PSOC 5LP Hat for a Raspberry Pi

I am the original designer so I have a biased opinion. Hoping maybe I can answer some of the questions/objections above.

Is it worth the money?

It is a bit pricey at $50 (for early backers). The PSoC chip goes (as noted) for around $7. The PCB costs a few dollars (in the under 100 volume), the connectors and hat EEPROM cost another few dollars. It costs around $10 additional to assemble and test the card. So the selling price of $50 is around 2-3x the costs. That's high if it was made in volume and purchased from China. It's not. We build them in my living room. I am going to build around 50 or so of these cards (maybe 100 of them if the Kickstarter goes well) so there's not all that much economy of scale involved. In the end I might end up with $1000 or so profit which will go into my next project.

Why hook up all of the Raspberry Pi GPIO lines?

Think about this one a bit. If you want an input pin connected just connect it inside the PSoC to one of the I/O pins. Same goes for output pins. Just route them through the PSoC. What about Bi-directional pins? Someone mentioned One Wire. The Pi really stinks at interfaces which have specialized timing. You get a packet in on the Ethernet and all of a sudden you are preempted - ouch. That's where the beauty of the PSOC comes in handy. The ARM processor inside the part does a great job in those situations. Try handling the timing of a chain of NeoPixels with the Pi. It's not a great bit-banger. The ARM CPU in the PSOC is great at bit-banging.

Is this just a Marketing Ploy?

Hardly, I'm horrible at Marketing. I just put up projects that I do for myself. If someone else made the card for $30 or $35, I'd buy it and not build it. I think an entire University class could be built around this card. Can you imagine the possibilities of breadboarding a fairly complex design with minimal breadboarding?

But I can just buy the $10 Dev Kit from Cypress!

Of course you can. We did too. That's what got us interested in the part to begin with. Is it a Raspberry Pi Hat or is it a breadboarding tool? If you want to do breadboards, it is just fine. If you want to build something beyond a breadboard that's another story.

Where's the innovation in this project?

That was a real challenge. Took me a couple of weeks of solid Engineering time to figure out just how to program a PSOC from a Raspberry Pi. You see the limitations of the Pi are the unpredictable timing. The Pi doesn't make a great bit-banging programmer. If someone else had already figured out how to program the part, I would have just used their code. Nobody did, though. Took pouring through the Application notes and the answer was there. But not easy to figure out. And what did we do with the answer? We put it up on our github site for all to see. Some clever stuff there? Surely someone else will come along and say that they would have done it better/differently. But they didn't do it. We did.

Doug at Land Boards, LLC (just a guy and his sons who do stuff together).

I am the original designer so I have a biased opinion. Hoping maybe I can answer some of the questions/objections above.

Is it worth the money?

It is a bit pricey at $50 (for early backers). The PSoC chip goes (as noted) for around $7. The PCB costs a few dollars (in the under 100 volume), the connectors and hat EEPROM cost another few dollars. It costs around $10 additional to assemble and test the card. So the selling price of $50 is around 2-3x the costs. That's high if it was made in volume and purchased from China. It's not. We build them in my living room. I am going to build around 50 or so of these cards (maybe 100 of them if the Kickstarter goes well) so there's not all that much economy of scale involved. In the end I might end up with $1000 or so profit which will go into my next project.

Why hook up all of the Raspberry Pi GPIO lines?

Think about this one a bit. If you want an input pin connected just connect it inside the PSoC to one of the I/O pins. Same goes for output pins. Just route them through the PSoC. What about Bi-directional pins? Someone mentioned One Wire. The Pi really stinks at interfaces which have specialized timing. You get a packet in on the Ethernet and all of a sudden you are preempted - ouch. That's where the beauty of the PSOC comes in handy. The ARM processor inside the part does a great job in those situations. Try handling the timing of a chain of NeoPixels with the Pi. It's not a great bit-banger. The ARM CPU in the PSOC is great at bit-banging.

Is this just a Marketing Ploy?

Hardly, I'm horrible at Marketing. I just put up projects that I do for myself. If someone else made the card for $30 or $35, I'd buy it and not build it. I think an entire University class could be built around this card. Can you imagine the possibilities of breadboarding a fairly complex design with minimal breadboarding?

But I can just buy the $10 Dev Kit from Cypress!

Of course you can. We did too. That's what got us interested in the part to begin with. Is it a Raspberry Pi Hat or is it a breadboarding tool? If you want to do breadboards, it is just fine. If you want to build something beyond a breadboard that's another story.

Where's the innovation in this project?

That was a real challenge. Took me a couple of weeks of solid Engineering time to figure out just how to program a PSOC from a Raspberry Pi. You see the limitations of the Pi are the unpredictable timing. The Pi doesn't make a great bit-banging programmer. If someone else had already figured out how to program the part, I would have just used their code. Nobody did, though. Took pouring through the Application notes and the answer was there. But not easy to figure out. And what did we do with the answer? We put it up on our github site for all to see. Some clever stuff there? Surely someone else will come along and say that they would have done it better/differently. But they didn't do it. We did.

Doug at Land Boards, LLC (just a guy and his sons who do stuff together).

Hello Doug,

despite the fact that it is my opinion that this post was the worth you wrote it just for first after the presentation article of your project, I consider one of your point of special interest:

It is a bit pricey at $50 (for early backers). The PSoC chip goes (as noted) for around $7. The PCB costs a few dollars (in the under 100 volume), the connectors and hat EEPROM cost another few dollars. It costs around $10 additional to assemble and test the card. So the selling price of $50 is around 2-3x the costs. That's high if it was made in volume and purchased from China. It's not. We build them in my living room. I am going to build around 50 or so of these cards (maybe 100 of them if the Kickstarter goes well) so there's not all that much economy of scale involved. In the end I might end up with $1000 or so profit which will go into my next project.

IMHO this changes the perspective as (at least by my point of view) the kickstarter phenomenon - and all related backing sites like indiegogo etc. - has none or very low trustability to my eyes. The product - as the facts are - remain almost expensive but probably the most important aspect is that there is no mention of the building details nor the possibility to hack the original version, opening at least a part of the project (or, better, the entire design idea). I think anyway that, maybe with the help of this community a way to make it cheaper with the same reliability can be found.

To be honest I saw the video presentation in your kickstarter page and this was almost evident the discordance with the Kickstarter presentation standard of the other electronic projects / gadgest: all the other are so nice and perfectly done that the first question I ask to myself is always the same: if these guys invested so much creating a presentation of something that does not exists, why don't they save their money and make the project the same? Just the opposite of your case.

About the second point of special interest, where you explain your reasons routing all the 40 PI pins, this sounds reasonable to me but frankly I'd like to have some more details. But I am almost sure that your post got a boost to this discussion and these things maybe clarified soon. If you like this, obviously.

Enrico

I don't know. I think I'd feel guilty charging 2-3 times cost for something that I knew was very similar to a dev-board that was one-fifth of the cost unless there was significant value-add.

Especially if you're targetting university classes - in that case isn't it better for them to save money and use the dev-board for a fifth of the cost?

I understand that you went to effort to port a programmer to the 'Pi, but Cypress Semi created the code that it was based on. Many of us here have ported code and share it

for free too.

I still don't get why you'd connect all the Pi's I/O to your project. The reasons mentioned would all still apply even if you just connected two or three I/O between the two boards.

Doug

Well done for fronting up and answering the questions posed.

We've seen many others simply hide and hope the questions go away and the other backers don't see them.

I understand the need to ensure you dont loose money, but I do question any add-on which costs a lot more than the board it's going on without adding significant extra hardware.

I have the same issues with Freedom Boards and they are a large corporate who could produce a simple prototype board for a few $.

Your breadboard v Hat comment is right, but this site is about Engineers and makers so that argument will always appear unless the cost makes it go away.

Good luck

I won't be backing it ... because I don't have a use for it.

Mark

Trying to understand - but not agreeing - I think that the question is to route (if I am not wrong he means managing pass-through) the PI on the PSoC. The point is that this is an apparent benefit, as the most powerful machine (that in all my project drives the PSoC) is just the Raspnerry, not the PSoC itself.

About the cost, I should say that what I have explained about the cost in my previous post is valid except the multiply factor; btw I suppose that this depends on the fact that the creator is anyway in a commercial perspective where his job (done at home if I am not wrong) is valuable and part of the product cost.

Enrico

Hi Enrico!

Everyone may implement it differently I suppose, but personally I'd route it all using a few pins, e.g. SPI or UART. For example, there might be 5 million phones in London, but there are not 5 million fiber-optic cables connecting London to NY.

I also get the time is valuable issue, but (personally, I don't expect others to feel the same way) I don't like projects that have a risk that others could be unaware that near-similar functionality products are available at far lower cost. It is the same reason I dislike supermarkets that offer a 500gm bar of soap at a higher cost than two 250gm bars of soap, i.e. relying on people not noticing, and the people that are affected the most are those who are poor at sums. I'm not saying that is occurring here intentionally, it is just an example. Supermarkets do it intentionally as far as I am concerned.

Enrico,

We could probably have as much as 2-3 times the backers if we sold the card at $34.00 (less than the Pi which is purely a psychological point). But would that make any actual economical sense? We've done more than a dozen successful Kickstarters and a few unsuccessful one as well. One of the successful ones, the I2C bridge card, costs us about $2 in parts but we Kickstarted it at around $25. Did pretty well economically. Didn't sell nearly as many cards as if we had a lower price but that's OK. We didn't have to build/ship as many cards either.

Other cards we've done with very little margin. Some folks just back because that is what they do (back hundreds of projects) and others back because the card does something that they've needed but have not found another card to do. Some of the projects have been similar to something that someone else did in a more breadboard form and we turned it into a field-able product. We've played with price points and product margins all over the place. It is easy enough to see the history of what we did here. We do cards because the folks at places like Adafruit and Sparkfun haven't seen the market for them. When our cards do well, we see Adafruit and Sparkfun do similar cards not long afterwards.

If someone has actual some marketing sense I would love their input. We have no marketing skills. I've been tempted to ask at the local college in the Marketing Dept if we could hire a cheap intern. Our video production quality (as you aptly noted) shows that we need help..My sons and I are two Electrical Engineers, one Computer Scientist and a High School kid with aspirations of Computer science. Oh, and a teenage daughter who does some shipping if she wants some extra money to go to the corner store and buy a Monster drink.

Why all 40 pins?

Well that one is pretty simple. We have male pins which end up getting wired to female cable pins which are typically routed to external hardware. There's not enough room above for us to mount a card above it so why not use all of the pins? We are not going to have a pass through female connector for mounting yet another card above ours. We did that in some particular examples like our Console Card for the Pi. It made some sense there. Problem is people order it that probably don't really need it and it is a pain to accommodate it. If we have a stackable card then we can't ship in a padded envelope, we have to use a box. That mattered more before the USPS messed up our entire system. So, if there's a card stack we are at the top. The vertical pins make it so.

What's your gripe with the USPS?

Used to cost is around $7 to ship anywhere in the world. They charge by weight so the difference between a 3 oz padded envelope and a 6 oz box mattered. Not anymore. We can all buy something from China for $1 shipped postpaid and yet we can't ship 1 oz anywhere in the world for less than $13.50. Used to cost half. Cut out most of half of our narrow market. I hate charging $15 to ship worldwide. I don't care that I now get 1-8 ozs for the same price. Who wants a $20 card for $15 in shipping? Someone that has the power needs to get a clue.

US shipping is not so bad. Was $2.30 for a few oz. Now it's $2.60.  Why are we letting foreign shippers get such cheap prices in our own country? Sounds like suicide to me (enough political message, but this is real stuff).

Doug at Land Boards, LLC0

University classes often use National Instruments hardware. They are way more expensive than this solution. This is not what we are marketing to. If it was we would have tried to reach that market. I give it as an example since I have a son at University in Electronic Engineering and the hardware they use is far more primitive and expensive and much less state of the art than this.

I am not hoping that others don't see the breadboard/dev kit that Cypress offers. I am hoping that they do and want something that is married to their Raspberry Pi in a way that doesn't require a bunch of messy wiring and helps move them a step closer to a product and away from a breadboard. Our motto is "taking people beyond breadboards". Yes you could solder the pins on a Cypress Dev module and use a breadboard and for some who just wanted to breadboard something get close enough for someone who is playing around with a breadboard. And, it's still a breadboard.

Now try the next step. Take the Bill of Materials for the Cypress Dev module and calculate the cost it would take YOU to produce that card. What do the two chips cost? What does the PCB, the connectors, etc go for? Cypress has a great price on the dev kit hoping that people will use it to develop an embedded product. You can't reproduce the Cypress Dev kit at 3-4 times the price. Great deal for the developer who likes breadboards. Not good for a real deployable product.

So where, other than the breadboard, is there a comparable product that is near the cost? The Sparkfun PSOC card is $50 and it's claim to fame is that it has Arduino headers.

Doug at Land Boards

Powerful is relative. The Raspberry Pi certainly is faster at 1 GHz or so than a 65 MHz processor if you are counting raw clock speed. But there's something to be said for a coprocessor which is not running a multitasking OS with interrupts from Ethernet, etc. Otherwise the BeagleBone Black with it's two PRUs would have not carved a nitch and the Pi would have swallowed the entire market.

Is the Pi really powerful if an Arduino does a better job of driving a NeoPixel chain with it's tight timing requirements? There's a certain power to be had in doing things with a single mindedness and the ARM in the RPPSOC does that nicely.

There are certain pieces of this market that this card meets. For example, we have sold hundreds of Pi cards which do nothing more than 3.3V to 5V conversion. We've set up this card do to that well. For uni-directional buses this card will hit that target. There's a whole world of legacy 5V cards out there which can't be accessed with most modern processors. The test market for legacy equipment alone is huge. The Pi can't talk to them without voltage translators and this card does that effortlessly.

Not a solution for everyone...

Doug at Land Boards, LLC

What other hat is there for the Raspberry Pi that has a PSOC?

I don't know what to say I feel that's like asking what possible other cars are there with F, e, r, r, a, r, and i in the name.

The dev-board has near-identical functionality, at $10. (In fact a higher-end PSoC) and fits the Pi with 5 jumper cables.