[time-nuts] Brooks Shera
jimlux at earthlink.net
Mon Mar 25 09:36:09 EDT 2013
On 3/24/13 8:22 PM, Chris Albertson wrote:
> On Sun, Mar 24, 2013 at 7:34 PM, EB4APL <eb4apl at cembreros.jazztel.es> wrote:
>> I wanted to build a GPSDO using the Brooks Shera design since I read the QST
>> article. I asked him in Jan 2009 about his source code, because I wanted to
>> change the PIC to a more modern one and add some functionality....
> This is a perfect example of why people need to publish the source.
> Make it GPL or whatever.
That's a decision that the author gets to make. I've been on both the
supplier and consumer side of that aspect. Sometimes I've published
source, sometimes I haven't. There's a lot of factors involved, and the
consumers need to respect the author: only the author knows all of them.
One reason is that if one DOES release source, one will wind up
supporting it, because generally, we all nice people and helpful, and
it's hard to tell someone no when they send an email asking how to get
it to compile on Version N+3 when you used version N, etc. This can be
a real distraction from whatever else you are doing.
One also DOES get a remarkable number of abusive emails when you don't
provide support and you nicely tell them that. I guess, the first one is
remarkable, after that, you go, "some people are just jerks" and let it
And then there's the folks who argue with you about your implementation
or coding style. "I've rebuilt your program using .net 2.0 and C#
instead of VB6". This is a real quandary.. On the one hand, you have no
desire to support this offspring, but on the other, you would like
credit for the underlying algorithms and initial work.
But the version that was posted to this
> thread is close enough to the final version and it is easy to modify
> any why you like.
> I was going to build one of these too. Then I found out about a
> PIC-TIC which is quite a bit better at measuring the phase difference
> and today we have better parts and cheap FPGAs. The value of the
> Shera design is that it is simple enough to study and understand and
> then go build something that builds on Brooks' idea. Now that you can
> read the code and see the schematic people can study this for years.
> This is another example of a dead-end project. No source code. You'd
> have to start over from scratch to make even the smallest change (such
> as getting it to run on a new version of the OS or an iPad or
The algorithms are well known and understood, and not very complex.
Perhaps starting from scratch wouldn't be a bad thing if you're going to
an entirely new platform. What works well in a dedicated
microcontroller may not work so well in a multithreaded smartphone or
media display device.
I've been fighting a problem with what I thought would be a trivially
simple web browser user interface. Turns out that modern browsers are
entirely too smart, and it's very difficult to force the browser to not
cache an image (e.g. I have an IMG tag pointing to "image.jpg" and I
want to have it pick up the latest rev whenever the page refreshes)
(Firefox is what I'm using, but they all have peculiarities)
The point is that when you move to a qualitatively different platform
(e.g. not just changing from a PIC model X to a PIC model Y), some
architectural thought might be useful.
More information about the time-nuts