Our answer is Raspberry Pi
Most people involved in or interested in the development of electronic devices running ARM processors are well aware of the Raspberry Pi project. In fact, this is the first cheap development board with an open architecture, good technical support, as well as a large and friendly community of enthusiasts.
Our small company (start-up) almost accidentally decided to compete with this wonderful berry. Well, I decided to write about it after I discovered an interest in this topic right here .
In general, single-board computers existed before the “raspberry”, but were considered “professional tools” intended primarily for industrial use. In addition to the cost typical for industrial solutions, such boards were traditionally quite closed, technical information about them was usually strictly metered (up to the signing of NDA with the manufacturer as a necessary condition for the purchase), and developing their own (custom-made) solution on their basis The task is non-trivial, especially for the amateur single.
Raspberry Pi is the first real project, designed for enthusiasts, amateurs, students and other "single handicraftsmen with a motor." What is called “cheap and cheerful” - for $ 25 or $ 35 (not including shipping costs) you get a single-board computer with an ARM11 processor and Linux on board, with a bunch of different interfaces. On the knee you can collect a lot of funny devices, and the "as is" board can be used for many different applications.
The cheapness of the board is mainly due to the orientation of the project to education, i.e. initially a high rate of return was not planned. By my estimation, manufacturers earn hardly more than $ 5 on one pay.
')
It would seem that everything is just fine, but ... You can always make it even better, right?
Our company is a typical start-up, i.e. created specifically for the implementation of one rather peculiar project (I will not talk about it here, because for a number of reasons our project is an almost ideal target for criticism of IT specialists, that is, the main audience of Habrahabr). In the process of working on our main project, we had (yes, we were forced, damn! In fact, we are programmers, but here ...) to develop our own board on the ARM processor. And the board turns out, in our opinion, very good. Those. for our project, it is generally perfect, I think, but after all, after minor modifications (mainly geometry changes, since we make our board for a very specific case), we can use it for other purposes too! That was born a "side" project - our answer Raspberry Pi.
We make a compact, modular, productive (ahem ... well, relatively - details below), a universal and inexpensive basic platform for the development of electronic devices for various purposes.
The current dimensions of the Paspberry Pi are 8.6x5.4x1.5 (1.7) cm. And with the first two, everything is almost OK (this is the size of a credit card, and “almost” because some connectors still crawl out of this size) height (due to all the same connectors) is too big. We are going to make the main processor module (about modularity - below) with a thickness somewhere around 5-6 mm. Most likely, we will do the length and width as in Pi, but there will definitely be nothing hanging around. On the other hand, we simply don’t need so much space (see “modularity”) ... So, perhaps, the width with height will be smaller.
The most interesting, in our opinion, feature of our project. Why pay for something you don't need? And where to get what you need, if it is not on the board?
In our “basic” project on the board there will be just a dof of everything interesting - a bunch of controllers, interfaces, etc. It's just that we need it for a number of reasons. As a result, we will have support, for example, LCD, touchscreen, Ethernet, WiFi, GPS, 3G and others. But all this can in fact not be placed on the processor board, but connected to it as needed using cables or connectors (the main thing is that there will already be support for all these devices, including the driver). Therefore, on the main - processor - module, we will place only what can not be connected to each other without soldering, namely: the processor and memory. And the buyer himself can connect the rest as he wants (and we will help with information, modules and components - with the same connectors).
An additional advantage of this approach: when a new processor module is released (progress does not stand still), most of the plug-ins continue to work with it without changes (we will simply keep the specification of the connection interface from version to version).
Raspberry Pi is based on the Broadcom BCM2835 processor, this is ARM11 (ARMv6 instruction set, there is no NEON support), the frequency of the processor is 700 MHz. In modern times - not the quickest "heart".
We are developing a board based on the Samsung S5PV210 processor (Exynos 3). This is a rather “old” processor. Used, for example, in the Samsung Galaxy Tab 7.0 Plus. 45nm, Cortex A8, support for ARMv7 + NEON, 1GHz, graphics PowerVR SGX540, hardware support for JPEG encoding / decoding and 720p video.
Why this processor? It is very affordable, at a price close to the Chinese ARM Cortex A8, and in terms of performance (especially in three-dimensional graphics), it covers them like a bull sheep. At the same time, unlike the same "Chinese", technical information on the S5PV210 is more or less available. Well, a number of other buns (large L2 cache, 45nm vs 65nm process for most “Chinese”, etc.) ...
Memory - DDR2, dual-channel mode, 512MB or 1GB - have not yet decided: either we will make two versions of the processor module, or there will simply be 1GB always (the price difference is small).
In the future, it will be possible to offer several variants of processor modules on various SoCs. Just, for example, such a module on the quad-core Exynos 4412 will now cost about 2.5 times more expensive ... And for most tasks, the S5PV210 is more than enough.
We initially want to make out of the box support for our motherboard the three most popular embedded OS: Linux, Android 4 and WindowsCE. This will make it possible to use the most convenient option for specific purposes. Freedom of choice taxis!
It is clear that a license for Windows CE will have to pay a few dollars (purely voluntary - an option), but do not say “fi, byaka!” - it has a number of advantages over its competitors. Exynos 3 - alas, Samsung's latest ARM processor, officially supporting Windows CE ...
Versatility in terms of connecting external devices is provided by modularity. Well, on the processor board itself, all the main interfaces will be accessible to a person familiar with the soldering iron.
The processor itself costs about $ 9 in China for a batch of 1,000 pieces. Those. the cost of the main module - the processor board - will be about 20-25 USD. Plus VAT 18%. Plus our interest. In general, our target retail price is 1000r, excluding delivery costs. It is unlikely to be cheaper to do, but there is no sense in setting a higher price either.
Our main goal is to make the platform common enough to be turned to us for the manufacture of bulk batches of customized solutions. Well, actually, we are always ready to help in developing such solutions (for a reasonable reward). And since we recently became residents of one special economic zone, plus we have experience in delivery, customs clearance and electronics assembly as part of our main project, there is hope that our service will be in demand.
Since the project is in its initial stage, it can still be painlessly changed. You have the opportunity to participate in the process, to offer something or to criticize our choice. We are pleased to hear from you.
Any ideas on sizing? In our opinion, the “credit card” form factor is slightly out of date - so much space is not required for modern electronics, everything can be made more compact.
The range of plug-ins. We are definitely planning to do the following:
1. Audio / video module. HDMI, component video, 3.5mm headphone and microphone connectors, space for a small speaker and microphone (you can solder yourself if you want). Planned price - 200r.
2. Flash memory modules. On the processor module, most likely, there will be no flash memory (there will be a slot for a microSD card). But a separate module, for example, 4GB eMMC can be added (approximately - 400r).
3. Ethernet module (~ 250r).
4. WiFi + Bluetooth module (~ 500r).
5. 3G module (until the cost is unknown).
6. Module with Li-Ion rechargeable battery. In fact, the “standard” power is planned from the microUSB connector, but in some applications the battery will definitely not be superfluous (~ 400r).
7. Module for connecting an LCD monitor (LVDS interface + connector for our 7 "LCD module).
8. GPS module (with the price, too, is unclear).
9. USB hub (4 USB ports). We think that no more 200r.
We are waiting for critics and suggestions for a set of modules from you.
Stages of the project while seen in this way:
1. Somewhere in April there will be a prototype.
2. By the summer we will finish off a software and we will make an industrial sample.
3. By the autumn we will make the first batch (1000 pieces).
This is an optimistic option. Unfortunately, we have a small budget, because the company is small and exists solely for the money earned (we have no investors). But we hope that everything will work out.
Our small company (start-up) almost accidentally decided to compete with this wonderful berry. Well, I decided to write about it after I discovered an interest in this topic right here .
In general, single-board computers existed before the “raspberry”, but were considered “professional tools” intended primarily for industrial use. In addition to the cost typical for industrial solutions, such boards were traditionally quite closed, technical information about them was usually strictly metered (up to the signing of NDA with the manufacturer as a necessary condition for the purchase), and developing their own (custom-made) solution on their basis The task is non-trivial, especially for the amateur single.
Raspberry Pi is the first real project, designed for enthusiasts, amateurs, students and other "single handicraftsmen with a motor." What is called “cheap and cheerful” - for $ 25 or $ 35 (not including shipping costs) you get a single-board computer with an ARM11 processor and Linux on board, with a bunch of different interfaces. On the knee you can collect a lot of funny devices, and the "as is" board can be used for many different applications.
The cheapness of the board is mainly due to the orientation of the project to education, i.e. initially a high rate of return was not planned. By my estimation, manufacturers earn hardly more than $ 5 on one pay.
')
It would seem that everything is just fine, but ... You can always make it even better, right?
Our company is a typical start-up, i.e. created specifically for the implementation of one rather peculiar project (I will not talk about it here, because for a number of reasons our project is an almost ideal target for criticism of IT specialists, that is, the main audience of Habrahabr). In the process of working on our main project, we had (yes, we were forced, damn! In fact, we are programmers, but here ...) to develop our own board on the ARM processor. And the board turns out, in our opinion, very good. Those. for our project, it is generally perfect, I think, but after all, after minor modifications (mainly geometry changes, since we make our board for a very specific case), we can use it for other purposes too! That was born a "side" project - our answer Raspberry Pi.
main idea
We make a compact, modular, productive (ahem ... well, relatively - details below), a universal and inexpensive basic platform for the development of electronic devices for various purposes.
Compactness
The current dimensions of the Paspberry Pi are 8.6x5.4x1.5 (1.7) cm. And with the first two, everything is almost OK (this is the size of a credit card, and “almost” because some connectors still crawl out of this size) height (due to all the same connectors) is too big. We are going to make the main processor module (about modularity - below) with a thickness somewhere around 5-6 mm. Most likely, we will do the length and width as in Pi, but there will definitely be nothing hanging around. On the other hand, we simply don’t need so much space (see “modularity”) ... So, perhaps, the width with height will be smaller.
Modularity
The most interesting, in our opinion, feature of our project. Why pay for something you don't need? And where to get what you need, if it is not on the board?
In our “basic” project on the board there will be just a dof of everything interesting - a bunch of controllers, interfaces, etc. It's just that we need it for a number of reasons. As a result, we will have support, for example, LCD, touchscreen, Ethernet, WiFi, GPS, 3G and others. But all this can in fact not be placed on the processor board, but connected to it as needed using cables or connectors (the main thing is that there will already be support for all these devices, including the driver). Therefore, on the main - processor - module, we will place only what can not be connected to each other without soldering, namely: the processor and memory. And the buyer himself can connect the rest as he wants (and we will help with information, modules and components - with the same connectors).
An additional advantage of this approach: when a new processor module is released (progress does not stand still), most of the plug-ins continue to work with it without changes (we will simply keep the specification of the connection interface from version to version).
Performance
Raspberry Pi is based on the Broadcom BCM2835 processor, this is ARM11 (ARMv6 instruction set, there is no NEON support), the frequency of the processor is 700 MHz. In modern times - not the quickest "heart".
We are developing a board based on the Samsung S5PV210 processor (Exynos 3). This is a rather “old” processor. Used, for example, in the Samsung Galaxy Tab 7.0 Plus. 45nm, Cortex A8, support for ARMv7 + NEON, 1GHz, graphics PowerVR SGX540, hardware support for JPEG encoding / decoding and 720p video.
Why this processor? It is very affordable, at a price close to the Chinese ARM Cortex A8, and in terms of performance (especially in three-dimensional graphics), it covers them like a bull sheep. At the same time, unlike the same "Chinese", technical information on the S5PV210 is more or less available. Well, a number of other buns (large L2 cache, 45nm vs 65nm process for most “Chinese”, etc.) ...
Memory - DDR2, dual-channel mode, 512MB or 1GB - have not yet decided: either we will make two versions of the processor module, or there will simply be 1GB always (the price difference is small).
In the future, it will be possible to offer several variants of processor modules on various SoCs. Just, for example, such a module on the quad-core Exynos 4412 will now cost about 2.5 times more expensive ... And for most tasks, the S5PV210 is more than enough.
Versatility
We initially want to make out of the box support for our motherboard the three most popular embedded OS: Linux, Android 4 and WindowsCE. This will make it possible to use the most convenient option for specific purposes. Freedom of choice taxis!
It is clear that a license for Windows CE will have to pay a few dollars (purely voluntary - an option), but do not say “fi, byaka!” - it has a number of advantages over its competitors. Exynos 3 - alas, Samsung's latest ARM processor, officially supporting Windows CE ...
Versatility in terms of connecting external devices is provided by modularity. Well, on the processor board itself, all the main interfaces will be accessible to a person familiar with the soldering iron.
Cost of
The processor itself costs about $ 9 in China for a batch of 1,000 pieces. Those. the cost of the main module - the processor board - will be about 20-25 USD. Plus VAT 18%. Plus our interest. In general, our target retail price is 1000r, excluding delivery costs. It is unlikely to be cheaper to do, but there is no sense in setting a higher price either.
Our main goal is to make the platform common enough to be turned to us for the manufacture of bulk batches of customized solutions. Well, actually, we are always ready to help in developing such solutions (for a reasonable reward). And since we recently became residents of one special economic zone, plus we have experience in delivery, customs clearance and electronics assembly as part of our main project, there is hope that our service will be in demand.
Any different
Since the project is in its initial stage, it can still be painlessly changed. You have the opportunity to participate in the process, to offer something or to criticize our choice. We are pleased to hear from you.
Any ideas on sizing? In our opinion, the “credit card” form factor is slightly out of date - so much space is not required for modern electronics, everything can be made more compact.
The range of plug-ins. We are definitely planning to do the following:
1. Audio / video module. HDMI, component video, 3.5mm headphone and microphone connectors, space for a small speaker and microphone (you can solder yourself if you want). Planned price - 200r.
2. Flash memory modules. On the processor module, most likely, there will be no flash memory (there will be a slot for a microSD card). But a separate module, for example, 4GB eMMC can be added (approximately - 400r).
3. Ethernet module (~ 250r).
4. WiFi + Bluetooth module (~ 500r).
5. 3G module (until the cost is unknown).
6. Module with Li-Ion rechargeable battery. In fact, the “standard” power is planned from the microUSB connector, but in some applications the battery will definitely not be superfluous (~ 400r).
7. Module for connecting an LCD monitor (LVDS interface + connector for our 7 "LCD module).
8. GPS module (with the price, too, is unclear).
9. USB hub (4 USB ports). We think that no more 200r.
We are waiting for critics and suggestions for a set of modules from you.
Stages of the project while seen in this way:
1. Somewhere in April there will be a prototype.
2. By the summer we will finish off a software and we will make an industrial sample.
3. By the autumn we will make the first batch (1000 pieces).
This is an optimistic option. Unfortunately, we have a small budget, because the company is small and exists solely for the money earned (we have no investors). But we hope that everything will work out.
Source: https://habr.com/ru/post/170527/
All Articles