Quote from: picowatt on May 08, 2012, 05:58:17 PM
@.99,

With the new "Altoids" circuit, how low were/are you able to get the DC bias current?

It's presently at about 18mA...just about right to drive that LED to full brightness.

Quote from: TinselKoala on May 08, 2012, 06:23:59 PM
Saaayy... do either of you  know anything about microprocessors? I have these Arduino clones with 16 and 32MHz Atmels in them,  and some other peripherals here... I'll bet it would be possible to program an Arduino to make the right frequency oscillations and interface its PWM output into an easily tunable little filter or pulseforming network that would light up an LED when OU was attained...... What do you think?

I've got some good opamps on hand but no AD834s.

TK,

The AD834 (and its slightly cheaper cousin AD835, they're pricey in singles) would be a decent multiplier for the analog signals.  The AD835 is good to 250MHz (and available in DIP).  A pair of fast precision buffers driving the inputs and for level scaling (the 834/5 have pretty high input bias currents) and then a LP at the output for averaging could be a good beginning to a neg/pos mean power measurement circuit.

An analog zero center or a digital multimeter at the output could be used as an indicator.

Possibly a lower cost, lower BW multiplier could be used.  National and others makes some multipliers as well, but I'd have to check the BW.  I was just "thinking out loud".

And yes, sometimes those gremlins do things while you sleep...

PW

Quote from: TinselKoala on May 08, 2012, 06:23:59 PM
Saaayy... do either of you  know anything about microprocessors? I have these Arduino clones with 16 and 32MHz Atmels in them,  and some other peripherals here... I'll bet it would be possible to program an Arduino to make the right frequency oscillations and interface its PWM output into an easily tunable little filter or pulseforming network that would light up an LED when OU was attained...... What do you think?

I've got some good opamps on hand but no AD834s.

TK,

As for the microprocessor approach, if you had fast enough AD converters for inputs, you could perform the math in code.  But I don't think the uPC's your discussing have fast enough conversion rates for any AD converters that may be available on chip.

With the NERD, TB, and now "Altoid's" circuit, even though the fundamental of Fosc is fairly low (the highest being .99's new 4.5MHz circuit), there are a lot of harmonics due to the non-sinusoidal waveshape.  I would want to AD convert at at least 50MSPS, but would be more comfortable with 100MSPS or better.  Most onboard AD's won't do that.  External high speed 8bit converters are fairly low cost, but even then, the microconrtroller would have to be able to input at that rate and still have enough machine cycles available to do the math, housekeeping, etc.  It would be an excellent way to go, but would require a pretty fast controller.

There are some IC solutions out there that would be suitable.

PW   

To be one of the 3 designated readers of this saga I had to sign a contract, and that
contract says I have to contribute occasionally, or face a fairly hefty fine. It doesn't say
anything about following any protocols or that I have to include attachments, just that
it be short and preferably not inflammatory. I figured a quote would satisfy my obligation:


"Ideas not coupled with action never become bigger than the brain cells they occupied."
Arnold H. Glasgow
tak

Quote from: picowatt on May 08, 2012, 07:08:38 PM
TK,

As for the microprocessor approach, if you had fast enough AD converters for inputs, you could perform the math in code.  But I don't think the uPC's your discussing have fast enough conversion rates for any AD converters that may be available on chip.

With the NERD, TB, and now "Altoid's" circuit, even though the fundamental of Fosc is fairly low (the highest being .99's new 4.5MHz circuit), there are a lot of harmonics due to the non-sinusoidal waveshape.  I would want to AD convert at at least 50MSPS, but would be more comfortable with 100MSPS or better.  Most onboard AD's won't do that.  High speed 8bit converters are fairly low cost, but again, the microconrtroller would have to be able to input at that rate and still have enough machine cycles available to do the math, etc.  It would be an excellent way to go, but would require a pretty fast controller.

There are some IC solutions out there that would be suitable.

PW

After a bit more thought, since we would want a long average at the output, it may be possible to undersample the signals as long as both the DUT and the uPC are asynchronous to each other and if very large numbers of cycles were to be averaged.

Additional thought needed.

PW