When I finally get round to building one of these again, I'm sure that PSoC will be a fantastic supervisory controller, with its flexible analog and programmable digital hardware.  This will allow the monitoring and calculation of temperatures, power dissipation levels, overload and operating points, built-in timers, power quality analysis, earth loop impedance measurement... so many useful things a great audio power amp might need.  If any of you out there are doing or have done any power amplifier supervisory stuff, check out PSoC3, and ping me if you have any questions.

After I've got over the shock of the beginning of the new quarter, my next published Filter Wizard article will likely be on circuits that generate voltage gain using only pure lumped resistors and capacitors.  Useful?  Don't know, but certainly unexpected!  best / Kendall

Meanwhile, a retro analogue synth is full of all kinds of circuitry that demonstrates interesting design technique.  I'll be making some more contributions to the Elektor PSoCaMorph project soon (the Editor's name, but my idea).  I wanted to design a really good VCO to use as a master oscillator, and I think the basic circuit techniques are well worth deploying in other non-musical applications.  Check out http://www.elektor-projects.com/project/the-psocamorph.12269.html, sign up and register your vote!  Our lovely marketing department has agreed to make some development kits available to people who want to collaborate on bringing this project into being.  All for now!  best / Kendall

We have several reference designs; some incorporate power factor correction (PFC) and some are compatible with dimming systems but currently we don't have a design with both.  Someone mentioned that another supplier had gone to great lengths (and cost, and board area) to do this, so why didn't we?  PFC is not mandatory for devices that consume this level of power.  But wouldn't it be more "good citizen" to include PFC in our dimmable designs, to take advantage of the cleaner input current waveform, despite the costs?  Well, incorporating PFC into a dimmable LED light bulb is less valuable than you'd think.

It's true that poor load power factor causes significant extra wasted energy in the infrastructure that connects you to your electricity supplier.  I studied this in brain-hurting detail when I looked at how to use the new PSoC3 and PSoC5 devices in smart electricity meter applications.  Clarifying the misinformation about this out there on the web is worth an entire long article on its own!  But, to stay on-topic for today's blog, it's important to know that a major contributor to the "dirtyness" of the current drawn by many homes and offices is... the traditional light dimmer.  People often ask for dimmability in a modern LED bulb, and consider that the ultimate LED bulb would be both dimmable and PFC'd.  Let me tell you why I think that could be a waste of time and effort.

It's simple.  Whatever load you attach to a conventional triac-based dimmer (the vast majority of dimmers out there), the average current is controlled by switching off the current flow after a certain percentage of the AC cycle has occurred.  The resulting "spiky" current has a power factor that degrades dramatically as the dimming effect kicks in.  It's only an empirical observation, backed up by some informal canvassing, but a great many modern homes, particularly in the US, seem to have lighting installations comprising a multitude of very bright bulbs that are consequently run "very dim" by turning the dimmers most of the way down.

As a result, the otherwise perfect power factor of the ubiquitous filament bulb is competely squandered by the phase-angle dimming technique.  And it shouldn't be much of a surprise that replacing the filament bulbs with PFC'd LED bulbs will make no difference to the power factor.  You'll still be causing the power utility's transformers and cabling to get hotter than would be expected from the actual billable wattage you're consuming (which will be reduced by using the LED lamps, obviously).  The ideal situation would be to rip out all these triac-based lighting dimmers and replace them with a much more intelligent solution to lighting level management (ideally using PFC'd bulbs).  If you're going to keep these dimmers, then for sure you need a dimmable light bulb design, but you definitely don't need a PFC'd design. 

So, not much about filters this week, then.  Just a cautionary tale about not getting over-excited about solutions to non-problems - a message I could sometimes do with contemplating myself in the filter world...  Happy illuminating! 

Best - Kendall

http://www.eetimes.com/electrical-engineers/education-training/courses/4237731/Wizard-Filters-for-all-using-the-built-in-hardware-digital-filter-in-a-programmable-SoC

It's really too short a piece to carry any in-depth (or even in-shallow) technical perspective, but it was fun to do, and whetted my appetite for other forays into this format.  Let me know what you think!  best / Kendall

http://www.analog-eetimes.com/en/now-synthesize-your-fir-filters-using-high-school-algebra.html?cmp_id=71&news_id=222903141

and see just how easy it can be to make an FIR filter that has notches where you want them to be, not scattered randomly by some mystery filter design process.  Check out my 'null hypothesis'!  best / Kendall

http://www.analog-eetimes.com/en/analyze-fir-filters-using-high-school-algebra.html?cmp_id=71&news_id=222903041

Dust off that high school algebra about polynomials and start digging!  The second part will be along in a few weeks' time!  best - Kendall

I'm fleshing out some topics for the Filter Wizard articles that I publish on EE Times Europe (http://www.electronics-eetimes.com/en/News/filter-wizard.html).  Why not tell me which of these seem the most interesting or relevant to you, and I'll see if I can arrange my schedule around them.  Or let me know what else you'd like me to write about.  The "Five things you should know about" format of the last one proved popular and I may adopt that for many other pieces.  Here's a list of some things I'm looking at:

Delay - time, phase and group.  What is it, what gets delayed, how to create filters with zero or even negative group delay (at some frequencies) and why you'd want to.

AC line interference - approaches to filtering it out (if you couldn't stop it getting in).  The "dial-a-notch" method of FIR filter design.

Ninety degrees of separation - quadrature: how to get it and where you use it

Waveform distortion - big deal or no deal? What information is there in a waveform, and what damage might different types of filters do?

Electrical energy measurement: how good do the ADCs need to be? Can we use just one ADC and get good results on reactive loads?  How custom filters help here.

The new PSoC Creator Filter component using the DFB - now does IIR filters too, with pole ordering for best SNR, and direct access to the optimized coefficients.  Out beginning of February.

So - give each of these a score 0 to 5 and send the result to filterwizard@cypress.com.  And let me know what else I should be writing about - theoretical or practical.  My deadline for the first article of the year is approaching fast, so get slewing!  best - Kendall

http://www.analog-europe.com/en/sample-multiple-channels-simultaneously-with-a-single-adc.html?cmp_id=71&news_id=222902424

http://www.analog-eetimes.com/en/a-fast-settling-bias-voltage-filter-with-high-ripple-rejection.html?cmp_id=71&news_id=222901909

Oh, and for another trick to help reduce the size of low-frequency gain rolloff caps in compact audio circuits, see my article in Linear Audio (www.linearaudio.net ; you actually have to pay money for it and get a paper version, how retro!  But well worth it.)

Keep on filtering!

If you've ever designed a high order analog active filter from a cascade of 'biquad' sections, you'll probably have discovered how important it is both to get those sections in the right order and to get the individual gains set properly.  It's the only way to avoid premature overload (sometimes hard to detect when it occurs internally) and excessive build-up of noise.

It seems like many digital filter designers are unaware that the issues of hidden overload and noise build-up apply just as much to digital IIR filter cascades as they do to analog filters.  In fact, although the output noise power of an analog filter falls with reducing cutoff frequency, that of a digital filter actually increases!  It's to do with the magnification of the inevitable quantization of the filter calculations, and I'll put a Filter Wizard piece out on the web on that soon.

These days, the presumption is that you'll have enough spare bits both above and below your working dynamic range that you can completely ignore this.  But these bits aren't free, either in area or power consumption.  In practical, commercially viable systems, you often have to push the hardware you can afford close to its theoretical limits to get the job done.  The PSoC3/5 DFB is a 24bit fixed-point engine, but you should still use such capacity wisely to have a chance of achieving respectable performance in your application.

So, I've put a lot of effort into ensuring that the IIR filters that the next Creator FILT component will generate are carefully sequenced and gain-adjusted to get the best (or close to the best, at least!) dynamic range possible, for any given response shape and cut-off frequency.  Low-frequency filters in particular benefit from this.  Internal safety margins are set to permit the worst case digital input signals - signals that are highly unlikely to originate from any physical system through a sampling process.  DC offset buildup - sometimes a problem with carelessly implemented IIR filter code - has been eliminated.  So, if you want to use any of the IIR filter types that we'll be supporting in the next release, please do use the Creator internal routines to do the design.  Don't assume that because you paid $10k for the math package you use, that it will do a better job - because it quite possibly won't!

Meanwhile, for the last few days I've been in San Francisco at the 129th Convention of the Audio Engineering Society, catching up with the latest technology and product trends in high quality and professional audio. Seeing the wide spread of equipment on show at the associated exhibition brought home just how useful our new PSoC3 devices will be in this space. Of course, all the micro guys can do a mix of buttons, lights and displays. But the programmable digital hardware of PSoC3 will give it a real edge for people who are doing custom signal and control interfaces, which abound in pro audio.

Audio amplification has always been a favourite subject of mine, and I'm also struck by what a good supervisory and control processor for amplifiers that PSoC3 will make. The flexible analogue capability can be deployed for voltage and current monitoring (of the amp and the psu), power device safe area protection and thermal management, and control of internal bias currents and voltages. Novel switchmode and hybrid power supply architectures can be developed and optimized for the particular application. Programmable logic enables no-code protection interlocks, and you can use the CAN bus to form a network between multiple amplifiers to ensure synchronized performance in high channel count systems. And for many applications, the audio signal processing, routing and management possible with UDB and DFB support will be more than sufficient.

All in all, I think this is a really exciting application and close to my heart, and I hope many of you out there will take up your DVK and your copy of PSoC Creator and develop something wonderful in the audio amplifier space!

But that comment also took me back to when I was a young, green engineer, designing an antialiasing filter module for a particular customer.  It turned out that the customer also needed some gain in the signal path.  "No problem, boss!", I said, "I can incorporate that in the filter".  "Kendall," he replied, "you don't understand.  That's his job.  He's got to be able to contribute something to the project or he'll get the sack".  The particular moral of that story, brought up to date, is that when you design something with PSoC3, you should really try to do something yourself, rather than just take a completely engineered example circuit provided by us (or by one of your colleagues).

And sometimes, the idea that every single component can be internal to the device is too good to be true.  I've designed quite a few active filters to run on PSoC3's analogue blocks, and they require a sprinkling of external passive components - something extra for you to add, that might differentiate your solution from someone else's.  Just recently, I've spread my wings a little, thinking what cool things could be done if you allowed for instance one external transistor.  Now that transistors are available in such tiny packages (SOT-1123 is the smallest I've seen so far, at 1x0.6mm), they take up negligible extra space and cost.  So I'm thinking of running a little 'just add one transistor" contest, to see what people can come up with.  OK, and a two transistor one as well...

And in a funny way, that's why PSoC3 is related to cake, and not just by the daft pun in the blog title.  Take the classic marketing tale of Betty Crocker cake mixes.  They got far greater engagement from their customers when they adjusted the mix formulation so that you needed to complete it by the addition of an egg.  If the whole thing came out of a packet, housewives felt rather disenfranchised from their cooking territory.  So, I'm a great fan of designing things where it's almost all done for you, but you still need to flex a few brain cells to actually get it to work out.  You actually feel like you've done something then.

My daughter gave me another example that came right out of left field.  Making a cake out of cake mix (with or without an egg) is somehow rather conventional in comparison.  Why not make your cake - or your next electronic product - out of something completely unexpected.  Meet the Ice Cream Muffin:

Ingredients: one tub of good ice cream, any flavour you want; same-sized tub of self-raising flour

Method: soften ice-cream slightly; beat in flour until mixture is smooth; bake in a medium oven until cooked.

Just as easy as a cake mix, more 'programmable', more interesting, less expected... The moral: PSoC3 might well be a solution to your problem, even if at first glance the packet seems to say something completely different.  Right, now I'm going to bake some Caramel Chew-Chew muffins for the PSoC3 team!  Happy PieSe-of-Caking - Kendall

However, human-scale reaction times are just no match for the propagation behaviour of modern social networks.  You can delete something from the Twitter public timeline, but Google seems to cache the timeline dynamically, so what you typed will get 'crawled' eventually.  Sure enough, next morning my deleted mis-Tweet turned up both as a Google Alert and in a LinkedIn summary emailed to my wife... how embarrassing!  Of course, it could have been much worse; I could have been exchanging personal information, or confidential trade secrets.  Check before you click!

But there's definitely an upside.  I get a buzz from helping people to solve their difficult signal management problems, and it's encouraging to reflect on how easy it is these days for us to bounce questions and answers around.  If you have a question on filter or system design, there are many ways for you to punch that question right through to my beloved iPhone.  Of course, everyone has some limitation to their bandwidth, I can't give an instant answer to every question.  But it's often really quick and easy to point people in the right direction with a short email; it doesn't take a magnum opus.  So, don't be shy - if you're stuck on some filtering task, drop me a line, on Twitter, or LinkedIn, or email (filterwizard at cypress dot com).  My response time will probably be inversely proportional to the interestingness of your question - the Bandwidth Theorem at work again, maybe!  Always happy to get subject requests for my Filter Wizard column as well.

Right, now back to the lovely pseudobandpass filter I'm designing for the front end of a PSoC3 solution which replaces an old chip that just suddenly became hard to get.  I can remember the days when component stores took up half a factory... now, you just need a few tubes of PSoC devices and a good "capsule wardrobe" of basic passives, diodes and transistors.  Maybe it's time to have your own "Independence Day" moment!  best - Kendall

My family were skeptical about this claim, and demanded further explanation.  With calculus off-limits in a car full of artists, and the simultaneous requirement to concentrate on the driving, I think I pulled it off - but I decided not to push my luck by setting any homework questions...

Reflecting on this later, I realized that, with a periodic modulation of the tyre tread, the frequency at which tread blocks are passing a particular point in space is being frequency-modulated, and with a unit modulation index.  The tread blocks in contact with the asphalt aren't moving at all relative to it, while the ones at the topmost part of the tyre are passing at twice the rate that an unrolled piece of tyre tread strapped to the car would do.  The tread blocks travel an approximately cycloidal path; points on the tyre's sidewall that are closer to the axis travel on a smoother curve (a curtate cycloid) that tends towards being a sinewave as you get nearer to the axis.

Does this have a PSoC3 connection?  Well, there's some interest in creating quadrature clock oscillators using the Digital Filter Block.  This will use a well-known technique - see for instance Lyons - but I'm interested in adapting the technique so that the processor can actually run on the varying clock that it creates.  Looking back on my weekend tyre experience, I now see that the equations that fall out of that look very like the parametric equations for a cycloid.  This tells us how to modify the process to correct the modulation non-linearity.

So next time you see a beat-up old truck doing something unusual on the highway, ask yourself - what does this tell me about my project?  Happy treading!  best - Kendall

When combining stop-frame animation with live performance, the live footage is back-projected a frame at a time behind the animated models.  Their small incremental movements are rephotographed together with the 'live' backdrop.  When the resulting shots are played back, you get the illusion of animated models 'acting' in perfect sync with the live cast of the movie.

As I was watching this, it occurred to me that there are significant parallels between this sort of approach and the common use of sampled data techniques in signal processing these days.  The 'live' footage is obtained by sampling the analogue dynamics of the actors at typically 24 frames per second.  When it's replayed in the cinema, our persistence of vision acts as a lowpass reconstruction filter.  We don't see the jerkiness of the sudden shot-to-shot changes, just an impression of the original dynamics (falsified occasionally by aliasing, but that's another story...).

We can do the same sort of thing in a mixed-signal processing engine like PSoC3.  The example that came to my mind was creating a little digital synthesiser to 'play along' with an incoming audio signal.  Imagine singing into a microphone which is driving PSoC3's main ADC, set to sample your analogue warblings (Bernard Herrmann's great score for the film, perhaps?) at say 48kHz/16bit.  We can make a little programmable digital oscillator running on the Digital Filter Block, combined with some envelope shaping and a little 70's synth-rock filter.  For the 20.8us that each input sample is sitting there - the equivalent of the film frame - the program running on PSoC3 is acting as a little stop-motion animator, making numerous small changes to parameters of another new sound.

Then we mix them together and send them back out - perhaps up the USB interface to your PC where you're re-recording this.  The result - a live actor (you) and a stop-motion skeleton with a sharp sword and a bad attitude (well, here's hoping the synth sounds that good), acting together in perfect sync!

Think about what other cool things you can do with PSoC3 that are  analogous (or should that be digitalogous?) to Ray Harryhausen's masterful animation.  Happy sampling! - Kendall

It's the result of a lot of work from the A-team (A for Audio, natch), including some Universal Digital Block magic from my Cypress co-blogger PSoC Sensei.  We're still cleaning up the PSoC Creator components for public consumption, so they won't be part of the distributed software for a while.  But eventually, you'll be able to drag-and-drop USB audio into your Creator schematics as easily as an op-amp or a NAND gate.  Is that cool, or what?

Next step is to get the built-in Digital Filter Block integrated and enabled.  We've got a 10-band-per-channel stereo graphic equalizer (yes, Old School, I know...) running, and it looks like we should be able to double the band count.  I got that general biquad ordering algorithm working that I blogged about last time.  So I'm looking forward to doing some high-order arbitrary bandpass filters for the comms applications people are asking about.  I'm also quite stimulated by a request from a correspondent who's trying to do a massively parallel bank of very narrow bandpass filters and detectors.  So much to do, so little time.  Actually, that can't be true, because little times imply large bandwidth, and I certainly don't have enough of that!

Oh, and we now have an email address: filterwizard@cypress.com gets you straight through (subject to our spam filters, of course).  Happy Filtering! - Kendall

If you've already started working with our graphical design tool PSoC Creator, you may know that we're already supporting FIR digital filters running on the embedded Digital Filter Block found within some models in the PSoC3 and PSoC5 family.  People regularly ask us when we'll have support for some other filter types, IIR being top of their list.  Well, we're working on providing some tools for that, of course.  But meanwhile, a few of us here get to design these filters one by one, for the applications that already benefit from them.  And the results from some of the IIR filter design tools out there (both free and expensive!) leave a whole lot to be desired.  One of my current projects needed a pretty sharp lowpass filter with good phase linearity in the audio band, so I dug up an ancient analogue design I'd done years ago, as an exercise in conversion.

Filter Wizard article readers may have spotted that I'm a fan of the bilinear transform method for migrating good old s-domain analogue transfer functions into z-domain digital ones.  A significant benefit of this approach used on a cascade of biquad sections is that the cumulative amplitude responses at the successive outputs of the biquads is preserved (allowing for the frequency 'warping', of course).  So, if you take a well-designed analogue transfer function that has been optimized to get the best dynamic range out of a cascade of analogue sections, it usually transforms into a digital design of close-to-optimum performance.

And this is the reason for my witty and amusing blog title (yeah, right - Ed.).  For it's the order of the poles and zeroes, and also the stage-by-stage gain scaling factors, that determine how much of the cascaded filter's dynamic range potential is available.  The frequency-dependent feedback built into a second order filter section has a magnifying effect on the internal noise created by that filter, whether it comes from an op-amp's input stage or from the truncation of numerical results after all that multiply/accumulating has been done.

I sort of 'cracked' this problem years ago for analogue filters when they used to be my 'day job'.  Basically you need to make sure that the signal level in the passband never gets either too big (you'll overload early) or too small (the gain needed to make this up boosts the nose as well).  Given all the poles and zeroes in a transfer function, there's generally a best arrangement which minimizes the output noise with respect to the maximum undistorted signal level.  There are some subtle differences working with standard digital biquads, and I'm looking at adapting the method to work under these conditions.  So I spent much of this morning cranking add-on functions to an old program of mine that contains code dating back over twenty years... no point in reinventing the wheel when all that's needed is a slightly different hub-cap, eh?

If I can get these tools easy enough to use (if you can get them working, you mean - Ed.), I might make them available to you on a case-by-case basis before our proper DFB support is ready.  If you're designing a PSoC3- (or PSoC5-) based system and you think you need some particularly good or critical digital filters, send your comments and requests through to... well, this column is so new that it doesn't have an email address yet, but I'm sure it soon will, and then you can use it!

That's all for this week.  Keep filtering!  best - Kendall