Whilst it is useful to know what is causing the drift, my discovery has not indicated an easy solution. Either I use a temperature compensated 10MHz crystal oscillator as suggested by one of my readers, or I use the Prop to control an outboard and more stable synthesized oscillator. The first option looks the easiest, but the TCXO is not an inexpensive component.


Richard G3CWI said...

There are a number of things that I don't like about my Prop USB board too. I suspect it has been designed by a software engineer!



Kenneth Finnegan said...

What about separating the PA supply from the board supply, and then have an external regulator farther from the crystal drop the 9V down to 6V for the board.

Isn't the drift more due to the change in temperature than the elevation in temperature? What if you were to attach a PTC heater to the crystal and simply hold it well above the 5V regulator all the time?

Julian Moss said...

I did think along those lines, Kenneth. I think that would reduce the drift but possibly not eliminate it to a sufficient extent that it would be no more than 1 or 2Hz on 28MHz or even 50MHz where it would be 2 or 4 times the drift on 14MHz. The fundamental problem is that it isn't a good idea to have heat generating components close to frequency determining ones. The clock crystal is not intended to be a frequency reference, of course, but I think Richard's comment is spot on! :)

g4fre said...

The problem of voltage regulator being close to the crystal is similar to the problem I had with my NJQRP DDS60 board on QRSS where the 30Mhz oscillator is on the flip side of the PCB. see After a lot of time making a polystyrene nest for the oscillator I discovered the best solution was to supply lower voltage to the board than the 13.8V i was using. 6V worked fine and cured the drift