Don't forget that a patent only gives you the right to take the issue to court. Going that route you should assign it to a large company that will provide your protection for you. Plus, you get paid up front. 🙂
Carefully read any contract though!
-Chris
Carefully read any contract though!
-Chris
For folks across the pond Farnell (tied in with Newark these days) has 319 of the "with lead" (not /NOPB) version of the LME49990MA in stock:
LME49990MA - TEXAS INSTRUMENTS - Operational Amplifier, Single, 110 MHz, 1 Amplifier, 22 V/µs, ± 5V to ± 18V, SOIC, 8 Pins | Farnell element14
LME49990MA - TEXAS INSTRUMENTS - Operational Amplifier, Single, 110 MHz, 1 Amplifier, 22 V/µs, ± 5V to ± 18V, SOIC, 8 Pins | Farnell element14
Hi Edmond,
Good to see that you are still active on this.
First, I think that perhaps a combination of your software and JensH hardware could be very interesting for us DIY guys.
Second, I wouldn't mind paying some money for a piece of software that actually brings me some value. And at a later time when you don't want to have the burden of support, and such anymore, I think you could consider making it open-source.
One thing you could consider was to make an interface public so a third party would be able to write code that allowed control of external interface box, if that makes sense in your setup. Just a thought.
Cheers,
Mogens
Yes well GFs think whatever time we put into these things is taken away from them.
With that said then how do we profit from watching sports?
Like my aunt Anna Rosanna Danna used to say....
Looking at your output plots and the description it works very similarly to the Shibasoku distortion analysis. Reading between the lines it seems you are synchronizing with the fundamental and multiplying so only harmonics survive and noise drops out.
I'll beta test it if you need.
I found some spreadsheets for a no window perfect measurement setup but its not easily.
Demian
I was about to mention a similar technique, just phase lock the A/D sampling to the input forcing an integer ratio, then average over single cycles probably a big processing gain over averaging long FFT's.
Actually the no window technique is trivial by forcing the stimulus to be an exact ratio to the clock. Within the same interface the D/A and A/D clocks should be the same forcing the fundamental and all harmonics exactly into FFT bins with no leakage or amplitude error. ARTA does this with its multitone.
But isn't this a little cart before the horse i.e. you still need a stimulus that supports the THD levels you want to measure.
Last edited:
And not useful for an asynchronous analog source that won't have either the frequency accuracy or stability to fit exactly. However some sample tweaking, finding exact start and stop points on samples and some other fudging could emulate the synchronous equivalent. If you are sampling at say 192X there should be enough resolution to do some meaningful coherent averaging. Or autocorrelation?
I meant PLL and the system becomes what I think they used to call "orders" as when studying machine rotation where one rev becomes the first order and absolute frequency becomes irrelevant. Also simply clock the DAC and the A/D from the same clock which is possible when doing amplifier THD. This is not the same as playing say a CD player and measuring it with a separate sound device.
Last edited:
That is essentially how the Shibasoku analysis works. It has a PLL that locks to the fundamental driving the sample clock.
I've never tried this but will the 725 operated in analysis mode when in level mode?
How do you get rid of 3H in an amplifier? I retuned the ShibaSoku AG16B generator using the ShibaSoku 725D and its monitor output going to a QA400 FFT.
Got the 2nd down into the noise floor of the analyzer but 3H is still hanging tough at -136dBv.
BTW -- how do i also get rid of the 498Hz 'clock' or signal which is always in the data.
THx-RNMarsh
Got the 2nd down into the noise floor of the analyzer but 3H is still hanging tough at -136dBv.
BTW -- how do i also get rid of the 498Hz 'clock' or signal which is always in the data.
THx-RNMarsh
And this is how DiAna does not work. No need for physically locking the phase of the sampling clock to the fundamental. You just need to know the exact phase and frequency of the fundamental (with respect to sampling rate of the ADC of course) and use this information for further data processing.
Cheers, E.
Edmond it is a similar technique to Bill Waslo's Diffmaker?
BTW On business models: the ARTAs and TrueRTAs of this world seem to make some money on giving away a crippled version, and a paid-for full feature version you can buy from without the free version. Could be a good incentive.
You might want to consider that.
Jan
BTW On business models: the ARTAs and TrueRTAs of this world seem to make some money on giving away a crippled version, and a paid-for full feature version you can buy from without the free version. Could be a good incentive.
You might want to consider that.
Jan
Hi Jan,
Perhaps Hanspeter, the author of HpW Works, could tell me if it's worth all the effort.
Cheers, E.
Basically, yes it is. Input signal respectively output signal of a DUT are compared and the difference analyzed for harmonics. Sounds simple, but doing it right is quite a challenge. Note that this technique, which I call -for the sake of convenience- "differential mode" is just an extension of DiAna, as more regular THD and IMD measurements are possible too.
That was exactly my initial plan. But I'm dreading the idea of implementing a product key or dongle for the paid version and building a website + payment system. Too much hassle for probably only a few customers.
Perhaps Hanspeter, the author of HpW Works, could tell me if it's worth all the effort.
Cheers, E.