[sdiy] Reality versus simulation?
rburnett at richieburnett.co.uk
rburnett at richieburnett.co.uk
Fri Nov 27 17:50:18 CET 2015
Interesting. Can you tie a whole load of the PIC I/O lines together and
then output the same PRBS signal to the whole group? For example, if
you can send the same PRBS state *simultaneously* to all of PORTB and
tie RB0-RB7 pins together then that should give you eight times more
current sourcing and sinking ability than a single pin alone. The lower
output impedance should be closer to what your simulator shows, and the
unwanted effect should be greatly diminished.
Or just make an estimate of the I/O line output impedance and then
adjust the filter resistance to take this source impedance into account,
The other thing that I was going to suggest was that it might be due to
"sinc rolloff" of the spectrum in the discrete-time white noise output,
but I see that you said it had an acceptably flat response before being
connected to the filter when you checked. It might be worth double
checking the "bit-rate" of the PRBS output is what you expect it to be
though, just in case it was significantly less for some reason!
On 2015-11-27 16:32, Tom Wiltshire wrote:
> Sorry about the lack of images. I was trying to get 'scope shots off
> the scope last night, but the non-standard USB connection and software
> to do it is terrible and I couldn't get it to work. Why they couldn't
> just get the scope to appear as an external drive filled with however
> many saved images it'll store is beyond me…but I digress.
> I tried swapping the filter stages around as per Florian's suggestion.
> No change.
> I tried altering the values as per Phil's suggestion (from 560R/10n to
> 5K6/1n), and that gave a marked improvement. So increasing the
> impedance of the circuit seems to help.
> On that basis, I'd expect that Sarah's suggestion of buffers between
> the stages as a test will give a still-better result, but I haven't
> done that experiment yet.
> The scope's FFT Y-axis is dB, so that's the same as I've got in
> LTSpice. Changing FFT window (Blackman, Hanning, etc) doesn't alter
> the result. Scope probes are set to x1.
> On 27 Nov 2015, at 15:50, Walker Shurlds <walkershurlds at gmail.com>
>> Probe capacitance is the first thing I thought also. In addition, the
>> relatively low impedance of the circuit could be shifting the oscope's
>> pole(s) into a lower frequency.
>> Another factor could just be windowing selection from the fft, or
>> maybe the axes aren't logrithmic but are in spice, this only affects
>> whether the response *appears* flat, of course. A few extra pictures
>> might help with diagnosis.
>> On Fri, Nov 27, 2015 at 9:09 AM, Sarah Thompson <plodger at gmail.com>
>> Another possibility would be stray capacitance from the breadboard
>> itself causing loss at higher frequencies. Also, are you using a X1 or
>> a X10 scope probe? If you are using X1, this might be the culprit,
>> though 20kHz is a bit low to start seeing much difference there.
>> As an experiment, try putting opamps before, after and between the
>> stages just configured as voltage followers. This will greatly reduce
>> the distance any high impedance signals need to travel, and tends to
>> make filter weirdness go away. I'm not necessarily suggesting this for
>> a production version, but sometimes going for overkill then backing
>> off is less pain.
>> Sent from my iPad
>> > On Nov 27, 2015, at 5:30 AM, Tom Wiltshire <tom at electricdruid.net> wrote:
>> > As per schematic, +/-10% for the caps, +/-1% for the resistors.
>> > Values wobbling about would shift the cutoff points a bit, but shouldn't give the significant downhill slope I'm actually seeing. Should it?
>> > For example, RC variations give me a lowpass cutoff of supposedly 28.4KHz, but actually 25.6 kHz – 31.9 kHz.
>> >> On 27 Nov 2015, at 11:52, Neil Johnson <neil.johnson71 at gmail.com> wrote:
>> >> Tom,
>> >>> So what's going on with the real-life filters? Can I improve their performance?
>> >> What are the actual values of the capacitors and resistors in your circuit?
>> >> Neil
>> > _______________________________________________
>> > Synth-diy mailing list
>> > Synth-diy at synth-diy.org
>> > http://synth-diy.org/mailman/listinfo/synth-diy
>> Synth-diy mailing list
>> Synth-diy at synth-diy.org
> Synth-diy mailing list
> Synth-diy at synth-diy.org
More information about the Synth-diy