Analog's answer to FPGA opens field to masses (2008) | Hacker News Reader

Analog's answer to FPGA opens field to masses (2008)

eetimes.com

78 pointspointfree39 comments

Fast forward 6 years and it looks like they are available from a company called Anadigm ( http://www.anadigm.com/fpaa.asp ).

This sounds like it could be a great fit for Software Defined Radio.

Dev kit is $200. Chip is listed at Arrow for ~$10, but out of stock.

I wonder if you could do some interesting analog audio stuff, like building low part count multi-speaker crossover networks, etc. Then again, for $10 apiece, it probably still makes way more sense to just put a cortex or something on there and do it all digitally.

You could market it to the analog-or-nothing crowd in the audio community. Decreasing production time of "boutique" analog clone effects pedals could make you some cash.

That would be neat— a reconfigurable effect pedal that you could download and install new patches for.

The _real_ trick would be providing the option to route the signals out of this thing and through vacuum tube stages, so you're not stuck trying to build a decent overdrive or compressor with silicon-only components.

You can buy the chips directly from them, for around ~$10 each.

They sell the dev kit for the current AN231E04 part, but at least the Asian distributor does not stock it (only back order), and over $200, but at least it is available from 1. If I'd just want the part, it comes in 1000 increments...

I'd have some more ideas for lab instruments, but under these circumstances it cannot really be used for product development, sadly

I just found that the part is available through distributors on alibaba one of them even selling in min orders of 1.

http://www.alibaba.com/trade/search?fsb=y&IndexArea=product_...

With all the fake electronic parts coming from china, are those companies a trustworthy long term source for such companies ? Not sure.

One common advice is to buy from authorized sources, and looking through octopart[1], arrow is the only distributor of anadgim, with 1000 units minimum quantities.

[1]http://octopart.com/partsearch#!?q=anadigm&sortby=avg_price%...

Rough estimates suggest there are around 3,000 analog engineers in the world

That's the real order of magnitude of the number of analog engineers?

Analog designer experience and skills are akin to "black magic", at least as whispered in the halls of electrical engineering schools 15 years ago.

Analog is just devilishly harder when compared to digital IC design; forget about hysteresis curve cutoffs and resulting 'clean' abstractions of 0 and 1.

Analog silicon designers? Maybe, although it may be underestimating the number outside the West.

It's much less prevalent than digital logic, much harder, and slower moving industry as it doesn't benefit from Moore's law in the same way.

I don't think they are talking about silicon designers.

The article quotes the professor as hoping FPAA's will allow "a wide range of less-skilled users to try out sophisticated, low-power (analog) techniques".

In other words, typical device-level design engineers.

Right, that's his target market, not his existing analog engineer population.

Very few people use analog signal processing techniques these days.

It used to be common up until about 10-15 years ago, but then DSP's came along.

Now, you take a signal, buffer it by the minimuum amount necessary to get it into the DSP, and call it a day.

The best case is, you don't touch the signal at all, because any caps, filters, op amps can all distort the original signal.

You can do it all digitally, and it's much easier to modify on the fly if your technique isn't working well.

Boards and boards full of touchy analog electronics have been replaced by a single DSP.

And not many people are looking back.

Signal conditioning is still an important part of EE design. Not all signals are TTL level and unipolar.

For example, this appnote[0] uses a Delta-Sigma converter to digitize a load cell, with the ability to "detect one sheet of copier paper dropped on a stack of 20 reams". In order to do this, it relies on an intrumentation amplifier and custom correlated-double sampling to get a measurement accurate to 200,000 counts.

And even if your input signal can be directly input to your ADC, you'll still need an analog anti-aliasing filter.

Signal conditioning is an important part of signal processing that has to happen in the analog domain.

[0] http://cds.linear.com/docs/en/application-note/an96fa.pdf

I agree that we haven't yet eliminated all analog conditioning.

Your example however, doesn't really bolster your point.

Delta-Sigma modulation is a digital signal processing technique.

That particular chip is also using a digital FIR filters to reduce noise.

You still need analog engineers for SoC chip design including voltage regulators and detectors, PLL and clock generation, IO pads.

You don't get that much analog per chip yet. See

http://www.anadigm.com/_doc/DS231000-U001.pdf

Basically, it's four op amps, plus programmable circuitry for all the accessory stuff needed to build most of the things you can do with four op amps. This is neat, but four op amps isn't much analog computation. Here's the block diagram of a classic, very successful analog control system: the F-16's original stability control system: http://thelexicans.files.wordpress.com/2013/07/yf16fcs.png

So this isn't for analog computation, it's for front-ending digital systems where you need to do some filtering at a faster speed than a DSP approach can handle.

The Cypress PSoC line seems to be a whole ARM processor with a little analog stuff added.

Too bad the market is still tough for low-volume FPAA's (vs. FPGA's). A product I was working on last year desperately needed FPAA's in low volumes (think 1k-2k MOQ), and was shelved largely because the lack of market availability.

The design was changed to reflect the lack of availability of a low-volume FPAA at a reasonable price point, but this drove the feature set down and the price up until the product was shelved. Until they make them as easily available and as low-cost and many comparable FPGAs, it's going to be hard to design niche-market products around them.

What's the difference between a FPAA and the programmable analog blocks on products like Cypress' PSOC? Those chips are pretty common.

In our experience, the primary issues was the available number of analog I/Os, we were doing large-scale matrix-like switching between analog sources. That is to say, the PSoCs contained a small number of analog I/Os that could be routed relative to a "dedicated" FPAA.

This article is from 2008. Is there any particular reason its here?

This. I was about to bring up the Anadigm source and then saw the date. Not sure why this is news since there are products in the market and demand scale has not surfaced to help ramp up the supply availability and product variations that could make it a healthy market.

I was expecting this to be about Silego products since that's what was featured in a recent newsletter. All of these parts seem pretty niche though.

I recently bought some FPAAs from Anadigm, specifically the 3.3V version: AN231E04-E2-QFNSP

Interestingly they use a switched-capacitor design, which means you can only input signals lower than the clock frequency used for switching the caps.

I had a look for continuous-time FPAAs but couldn't find any.

My goal is simply to make a guitar FX pedal with them, as they can alter gain, perform filtering such band-pass etc.

Potentially I could see it having lower latency than DSP, if anyone knows more about that, I'd be very curious.

This may be completely not what you're looking for, but take a look at Cypress' PSOC devices. They have an interesting mix of analog and digital programmable blocks.

If you are interested, I did some research on extremely fast, lower power, and low noise filtering using FPGAs or CPLDs.

I am interested if you are willing to share!

https://news.ycombinator.com/item?id=8592292

Link to my reply. :)

Sure, that sounds cool! (I've got a cheap FPGA dev board that's got an ADC/DAC on, that I wouldn't mind putting to good use)

Most of the research work was targeting a specific development paltform, so I don't have good instructions for doing it generally. Here is the paper we based our research off of: http://www.eecg.toronto.edu/~johns/ece1371/papers/1993_lewis...

Here is my presentation poster which should explain the motivation and theory a little more clearly than the paper: https://www.ece.ucsb.edu/academics/undergrad/capstone/event-...

I graduated before we got it releasable, and haven't had much time with my job to finish up the project. I don't have more time to add to this explanation, but if you want more info or implementation help, feel free to email me: alecdibble at gmail dot com.

You can do digital control of analog circuits, PGA, digital pots, etc.

Just had a fun idea, you could use an LED to charge a rotating photoluminescent medium and then read out the intensity some rotation angle later, it be like a magnetic tape loop but have its own decay as well. Modulate the speed of rotation, the read angle and the intensity of the write.

http://www.identi-tape.com/photolum-tape.htm on the edge of a turn table

Heh, that's a really cool idea!

I wonder if you could use it to build a synthesizer. That would be amazing - a modular analog synthesizer on a chip.

I believe I read a while that they demonstrated it in a synthesizer application at a music expo, and I've heard rumors that they are already used in some commercial analog synthesizers.

Audio-synthesis has been using these kinds of technologies for decades. Its not unusual for synth/audio-processor manufacturers to make their own custom ASIC's, FPGA's, even small-quantity chipfab runs, to provide their IP to the customer. Synthesis is very promiscuous in this regard; you can make sound with a surprising number of things in the universe.

I well aware of the prevalence of the technologies you list in the synthesizer industry, but custom ASICs or FPGAs are quite obviously not the same thing as FPAA.

If there's a way, the synth nerds will use it.

Two of the sub-comments here mention them, but anyone interested in this sort of chip should check out Cypress's PSoC line.

Reply on news.ycombinator.com