Maker Pro
Maker Pro

Need DDS module, 2nd round

J

josephkk

Jan 1, 1970
0
So do I and it dopesn't have modulation problems. But these are too big
even if I'd remove the enclosure and power supply.



The problem with the uC on these little boards is not the serial link.
Well, at least not if 100Hz sweeps suffice. The problem is that they are
too small to calculate the numbers on the fly and usually have to little
flash left to store them in a LUT. Plus that would be a major
time-consuming hack.

If i could get you to stop and listen i could show you haw to do the
operating side in very little code space, no lookup table, fixed point
math without divides. The user interface or system interface is another
matter though.

?-)
 
J

Joerg

Jan 1, 1970
0
josephkk said:
If i could get you to stop and listen i could show you haw to do the
operating side in very little code space, no lookup table, fixed point
math without divides. The user interface or system interface is another
matter though.

I am sure it can be done somehow. But it's not a slam dunk because the
little uC in there is already running an LCD and other things. One could
do some interrupt shuffling and all that. But ... I've now resorted to
the old analog method. It's my home turf and there's just too many other
tasks at hand to let one like this balloon into its own sizeable project.
 
N

Nico Coesel

Jan 1, 1970
0
josephkk said:
I can even do that in fixed point. Select a reference frequency, say 100
Hz, then scale for frequency at beginning of sweep, and sweep step size

Ofcourse but with floating point its a total no brainer so faster to
develop. In the end floating point is automatic fixed point without
having to think about where to put the decimal point for maximum
accuracy.
 
J

Jasen Betts

Jan 1, 1970
0
If its software then it is not a real sweep but a stepped sweep. If
you use that to test a resonator or a notch filter you might be in for
a nasty surprise.

Not if the software runs fast enough and is done right.

You can microstep a DDS's frequency by jittering the phase accumulator
step size.
 
N

Nico Coesel

Jan 1, 1970
0
Jasen Betts said:
Not if the software runs fast enough and is done right.

Fast enough and done right usually don't go well with an 8051. In an
ideal sweep (or FM modulation) you'd have to recalculate the step size
for every sample.
You can microstep a DDS's frequency by jittering the phase accumulator
step size.

Somehow I have a feeling that will look funny on a spectrum analyzer.
My guess is that you'll see two distinct peaks instead of an
intermediate frequency.
 
J

Jasen Betts

Jan 1, 1970
0
Fast enough and done right usually don't go well with an 8051. In an
ideal sweep (or FM modulation) you'd have to recalculate the step size
for every sample.

I understand that 8051 is the uC that everone that everyone loves to
hate, but the claim depends on the frequency you're running it all at.
Somehow I have a feeling that will look funny on a spectrum analyzer.
My guess is that you'll see two distinct peaks instead of an
intermediate frequency.

You have to do it between each sample. It's the same effect you'd get
by adding more bits to the phase accumulator, but not widening eange
of phase bits used for output.
 
J

josephkk

Jan 1, 1970
0
Ofcourse but with floating point its a total no brainer so faster to
develop. In the end floating point is automatic fixed point without
having to think about where to put the decimal point for maximum
accuracy.

I understand where you are coming from. But the nature of this
application, as stated, is inherently fixed point with an obvious radix
placement. Why put up with FP overhead when you don't need to.

Different viewpoint for this case.

?-)
 
J

josephkk

Jan 1, 1970
0
Cool, I've seen this phasor diagram that shows FM is just 90 deg phase
shifted AM. (but only at low amplitude) I never knew how to make it
work, 'in practice'.

George H.

- Hide quoted text -

The best way to learn is to just build one and play with it. It would be
fun to work with you if you want to explore this. I haven't had the
chance to actually build one, but i guessed the principle while still a
teen.

?-)
 
N

Nico Coesel

Jan 1, 1970
0
josephkk said:
I understand where you are coming from. But the nature of this
application, as stated, is inherently fixed point with an obvious radix
placement. Why put up with FP overhead when you don't need to.

If the overhead is not in the way then why bother trying to 'improve'
something which doesn't need improving? I'd rather spend my time on
adding features which are actually useful to the customer. New
technology brings new solutions to old problems.
 
J

Joerg

Jan 1, 1970
0
Nico said:
If the overhead is not in the way then why bother trying to 'improve'
something which doesn't need improving? I'd rather spend my time on
adding features which are actually useful to the customer. New
technology brings new solutions to old problems.

And somtimes it brings new problems to old solutions. Like the
electronics in many cars, which came at the price of reduced
reliability. Or what just happened with my Thunderbird. Tried an add-on
called Menhy or something to be able to jump to these weird NG post
links with the @-sign in there. Didn't work, no surprise there. But ...
it totally screwed up TB in that now every printed email has the first
3/4 page full of routing garbage. And no proper address field display
anymore. Now I'll have to do a re-install. Just great.
 
J

Joerg

Jan 1, 1970
0
Jasen said:
I understand that 8051 is the uC that everone that everyone loves to
hate, but the claim depends on the frequency you're running it all at.

Hate? It is just about the only uC that has 2nd source (in some of the
packages) and where you can find a capable programmer in almost every
village. What there to hate?

You have to do it between each sample. It's the same effect you'd get
by adding more bits to the phase accumulator, but not widening eange
of phase bits used for output.

Depending on the application some "spectral fuzz" is ok. Certainly would
be in my current one. I just don't want to program a uC :)
 
J

Joerg

Jan 1, 1970
0
John said:
The architecture? The instruction set? The speed?

So what? It's a practical uC, for many reasons. In the same way that I
won't complain that my car would not perform well on the Indianapolis
racetrack. But it does ferry 1/2 ton of pellets back home rather nicely
and that matters more to me than winning some race.
 
R

rickman

Jan 1, 1970
0
If one can program them in a reasonaable time frame which I can't (yet).

If I can find an appropriate eval board for an FPGA, I can gen up a
solution that should do everything you need. How much are you willing
to spend? Unfortunately the FPGA boards tend to be a bit more than MCU
boards. Nearly anyone should be able to do this on an MCU board, but
that isn't my forte.

Heck, I have an old sig gen that I need to repair. I just got a Renesas
MCU eval board for free. I'm sure it has the analog I/O needed. I'll
take a look to see if it will do this. When you are done can I get it
back so I don't have to repair my sig gen? I was just starting to take
a look at it today and got called away. Too fine detail and too hard to
get into... damn old eyes!
 
J

Joerg

Jan 1, 1970
0
rickman said:
If I can find an appropriate eval board for an FPGA, I can gen up a
solution that should do everything you need. How much are you willing
to spend? Unfortunately the FPGA boards tend to be a bit more than MCU
boards. Nearly anyone should be able to do this on an MCU board, but
that isn't my forte.

Thanks for the offer, that was nice. But meantime I already did it the
old analog way.

Heck, I have an old sig gen that I need to repair. I just got a Renesas
MCU eval board for free. I'm sure it has the analog I/O needed. I'll
take a look to see if it will do this. When you are done can I get it
back so I don't have to repair my sig gen? I was just starting to take
a look at it today and got called away. Too fine detail and too hard to
get into... damn old eyes!

Tell me about it, I just designed a board full of 0402 parts and I'll
get to debug it later :-(

So, a few months ago I caved in and bought one of those Donegan Optical
head goggle thingies. Took the 5x version because some of my work is now
much finer pitch than 0402. I can keep my reading glasses on while
soldering that. I was at the point where I occasionally wore two reading
glasses stacked.
 
N

Nico Coesel

Jan 1, 1970
0
Joerg said:
So what? It's a practical uC, for many reasons. In the same way that I
won't complain that my car would not perform well on the Indianapolis
racetrack. But it does ferry 1/2 ton of pellets back home rather nicely
and that matters more to me than winning some race.

An 8051 compares to ARM like a model T Ford compares to a Volkswagen
Golf / Rabbit. IMHO finding a second source for parts is becoming more
and more impossible. Just like the XR2206 8051 microcontrollers may
become extinct real soon as well.
 
J

Joerg

Jan 1, 1970
0
Nico said:
An 8051 compares to ARM like a model T Ford compares to a Volkswagen
Golf / Rabbit. ...


There are many situations where a Model T gets you there. When I rode in
a Model A a while ago I was pleasantly surprised how comfortable that was.

... IMHO finding a second source for parts is becoming more
and more impossible. ...


However, if it is possible one should always prefer the part that has
2nd source unless there are compelling reasons not to.

... Just like the XR2206 8051 microcontrollers may
become extinct real soon as well.


That's been said since over 15 years :)
 
J

Joerg

Jan 1, 1970
0
John said:
Jamie wrote:
[...]
The problem with the uC on these little boards is not the serial link.
Well, at least not if 100Hz sweeps suffice. The problem is that they are
too small to calculate the numbers on the fly and usually have to little
flash left to store them in a LUT. Plus that would be a major
time-consuming hack.

Fancy ramped, modulated DDS sinewave generation at 200K samples/sec would be no
sweat for a $4 ARM, and you get a mux'd ADC and a DAC for free. But yes, it
would take some code slinging.

And I'll probably do that if this becomes a product, with whatever
processor the local SW folks can handle without a major learning curve.
Just not right now, got to watch work load, expenses and time line. So
it'll be analog for now.
 
Top