Generic Analog Building Blocks

[arigy]

Dear friends,

I think, at some point, the following idea visits any pro-audio DIY artist. It's like, when you building your next uber-compressor or synthesizer, you always need some standard building blocks, that you use here and there, like balanced input stage or simple +/-10 dB stereo gain control. And, instead of focusing on the core idea of your new project, you spend a lot of time on 'copying' that generic design, tracing PCBs for them etc. All of this makes our PCB larger (== cost higher), increases a chance for error (==time and effort needed to debug circuits).

And there comes an (obvious) idea. Why not design some common building blocks with simple and clean design, modding ability, intuitive connections, and, instead of replicating them in each new project, just re-use them? What do you think? Look, arduino guys has this for several years, all these shields, sensors, that does not require some special understanding of underlying processes, and the modular synthesizers has this for several decades now. I am talking not only about hi-end all-discrete stages, but also on generic circuits, found in every analog mixing desk, dynamic processors or effect unit.

And there comes my vision of these building blocks:
- Thru-hole design (where possible) - to make modding simplier.
- Screw terminals for inputs/outputs and power - no crimping, faster assembly, possibility to use almost any wire.
- Tiny panel mounted PCB - again - for simplified assembly and lower cost.

As of now, I am planning to build the following blocks: Balanced TRS Stereo Input, Unbalanced RCA Stereo Input, Stereo +/-10 dB Gain Stage, Stereo Mixer with Master Level and Balanced TRS Stereo Output. For unification I would like to use 9mm potentiometers, which are available from different manufacturers: PTD90 by Bourns, RK097 by ALPS, RD09 by Alpha, P09x by BI Technologies and P9 by Vishay Sfernice.

 

[Andy Peters]

The cost of interconnects between smaller PCBs will outweigh the cost of a larger board.
These interconnects also reduce reliability.
NRE cost is generally per-PCB, so if you have two PCBs you pay more NRE.

Then there's the issue of mounting all of these little boards ...

 

[arigy]

The cost of interconnects between smaller PCBs will outweigh the cost of a larger board.
These interconnects also reduce reliability.
NRE cost is generally per-PCB, so if you have two PCBs you pay more NRE.

Then there's the issue of mounting all of these little boards ...

If we are talking of production - you are absolutely right. But sometimes we need a-proof-of-concept type design, or some other reason not to spend days on routing bigger board rather than get the result yesterday As of interconnects: any interconnect can fail with no exceptions. And out of many IDC or crimping connectors, I have found screw terminals the most reliable and requiring no special equipment (that's why there are widely used in Pro Audio public address equipment).

 

[JohnRoberts]

Andy is correct even for prototyping, but how about an observation and a suggestion.

Observation.. 1- What used to be useful building blocks like balanced inputs, balanced outputs, mic preamps, etc. That formerly involved multiple opamps and discrete circuity are now often available in dedicated single ICs or chip sets (like a mic preamp and  programmable gain control). .

Observation #2- pretty much all audio products require inputs, outputs, power supplies.

Suggestion #1- Build a couple balanced inputs, balanced outputs, and a power supply on one prototype PCB.

Suggestion #2-  Build an area into the generic prototype for circuits not yet envisioned. Perhaps some popular dip opamp footprints, switch and pot foot prints.


or not...

JR

 

[ramshackles]

It would be an interesting idea for newbie DIY-ers like me, in order to learn something more about creating/desigining/what goes into a unit.

For example, choosing an input from a variety of selections, an output and whatever goes in the middle...different eq sections for example.

It would allow a newbie to design & build from a 'block' diagram point of view.

However, it would be very expensive to do, compared to buying a single circuit board for a specific project.
I imagine it would also be a lot of work to make sure all these 'blocks' can be matched up, so they can be interchanged at will (impedance matching or whatever...).


However, rather than actually making the 'blocks' why not provide the schematic/pcb layout 'blocks'.

Am I right in thinking most schematci/layout editors allow you to save circuits as sub-circuits or components.
So you could create your line input and save it is a component, to be loaded into whatever project you are doing...

 

[JohnRoberts]

It would be an interesting idea for newbie DIY-ers like me, in order to learn something more about creating/desigining/what goes into a unit.

For example, choosing an input from a variety of selections, an output and whatever goes in the middle...different eq sections for example.

It would allow a newbie to design & build from a 'block' diagram point of view.

However, it would be very expensive to do, compared to buying a single circuit board for a specific project.
I imagine it would also be a lot of work to make sure all these 'blocks' can be matched up, so they can be interchanged at will (impedance matching or whatever...).

Have you actually quoted the cost of prototype PCB?

Generally there are fixed minimum costs that make the economics of buying one small PCB not much better than several small boards or a couple larger ones. One vendor I used priced prototype based on how many you could get from a standard PCB panel. 

However, rather than actually making the 'blocks' why not provide the schematic/pcb layout 'blocks'.

I used to manage a design engineering group for a decent sized company and that is the goal. Not to reinvent the wheel for every product but cut and paste proved circuit blocks together.

Unfortunately I found many products required revisiting the wheels, but that was always the goal to improve design efficiency.

Am I right in thinking most schematci/layout editors allow you to save circuits as sub-circuits or components.
So you could create your line input and save it is a component, to be loaded into whatever project you are doing...

Better ones should, I never worked with a modern SOTA CAD system, I worked with some very early very crude stuff decades ago.

JR

 

[ramshackles]

Have you actually quoted the cost of prototype PCB?

Generally there are fixed minimum costs that make the economics of buying one small PCB not much better than several small boards or a couple larger ones. One vendor I used priced prototype based on how many you could get from a standard PCB panel. 

Yes and you are knocking on $100 or more for a couple of PCB's. But I read it like the OP was thinking of making these available for purchase. In which case, if ordering just 50 at a time, then you are down to about $5 a board...

Still expensive compared to single board projects...I imagine the smallest number of boards you would need to complete something would be 3 based on what the OP said (an input, output and whatever inbetween), which would be $15 manufacturing costs at least...


Anyway...isn't that idea similar to what is already available from JLM? (dINGo kit especially, MAC compressor kit, & other 'small interface' kits)

 

[JohnRoberts]

I exited the kit business 30+ years ago because the economics didn't make sense to me. Consumers could buy assembled mass produced products cheaper than I could sell them parts.

Back then I sold some small building block sub kits, a +/-15V power supply, basic HP/LP filter, variable gain stage, etc.

The economics of DIY kind of doesn't really count, while sometimes it is the only way to make obscure one of a kind franken-products.

JR

 

[bruce0]

JLM audio in Australia has done some building blocks.  They use ribbon connector sized interconnects.  While they are not documented clearly as to how to arrange them he has a "go between" which provides phase/phantom and pad switching, and a DI interface. And he has various input and output buffers in different formats (including 500 series )... and other interfaces.  And psu cards.

I think it is a good idea, but in general I think the big time saver is 500 series which almost eliminates metal work, but also makes space an issue and thus make modular cards difficult.  Although I have been thinking about stackable cards that might work in 500 series format. 

Standards can save a lot of time. How about 500 series L bracket carriers with standard hole placements.

 

[dfuruta]

If you could keep each pcb rather small (just a few square inches), it wouldn't be very expensive to have some made at batchpcb or an equivalent service.

 

[JohnRoberts]

IC manufacturers provide lots of application support to improve chances of winning design-in's. For ICs where layout is critical (switching PS, A/D convertors. etc)  they will sometimes offer layout example advice.  If there were easily portable PCB blocks that could be integrated into larger designs, it would be a no-briner for the IC makers to supply them as electronic files.

I guess there is still some tower of babel, poor communication between different cad systems and difficulty to seamlessly integrate circuit blocks together. Components would need to be renumbered obviously, so BOM to node list, to board intelligence for part placement drill programs etc would need to be coordinated. 

Commercially such design data as a computer file would be hard to protect and sell for a profit. So makes most sense for manufacturers to do it free as a promotion to sell ICs.

JR

 

[bruce0]

John:

I agree with you, but I think the tower of babel is not that high, just that IC manufacturers can't climb or just are oddly not interested in the prototyping market. 

They don't even post footprints in scannable form, never mind footprints in Eagle, Kicad, and other formats, for their parts.  This seems the most reasonable thing (and on some of the more complex footprints would save a ton of time. ( I labored for hours trying to get a decent heat sinking version of a QFN preamp that THAT makes.  )).

IC makers should post footprints.  Perhaps in an intermediate format that designers could convert to their platform of choice.

 

[Andy Peters]

John:

I agree with you, but I think the tower of babel is not that high, just that IC manufacturers can't climb or just are oddly not interested in the prototyping market. 

That's because they're interested in the production market, where the products are used by professionals. Which shouldn't be a surprise.

I think there's a difference between what I would call a prototype -- a "first-article build" -- and what the hobbyist calls a prototype.

They don't even post footprints in scannable form, never mind footprints in Eagle, Kicad, and other formats, for their parts.  This seems the most reasonable thing (and on some of the more complex footprints would save a ton of time. ( I labored for hours trying to get a decent heat sinking version of a QFN preamp that THAT makes.  )).

What's a "Scannable form?"

Well, the professionals to whom they primarily sell don't use Eagle or Kicad. But many of the vendors do make footprints and symbols available for professional packages such as Altium and Cadence and Mentor.  I know that Altium has vast libraries of pre-made parts with symbols and footprints, too.

Perhaps the "community" that supports Kicad and Eagle has footprints you want? I've been playing with Kicad, and the kicad library website has quite a few footprints available. And creating a footprint in Kicad isn't all that hard.

IC makers should post footprints.

They do. And remember that the footprints are all generally JEDEC standards, and there are plenty of footprint calculators and creation tools available.

See here for Analog Devices.

See here for TI.

See here and here for Linear Tech.

See here for ON Semi.

Perhaps in an intermediate format that designers could convert to their platform of choice.

There's no such thing as an "intermediate format."

-a

 

[JohnRoberts]

I've watched the evolution of CAD and CAM integrating CAD into manufacturing. It is hugely better today than it has ever been.

Cut and paste design blocks, have long been an industry goal but you still may have issues cutting and pasting from MAC to a PC, but who knows even about that..

Getting back to my suggestions if you combine input, output , and power supply blocks onto a single PCB, or even if you don't, be sure to accommodate differential referencing between sundry local grounds since it impractical to send fully balanced signals around inside products.

JR

 

[Brian Roth]

This sort of reminds me of the early/mid 1970's when I first began building "gizmos" for sale.  While as a staffer at a pro audio company, I began building active crossovers for the various permanent/mobile PA systems which we sold.

In that case, I designed up a single PCB for various filter configs....mono 2-way, stereo 2-way, mono 3-way,  etc.  It was just a card which allowed various active filter circuits, and I added them as needed.  But, a separate card held the PSU stuff, so I could mix n' match.

A bit later, I also drew up a card which was a "universal balanced I/O" for concocting other gizmos, without reinventing the wheel.

The "other gizmos" were typically built on perf board, but I used the "Lego" elements (balanced I/O card, PSU, etc) as well.

Of course, back then, all of the PCBs were done at 2X using Bishop Graphics "donuts" and crepe tape stuck onto clear mylar film, which was then "shot down 50%" to actual size (1X) via a "stat" camera onto positive or negative litho film.

I did the same for litho film used to make photo silk screens in order to make panel legending.  I distinctly recall finding the MAN here in OKC who had a "stat" camera which could create lithos which were 19"...or wider....for panels.  IIRC, his litho camera and film stock went up to 30" (?????).

Wow..I'm now recalling such arcane info such as "emulsion side up", etc  etc........tis been a LOOONG time...LOL!!

I also used to make up my own silk screen frames, and breathed the silk screen ink and "screen wash" fumes as I printed various panels....

Perhaps why I am a bit bit bit bit....errrrrr........"thetched in the head now"....<g>

I also hear the sounds of "dinosaurs farting" as I mention all of this...LOL!!!!!

Bri

 

[JohnRoberts]

This sort of reminds me of the early/mid 1970's when I first began building "gizmos" for sale.  While as a staffer at a pro audio company, I began building active crossovers for the various permanent/mobile PA systems which we sold.

In that case, I designed up a single PCB for various filter configs....mono 2-way, stereo 2-way, mono 3-way,  etc.  It was just a card which allowed various active filter circuits, and I added them as needed.  But, a separate card held the PSU stuff, so I could mix n' match.

A bit later, I also drew up a card which was a "universal balanced I/O" for concocting other gizmos, without reinventing the wheel.

been there done that... A few of my kits back in the day used the exact same power supply board-sub assembly. Later kits used an external lump and integrated PS into main PCB. 

The "other gizmos" were typically built on perf board, but I used the "Lego" elements (balanced I/O card, PSU, etc) as well.

Sometimes features like bypass and turn on/off muting are added on as daughter boards instead of redesigning whole thing.

Of course, back then, all of the PCBs were done at 2X using Bishop Graphics "donuts" and crepe tape stuck onto clear mylar film, which was then "shot down 50%" to actual size (1X) via a "stat" camera onto positive or negative litho film.

Yup... I recall even seeing a rubylith(?) a similar oversized artwork for metallization (top) layer of semi-custom IC, reduced using similar photographic reduction to make masks. 

I did the same for litho film used to make photo silk screens in order to make panel legending.  I distinctly recall finding the MAN here in OKC who had a "stat" camera which could create lithos which were 19"...or wider....for panels.  IIRC, his litho camera and film stock went up to 30" (?????).

I recall them being called "copy" cameras. I have one ex-friend who may still have one in his basement. Back then many print shops had one. Good practice rule #1 about generating artwork that way.. provide large easy to read and measure marks clearly indicating final output size, the larger this dimension the more accurate the reduction. Copy camera reductions are continuously variable and getting a PCB several percent too big or too small is never fun. 

This was also a handy way to improve the quality of small ads and artwork before the days of high resolution computer printers. A crude dot matrix printout reduced 4:1 or 6:1 cleans up nicely. 

Wow..I'm now recalling such arcane info such as "emulsion side up", etc  etc........tis been a LOOONG time...LOL!!

Negative used emulsion down to the raw photo sensitive PCB or screen material for most accurate transfer. The thickness of the film stock can make a tiny error if between the image and surface being exposed.

However when the negative is shot in the camera you specify emulsion up or down so it ends up on the correct side for later when negative is finally used for transfers. IIRC that was specified as something like "right reading emulsion side down" Right reading refers to the words on the artwork being readable and not upside down.

I also used to make up my own silk screen frames, and breathed the silk screen ink and "screen wash" fumes as I printed various panels....

Silk screens ? Fabric or stainless steel screens? SS was better.  Silk screening was a great mass production tool for it's time but had it's accuracy limitations.

I may have told this story before but one of the several production issues I wrestled with at Peavey was minimum design rules for fabricated in house PCB. Peavey used something like 24" wide screens, in automatic screening machines to image standard PCB panels. The minimum dimension of spaces and traces depends on the accuracy of that pattern transfer across the 24" span.

I ended up in the middle of a pissing match between the production team in the factory who had to make these, and the director of digital engineering and his top PCB designer who was the self-appointed "expert" on making PCB, mainly because he had visited several outside PCB shops when buying double-sided digital PCBs (we only made single sided PCB in house).

The crux of the argument was that the factory was lobbying for temperature and humidity environmental control for the PCB work area to improve yield. In MS summer the temp and humidity swings were huge. The self-appointed expert opined it wasn't necessary, because the PCB house he visited often didn't use specific environmental management.

I tasked the PCB layout designer who worked for my (mixer) group to get me the technical specs for the screen material, not the screen itself but the photosensitive layer attached to the screen that dominates mechanical stability. Then I asked the factory to get me the range of temperature and humidity the department see's on a daily basis.

After I got the specs, and even ignoring some extra variables in the equation I couldn't easily come up with, it took less than 5 minutes to do the math that proved it was impossible to hold our minimum dimension (20 mils IIRC) over that 24" span with that much temperature and humidity variation daily. 
 
This made the factory my new best friend, and a reinvigorated political enemy out of the digital director I proved wrong, but that wasn't a new theme for him or me. It turns out his outside PCB shop that didn't need environmental controls was in southern FL and air-conditioned much of the year. The near constant air-conditioning kept temp and humidity variation within a small manageable range.

Perhaps why I am a bit bit bit bit....errrrrr........"thetched in the head now"....<g>

I also hear the sounds of "dinosaurs farting" as I mention all of this...LOL!!!!!

Bri

You are in the company of other dinosaurs...  8)

JR

 

[Andy Peters]

Of course, back then, all of the PCBs were done at 2X using Bishop Graphics "donuts" and crepe tape stuck onto clear mylar film, which was then "shot down 50%" to actual size (1X) via a "stat" camera onto positive or negative litho film.

When I was in college ('84 to '88) , a lot of the engineering profs allowed students to bring in a single 8.5" x 11"  double-sided sheet of paper for formulas.

I had a job working part time in the school's Publications department, so I got to know the guys in the print shop. And they had a stat camera and taught me to use it.

So for exams I'd get a big piece of poster board, draw my formulas on it, and then take it to the print shop and shoot it down to letter size. Perfectly legible but kinda small (I had better eyesight then). My friends were simultaneously impressed and pissed off.

Then there were the times when I forgot to bring a regular pencil or pen to class but I had a Non-Photo Blue pen for taking notes. That would annoy friends who didn't attend class but wanted my notes.

-a

 

[PRR]

> then "shot down 50%" to actual size (1X) via a "stat" camera

Copy camera.

They had a couple at Mom's newspaper. Output was full width of a newspaper web. That rig was not infinitely variable.

Dad knew a guy retired from that then set up his own copy-camera in the basement. Did a lot of small-job work. Often adjusting oversized ad masters down to a part-page size for the local publishers.

Later I worked in another paper shop. We got the full-page negatives, laid them on coated aluminum, flashed that with an arc-lamp, then scrubbed the plate with this smelly pink lotion. The emulsion came off according to exposure. However if a plate was not coming up crisp, we could scrub and scrub and maybe get something which would print. (Climate control was an issue.)

Those copy-cameras went out of style real fast. When I worked near an art school, one of the techs was *thrilled* to get a large used camera "for only $20,000!!". Well, yes, it cost 4X more a few years before, but I knew where the industry was going and that it might already be worth zero. Also that 2-year grad students would never gain proficiency in copy-work (it's a real skill). And that it might be hard to get supplies. Indeed they putzed with it for a year, had nothing to show, and the tech was not-rehired. Couple years later they paid $2K to get the beast hauled to scrap.

 

[tubegeek]

Ouch.

My first job out of college was with a "multimedia" production firm. We made, at great expense and effort, what today is done easily in Powerpoint or Keynote. Back then it took a small room full of graphic artists, a stack of computer-controlled Kodak carousel slide projectors, and a big budget. I learned how to register multiple projectors into a panorama among other esoteric skills.

Lots of artwork going to and from copy camera services, oh the traffic management, oh the expenditure! And the inevitable changes at the last minute were a huge deal to accomodate. And navigating between the dinosaur footprints was indeed a dangerous reality.