Bill, Maybe we can arrange a swap, as I have plans for a new series of articles using a new base. I've already slated one of the upcoming Beginnerbot articles around using the PropBOE, but this new base uses DC motors (not Tamiya) so it needs a motor driver. For this round of articles I'd lean toward boards with the driver built-in.
Bill, Maybe we can arrange a swap, as I have plans for a new series of articles using a new base. I've already slated one of the upcoming Beginnerbot articles around using the PropBOE, but this new base uses DC motors (not Tamiya) so it needs a motor driver. For this round of articles I'd lean toward boards with the driver built-in.
I own a boe-bot, S2 and a Stingray. The RoboProp looks good but I think Bill is pricing himself out of the market. If you look at the Arduino world, I can get an Arduino Mega on ebay for $35 and a L298N motor shield for $15 for a total of $50. The mega has many more I/O (digital and analog) than the Propeller. Yes, the propeller excels in generating VGA, etc. but how many of us carry a VGA monitor, keyboard and mouse on his mobile robot? Paying an extra $100 for the privilege of using a Propeller does not make sense to me. I think the fair price for a assembled RoboProp with L298N is around $75. BTW, the L298N is not powerful enough to support the Stingray motors. May be that is why the RoboProp is advertised for the Bot-bot? Bot-bot uses continuous rotating servos! I am planning to either dump the propeller board that comes with the Stingray or just use it as a slave processor to a Arduino Mega in a way similar to that of Andre's AVR Chameleon.
I own a boe-bot, S2 and a Stingray. The RoboProp looks good but I think Bill is pricing himself out of the market.
I can see why you may think so, however there are factors in play that I think you are not taking into account. I am on a break from some consulting, so I will elaborate.
RoboProp is not comparable to Arduino boards; they address different market segments.
I can make robot controllers closer in price to the Arduino - but not Propeller based, with a nice external 12 bit ADC, motor controller, and uSD support. Different ball of wax.
If you look at the Arduino world, I can get an Arduino Mega on ebay for $35 and a L298N motor shield for $15 for a total of $50. The mega has many more I/O (digital and analog) than the Propeller.
1) Propeller chip costs me, including shipping to my location, almost $8. The MCP3208 ADC is about $3.50 shipped. The L298 is about $2.50. Much more than an AtMega bought in 10k lots.
2) As a new startup, I cannot buy in 10k+ quantities - and thus my parts costs, pcb costs etc are much higher
3) I have to amortize development costs over a much smaller number of units
4) Ebay Arduino's have zero development costs, as it is open source - small companies just run the free gerbers by the thousands.
5) The AtMega is an 8 bit cpu, running at 16Mhz. The Propeller has eight 32 bit cores, each running at 100Mhz - not to mention much more ram.
Yes, the propeller excels in generating VGA, etc. but how many of us carry a VGA monitor, keyboard and mouse on his mobile robot? Paying an extra $100 for the privilege of using a Propeller does not make sense to me.
Ask anyone who is manufacturing products in small/medium quantities.
MSRP has to be 3x-5x raw parts cost to make a little profit.
I have to pay to get them assembled here in Canada - thus the price difference between the kit and assembled and tested; and I have to mark up my labor costs for getting them assembled so there is margin for distribution.
6) ordering takes time, shipping costs, inventory costs etc... duties... brokerage fees... all adds up
I think the fair price for a assembled RoboProp with L298N is around $75.
You are welcome to your opinion.
At $75, I would only get about about $37-$60 per unit from dealers / distributors.
An assembled and tested board (not including development costs) costs me significantly more than $37.
FYI, Parallax sells the Stingray controller for $99 assembled and tested, with fewer features (but better motor drivers) than RoboProp. And they make many more of them - and use their own CPU.
7) Dealers want 25%-30% discount of MSRP
8) Distributors want 30%-55% discount of MSRP
9) Marketing costs
10) documentation costs
I've considered selling a "Bare Essentials" kit (PCB, uSD socket, 6.25Mhz crystal) however I don't think there is a market for it, as I'd have to set the MSRP at around $35. I can already see the response "It should sell for $15!"
It is not possible to market a commercial product at a small percentage above raw parts costs without losing money.
BTW, the L298N is not powerful enough to support the Stingray motors.
I agree, and in that case the buyer can opt for the cheaper "Servo Edition" that leaves off the L298 and supporting parts, and add an off-board motor driver.
Bot-bot uses continuous rotating servos! I am planning to either dump the propeller board that comes with the Stingray or just use it as a slave processor to a Arduino Mega in a way similar to that of Andre's AVR Chameleon.
In your case, I'd use the Arduino as a slave (I/O expander) - makes much more sense as the Prop is far more powerful.
NutNut,
I do not mean to sound condescending, but your post indicates that you are unaware of:
- cost to design and bring a product to market
- all associated business costs
- cost of parts
- cost of assembly
- cost of marketing
- cost of distribution etc
and
the VAST difference in reproducing open source boards that use much cheaper parts in huge quantities, smt assembled overseas in huge lots versus low quantity (at least during the startup phase) assembly etc costs.
You don't have to believe me. Design some products of your own, start a new business, and see how much it costs you - and how much you have to charge.
Mikronauts is a real startup - not a hobby evening gig selling only direct.
The gadget gangster propeller platform kit is priced pretty close to an Arduino mega. The octodriver shield has some significant current capacity as well. So there is pairty between prop boards and arduino boards. But you would also need another shield for servo headers in either case.
I don't own one, but Bill's roboprop seems to fills a specific niche because it's form factor fits a BOE bot chassis and the Parallax PropBOE isn't shipping yet.
The gadget gangster propeller platform kit is priced pretty close to an Arduino mega. The octodriver shield has some significant current capacity as well. So there is pairty between prop boards and arduino boards. But you would also need another shield for servo headers in either case.
Agreed; and the feature set is closer between the Propeller Platform kit and an Arduino - except the PPK lacks an ADC.
The OctoDriver looks like a pretty nice motor driver shield! It can drive about twice the current that the L298 can.
So PPK is $34.95 (on sale now from $39.95), OctoDriver is $39.99... $79.94 for the pair.
Here is a comparison for NutNut:
RoboProp PropellerPlatformKit+OctoDriver
Kit Price $99.95 $79.94 ($84.94 after sale is over)
Clock Speed: 100Mhz 80Mhz
EEPROM: 64KB 32KB
uSD included n/a
ADC 8 channel 12 bit n/a
Motor Driver L298, 2A/chan stacked SN754410, 4A/chan
Servo headers 24 n/a
Analog headers 8 n/a
Those are the key differences.
RoboProp is on a single PCB, comes with tons of headers, uSD, more EEPROM and a nice 12 bit ADC.
PropellerPlatformKit+OctoDriver can drive significantly beefier motors, but is missing headers, ADC, uSD and has less EEPROM space.
Both are kits.
Which one is better for you? Whichever fits your application best of course!
I don't own one, but Bill's roboprop seems to fills a specific niche because it's form factor fits a BOE bot chassis and the Parallax PropBOE isn't shipping yet.
Actually RoboProp also fits a number of other chassis, I just have not announced which (other than the Magician that I carry) as I have not posted photos, videos etc yet.
I like what what I see about the BoeBOT, and look forward to playing with it.
I agree that small vendors manufacturing small batches will cost more than mass-produced products. One thing people say all the time is how much more powerful the propeller is than the mega 2560 or other controllers like PIC. The propeller has to have more powerful cpus because it does not have specialised hardware. Everything is done using software making it very flexible and at the same time negating much of its power. AVRs and PICs have specialised hardware UART (Mega has 4), PWM (14), ADC (14), SPI and TWI which offload the load from the cpu. Propeller uses its cpus in bit-banging and polling to emulate hardware.
Another thing that adds to the cost of the RoboProp is the lack of ADC (analog input) for the propeller. Owing to the lack of hardware multiply instruction, math is not that fast either. However, it still excels in generating VGA video. Also, OBEX is a great idea.
Regarding product planning and design, I think the proper way to do it is to research the market, set the product price point and then design your product taking into account the manufacturing cost, distribution cost and margin and not the other way round ie, design and manufacture a fantastic product, then add the various costs and margin to decide on the price. In the latter case, you may hit a sweet price point and you may not.
I do agree that it is horses for courses when you select the controller, It all depends on the project and requirements. There is no single solution that is optimal for all situations.
I did quite a bit of market research before deciding on the final feature set for RoboProp.
My original idea would have cost significantly more; and a version of it may surface in the future as a high end controller. In the future I will have lower priced robot controller boards - however they cannot have as rich a feature set as RoboProp due to costs.
My target was $99 for a full kit; which I met, by accepting that I would not have a great margin on products I sell through distribution.
By leaving off the L298 and related circuitry, I can sell a kit for $89.
I offer the more expensive assembled and tested versions as a convenience to buyers.
Should someone place a pre-paid order for 10,000 RoboProp's I can get significantly better pricing on components and assembly for a surface mount RoboProp at that volume, allowing me to sell it at a lower cost.
At least we agree that there is no single optimal solution for all situations :-)
I agree that small vendors manufacturing small batches will cost more than mass-produced products. One thing people say all the time is how much more powerful the propeller is than the mega 2560 or other controllers like PIC. The propeller has to have more powerful cpus because it does not have specialised hardware. Everything is done using software making it very flexible and at the same time negating much of its power. AVRs and PICs have specialised hardware UART (Mega has 4), PWM (14), ADC (14), SPI and TWI which offload the load from the cpu. Propeller uses its cpus in bit-banging and polling to emulate hardware.
Another thing that adds to the cost of the RoboProp is the lack of ADC (analog input) for the propeller. Owing to the lack of hardware multiply instruction, math is not that fast either. However, it still excels in generating VGA video. Also, OBEX is a great idea.
Regarding product planning and design, I think the proper way to do it is to research the market, set the product price point and then design your product taking into account the manufacturing cost, distribution cost and margin and not the other way round ie, design and manufacture a fantastic product, then add the various costs and margin to decide on the price. In the latter case, you may hit a sweet price point and you may not.
I do agree that it is horses for courses when you select the controller, It all depends on the project and requirements. There is no single solution that is optimal for all situations.
You've stacked the price deck against most any commercial product by selecting an eBay seller who is offering the product under distributor cost. The Mega2560 retails for about $60 to $65 through resellers willing to provide end-user support (a requirement for being an authorized Arduino reseller). Anything substantially less, without a quantity buy, *could* and likely does indicate gray market, reject product, or outright counterfeiting.
There's now a good number of counterfeit Arduino-branded boards coming from China. Though the board design is open source -- there are a lot of off-branded Mega2560, too -- the copy-catters are trading on the Arduino brand, and undercutting the spirit of open source hardware and software.
None of the above has anything to do with Bill's costs, his markups, the feature set of his board, variations in the cost of MCUs (an ATmega2560 is more expensive than a Propeller), or anything else, but simply comparing one valid pricing structure over another.
As for using separate cores to perform the jobs of built-in hardware, look at it this way: what if your application doesn't use serial or hardware interrupts? You've got hardware wasted, because it's dedicated to something you're not using. You still have to fit everything in that single thread running in the one core of your MCU. (And don't forget, at the end of the day UARTs bit-bang, too. They just do it in hardware.)
If, on the other hand, your use of the controller is more along traditional lines, and your application makes good use of the hardware built into the chip, and a single ,main thread is acceptable, then by all means the selection of appropriate MCUs is fairly wide open.
I had a lot of fun at Vancouver mini Maker Faire on the previous weekend. I had a 10x20 booth with two tables, showing off my robots, robot controllers and modular system.
Got tons of great feedback, made some great contacts ... and I was happy that Michael Park came by! He drove up from Seattle to come to the show.
Another forumista came by, but sorry, I forgot your name :-(
People loved the LIDAR demo from RoboProp (see youtube.com/mikronauts) and also liked my lower cost non-propeller based controller. Kids loved the panning IR head a bit too much, kept grabbing it, and three servo's later, I had to shut the demo down.
I am thinking of doing the main Maker Faire in San Mateo next year; hopefully it will not conflict with the Prop Expo.
Some other updates (after a LONG absence)
- I added sample "roamer" code and RoboProp.spin object as an attachment to post #1, it is from my Servo articles
- On my site I added a robot controller at roughly half the price of RoboProp (not propeller based, which is why I did not start a thread for it here, email for info)
For forum users who already have a lot of parts, I can provide a "RoboProp Essentials" (pcb, 6.250MHz crystal, uSD socket I use) for $15+s/h
I also have a new"RoboDeck" prototyping board that plugs into the two 10 pin connectors on RoboProp, and provides a 0.1" hole grid prototyping PCB. I use it with a small solder less breadboard on top :-)
(I'll post a RoboDeck photo in a day or so... after I take one)
Comments
-- Gordon
Let's take this discussion to email... but let's just say I have an assembled and tested RoboProp which now has your name on it
Bill
I can see why you may think so, however there are factors in play that I think you are not taking into account. I am on a break from some consulting, so I will elaborate.
RoboProp is not comparable to Arduino boards; they address different market segments.
I can make robot controllers closer in price to the Arduino - but not Propeller based, with a nice external 12 bit ADC, motor controller, and uSD support. Different ball of wax.
1) Propeller chip costs me, including shipping to my location, almost $8. The MCP3208 ADC is about $3.50 shipped. The L298 is about $2.50. Much more than an AtMega bought in 10k lots.
2) As a new startup, I cannot buy in 10k+ quantities - and thus my parts costs, pcb costs etc are much higher
3) I have to amortize development costs over a much smaller number of units
4) Ebay Arduino's have zero development costs, as it is open source - small companies just run the free gerbers by the thousands.
5) The AtMega is an 8 bit cpu, running at 16Mhz. The Propeller has eight 32 bit cores, each running at 100Mhz - not to mention much more ram.
Ask anyone who is manufacturing products in small/medium quantities.
MSRP has to be 3x-5x raw parts cost to make a little profit.
I have to pay to get them assembled here in Canada - thus the price difference between the kit and assembled and tested; and I have to mark up my labor costs for getting them assembled so there is margin for distribution.
6) ordering takes time, shipping costs, inventory costs etc... duties... brokerage fees... all adds up
You are welcome to your opinion.
At $75, I would only get about about $37-$60 per unit from dealers / distributors.
An assembled and tested board (not including development costs) costs me significantly more than $37.
FYI, Parallax sells the Stingray controller for $99 assembled and tested, with fewer features (but better motor drivers) than RoboProp. And they make many more of them - and use their own CPU.
7) Dealers want 25%-30% discount of MSRP
8) Distributors want 30%-55% discount of MSRP
9) Marketing costs
10) documentation costs
I've considered selling a "Bare Essentials" kit (PCB, uSD socket, 6.25Mhz crystal) however I don't think there is a market for it, as I'd have to set the MSRP at around $35. I can already see the response "It should sell for $15!"
It is not possible to market a commercial product at a small percentage above raw parts costs without losing money.
I agree, and in that case the buyer can opt for the cheaper "Servo Edition" that leaves off the L298 and supporting parts, and add an off-board motor driver.
Actually the Boe-Bot is just the first usage case; more are forthcoming.
RoboProp is far more advanced and capable than mass-produced Arduinos. If the Arduino meets your needs, by all means use it.
I have several Arduino's in my lab as well - they are very neat little boards; but they don't compare. Different board for a different market.
In your case, I'd use the Arduino as a slave (I/O expander) - makes much more sense as the Prop is far more powerful.
NutNut,
I do not mean to sound condescending, but your post indicates that you are unaware of:
- cost to design and bring a product to market
- all associated business costs
- cost of parts
- cost of assembly
- cost of marketing
- cost of distribution etc
and
the VAST difference in reproducing open source boards that use much cheaper parts in huge quantities, smt assembled overseas in huge lots versus low quantity (at least during the startup phase) assembly etc costs.
You don't have to believe me. Design some products of your own, start a new business, and see how much it costs you - and how much you have to charge.
Mikronauts is a real startup - not a hobby evening gig selling only direct.
Best Wishes,
Bill
I don't own one, but Bill's roboprop seems to fills a specific niche because it's form factor fits a BOE bot chassis and the Parallax PropBOE isn't shipping yet.
Agreed; and the feature set is closer between the Propeller Platform kit and an Arduino - except the PPK lacks an ADC.
The OctoDriver looks like a pretty nice motor driver shield! It can drive about twice the current that the L298 can.
So PPK is $34.95 (on sale now from $39.95), OctoDriver is $39.99... $79.94 for the pair.
Here is a comparison for NutNut:
Those are the key differences.
RoboProp is on a single PCB, comes with tons of headers, uSD, more EEPROM and a nice 12 bit ADC.
PropellerPlatformKit+OctoDriver can drive significantly beefier motors, but is missing headers, ADC, uSD and has less EEPROM space.
Both are kits.
Which one is better for you? Whichever fits your application best of course!
I class Nick as a friendly competitor :-)
Actually RoboProp also fits a number of other chassis, I just have not announced which (other than the Magician that I carry) as I have not posted photos, videos etc yet.
I like what what I see about the BoeBOT, and look forward to playing with it.
Another thing that adds to the cost of the RoboProp is the lack of ADC (analog input) for the propeller. Owing to the lack of hardware multiply instruction, math is not that fast either. However, it still excels in generating VGA video. Also, OBEX is a great idea.
Regarding product planning and design, I think the proper way to do it is to research the market, set the product price point and then design your product taking into account the manufacturing cost, distribution cost and margin and not the other way round ie, design and manufacture a fantastic product, then add the various costs and margin to decide on the price. In the latter case, you may hit a sweet price point and you may not.
I do agree that it is horses for courses when you select the controller, It all depends on the project and requirements. There is no single solution that is optimal for all situations.
My original idea would have cost significantly more; and a version of it may surface in the future as a high end controller. In the future I will have lower priced robot controller boards - however they cannot have as rich a feature set as RoboProp due to costs.
My target was $99 for a full kit; which I met, by accepting that I would not have a great margin on products I sell through distribution.
By leaving off the L298 and related circuitry, I can sell a kit for $89.
I offer the more expensive assembled and tested versions as a convenience to buyers.
Should someone place a pre-paid order for 10,000 RoboProp's I can get significantly better pricing on components and assembly for a surface mount RoboProp at that volume, allowing me to sell it at a lower cost.
At least we agree that there is no single optimal solution for all situations :-)
There's now a good number of counterfeit Arduino-branded boards coming from China. Though the board design is open source -- there are a lot of off-branded Mega2560, too -- the copy-catters are trading on the Arduino brand, and undercutting the spirit of open source hardware and software.
None of the above has anything to do with Bill's costs, his markups, the feature set of his board, variations in the cost of MCUs (an ATmega2560 is more expensive than a Propeller), or anything else, but simply comparing one valid pricing structure over another.
As for using separate cores to perform the jobs of built-in hardware, look at it this way: what if your application doesn't use serial or hardware interrupts? You've got hardware wasted, because it's dedicated to something you're not using. You still have to fit everything in that single thread running in the one core of your MCU. (And don't forget, at the end of the day UARTs bit-bang, too. They just do it in hardware.)
If, on the other hand, your use of the controller is more along traditional lines, and your application makes good use of the hardware built into the chip, and a single ,main thread is acceptable, then by all means the selection of appropriate MCUs is fairly wide open.
-- Gordon
I had a lot of fun at Vancouver mini Maker Faire on the previous weekend. I had a 10x20 booth with two tables, showing off my robots, robot controllers and modular system.
Got tons of great feedback, made some great contacts ... and I was happy that Michael Park came by! He drove up from Seattle to come to the show.
Another forumista came by, but sorry, I forgot your name :-(
People loved the LIDAR demo from RoboProp (see youtube.com/mikronauts) and also liked my lower cost non-propeller based controller. Kids loved the panning IR head a bit too much, kept grabbing it, and three servo's later, I had to shut the demo down.
I am thinking of doing the main Maker Faire in San Mateo next year; hopefully it will not conflict with the Prop Expo.
Some other updates (after a LONG absence)
- I added sample "roamer" code and RoboProp.spin object as an attachment to post #1, it is from my Servo articles
- On my site I added a robot controller at roughly half the price of RoboProp (not propeller based, which is why I did not start a thread for it here, email for info)
For forum users who already have a lot of parts, I can provide a "RoboProp Essentials" (pcb, 6.250MHz crystal, uSD socket I use) for $15+s/h
I also have a new"RoboDeck" prototyping board that plugs into the two 10 pin connectors on RoboProp, and provides a 0.1" hole grid prototyping PCB. I use it with a small solder less breadboard on top :-)
(I'll post a RoboDeck photo in a day or so... after I take one)
Bill