I am about 90 percent done with a design project with a LPC2148 and am
looking to the future to integrate ethernet into the project. I was
told by a couple of design friends that the LPC series is not a good
series for ethernet, because of all the overhead circuitry that the
LPC doesn't handle. They are using the STR9 series for their ethernet
design, which I was told has better infrastructure for handling
ethernet. If the LPC series is not recommended for this, then I will
finish up this design and migrate over to another micro, when this
project is stable and in production. Any thoughts and experiences out
there?
Sutton
LPC2148 with ethernet
Started by ●November 17, 2006
Reply by ●November 17, 20062006-11-17
>> Any thoughts and experiences out there?
>> Sutton
Well, here are my thoughts right now. After looking over both these LPC and STR parts, I am not happy with either of them. Neither part gives me what I am really after. I started out planning to use a Cypress CY68013 USB2 controller to get hi-spd USB, then some other ARM9 MCUs, then went to the STR912 route, and now looked at LPC2378.
Some of the other posters here made the comment that I was choosing the wrong part for the job. I agree with them. I can't find the right part for the job I want to do in one chip. I don't think it exists.
So, I just finished an SoC on an Altera Cyclone FPGA. It was fantastic. I could create exactly what I needed with soft cores in terms of bus control, perfs, and CPU. $4 part that does everything perfectly as I desired. Everything integrated in a single part.
I started thinking about this same approach last night for this problem, and checking on the available IP cores. It looks very doable. I can put in both a USB2-480MHz and Enet-1000 which is what I really wanted from the start. Easy to get external PHYs for both. Then I can create my own full 32 bit DMA chs between my external 32 bit bus, external 16 bit bus, and the ENET and USB exactly how I want it. Virtually all hardware piping, little or no CPU involvement. Blazing fast. Not only that, I can put all of the pins exactly where I what them. Hey, what a deal.
So my thoughts at this point are to roll-my-own SoC. I think that is the real solution I need. After discovering how much you can do with these and how easy they are, I probably will do a lot more of these in the future.
Chris.
>> Sutton
Well, here are my thoughts right now. After looking over both these LPC and STR parts, I am not happy with either of them. Neither part gives me what I am really after. I started out planning to use a Cypress CY68013 USB2 controller to get hi-spd USB, then some other ARM9 MCUs, then went to the STR912 route, and now looked at LPC2378.
Some of the other posters here made the comment that I was choosing the wrong part for the job. I agree with them. I can't find the right part for the job I want to do in one chip. I don't think it exists.
So, I just finished an SoC on an Altera Cyclone FPGA. It was fantastic. I could create exactly what I needed with soft cores in terms of bus control, perfs, and CPU. $4 part that does everything perfectly as I desired. Everything integrated in a single part.
I started thinking about this same approach last night for this problem, and checking on the available IP cores. It looks very doable. I can put in both a USB2-480MHz and Enet-1000 which is what I really wanted from the start. Easy to get external PHYs for both. Then I can create my own full 32 bit DMA chs between my external 32 bit bus, external 16 bit bus, and the ENET and USB exactly how I want it. Virtually all hardware piping, little or no CPU involvement. Blazing fast. Not only that, I can put all of the pins exactly where I what them. Hey, what a deal.
So my thoughts at this point are to roll-my-own SoC. I think that is the real solution I need. After discovering how much you can do with these and how easy they are, I probably will do a lot more of these in the future.
Chris.
Reply by ●November 17, 20062006-11-17
> So my thoughts at this point are to roll-my-own SoC.
I think that is
the real solution I need. After discovering how much you can do with
these and how easy they are, I probably will do a lot more of these in
the future.
good point, but you don't mention what is the cost of the high speed
USB and Giga Ethernet IPs? I guess 4 digit numbers? and Verilog/VHDL
guys usually charge more per design than the general micro embedded
guys, so if you don't do this yourself but have to subcontract the
design the NRE cost will go too high for most of general applications
IMO, so there is still too much room for the old fashioned MCUs, but I
may be wrong.
Best regards
Tsvetan
---
PCB prototypes for $26 at http://run.to/pcb (http://www.olimex.com/pcb)
Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
(http://www.olimex.com/dev)
the real solution I need. After discovering how much you can do with
these and how easy they are, I probably will do a lot more of these in
the future.
good point, but you don't mention what is the cost of the high speed
USB and Giga Ethernet IPs? I guess 4 digit numbers? and Verilog/VHDL
guys usually charge more per design than the general micro embedded
guys, so if you don't do this yourself but have to subcontract the
design the NRE cost will go too high for most of general applications
IMO, so there is still too much room for the old fashioned MCUs, but I
may be wrong.
Best regards
Tsvetan
---
PCB prototypes for $26 at http://run.to/pcb (http://www.olimex.com/pcb)
Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
(http://www.olimex.com/dev)
Reply by ●November 17, 20062006-11-17
> So my thoughts at this point are to roll-my-own SoC.
I think that is
the real solution I need. After discovering how much you can do with
these and how easy they are, I probably will do a lot more of these in
the future.
>
> Chris.
I like FPGAs and building special purpose hardware is what they are
all about. In some cases, the microcontroller has a lot of gadgets
that just aren't necessary (PWM, I2C, SPI, etc). There is certainly a
niche in the spectrum of digital hardware for everything. If there
wasn't a need, they wouldn't exist.
But the thing is, the LPCs are microcontrollers. Designed
specifically to control things in applications such as toys,
applicances and automotive. They are not high bandwidth
communicatiions channels.
OTOH, it would take a lot of effort to duplicate all of the features
of an LPC in an FPGA. In fact, I am not aware of ANY FPGAs with 128k
of user flash and 64k RAM like the lowly LPC2106 much less in a tiny
48 pin package.
Richard
the real solution I need. After discovering how much you can do with
these and how easy they are, I probably will do a lot more of these in
the future.
>
> Chris.
I like FPGAs and building special purpose hardware is what they are
all about. In some cases, the microcontroller has a lot of gadgets
that just aren't necessary (PWM, I2C, SPI, etc). There is certainly a
niche in the spectrum of digital hardware for everything. If there
wasn't a need, they wouldn't exist.
But the thing is, the LPCs are microcontrollers. Designed
specifically to control things in applications such as toys,
applicances and automotive. They are not high bandwidth
communicatiions channels.
OTOH, it would take a lot of effort to duplicate all of the features
of an LPC in an FPGA. In fact, I am not aware of ANY FPGAs with 128k
of user flash and 64k RAM like the lowly LPC2106 much less in a tiny
48 pin package.
Richard
Reply by ●November 17, 20062006-11-17
My problem with FPGAs like these is that there is little or
not analog on them, like A/Ds. Isn't that right ?
It would be nice if I was wrong. Hope I am and maybe I
just haven't looked hard enough.
boB
--- In l..., "tsvetanusunov" wrote:
>
> > So my thoughts at this point are to roll-my-own SoC. I think that
is
> the real solution I need. After discovering how much you can do
with
> these and how easy they are, I probably will do a lot more of these
in
> the future.
>
> good point, but you don't mention what is the cost of the high
speed
> USB and Giga Ethernet IPs? I guess 4 digit numbers? and
Verilog/VHDL
> guys usually charge more per design than the general micro embedded
> guys, so if you don't do this yourself but have to subcontract the
> design the NRE cost will go too high for most of general
applications
> IMO, so there is still too much room for the old fashioned MCUs,
but I
> may be wrong.
>
> Best regards
> Tsvetan
> ---
> PCB prototypes for $26 at http://run.to/pcb
(http://www.olimex.com/pcb)
> Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
> (http://www.olimex.com/dev)
>
not analog on them, like A/Ds. Isn't that right ?
It would be nice if I was wrong. Hope I am and maybe I
just haven't looked hard enough.
boB
--- In l..., "tsvetanusunov" wrote:
>
> > So my thoughts at this point are to roll-my-own SoC. I think that
is
> the real solution I need. After discovering how much you can do
with
> these and how easy they are, I probably will do a lot more of these
in
> the future.
>
> good point, but you don't mention what is the cost of the high
speed
> USB and Giga Ethernet IPs? I guess 4 digit numbers? and
Verilog/VHDL
> guys usually charge more per design than the general micro embedded
> guys, so if you don't do this yourself but have to subcontract the
> design the NRE cost will go too high for most of general
applications
> IMO, so there is still too much room for the old fashioned MCUs,
but I
> may be wrong.
>
> Best regards
> Tsvetan
> ---
> PCB prototypes for $26 at http://run.to/pcb
(http://www.olimex.com/pcb)
> Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
> (http://www.olimex.com/dev)
>
Reply by ●November 17, 20062006-11-17
--- In l..., "bobtransformer" wrote:
>
> My problem with FPGAs like these is that there is little or
> not analog on them, like A/Ds. Isn't that right ?
> It would be nice if I was wrong. Hope I am and maybe I
> just haven't looked hard enough.
> boB
I haven't seen any with A/D although there may be some. However, it
is easy to implement SPI and hang a whole bunch of very fast A/Ds on
the bus. That's probably a better solution anyway because you can get
just the A/D performance you want (speed, resolution, voltage range,
etc.).
Richard
>
> My problem with FPGAs like these is that there is little or
> not analog on them, like A/Ds. Isn't that right ?
> It would be nice if I was wrong. Hope I am and maybe I
> just haven't looked hard enough.
> boB
I haven't seen any with A/D although there may be some. However, it
is easy to implement SPI and hang a whole bunch of very fast A/Ds on
the bus. That's probably a better solution anyway because you can get
just the A/D performance you want (speed, resolution, voltage range,
etc.).
Richard
Reply by ●November 17, 20062006-11-17
On Friday 17 November 2006 18:20, s...@hotmail.com wrote:
> I started thinking about this same approach last night for this problem,
> and checking on the available IP cores. It looks very doable. I can put
> in both a USB2-480MHz and Enet-1000 which is what I really wanted from the
> start. Easy to get external PHYs for both. Then I can create my own full
> 32 bit DMA chs between my external 32 bit bus, external 16 bit bus, and the
> ENET and USB exactly how I want it. Virtually all hardware piping, little
> or no CPU involvement. Blazing fast. Not only that, I can put all of the
> pins exactly where I what them. Hey, what a deal.
>
Just a short remark: The free IP cores available aren't necessarily production
quality. The USB 2.0 core available from opencores.org for example is known
to have several show-stopping bugs, I suppose something as complex as a gbit
ethernet controller isn't going to be fully functional either.
Best regards,
Dominic
> I started thinking about this same approach last night for this problem,
> and checking on the available IP cores. It looks very doable. I can put
> in both a USB2-480MHz and Enet-1000 which is what I really wanted from the
> start. Easy to get external PHYs for both. Then I can create my own full
> 32 bit DMA chs between my external 32 bit bus, external 16 bit bus, and the
> ENET and USB exactly how I want it. Virtually all hardware piping, little
> or no CPU involvement. Blazing fast. Not only that, I can put all of the
> pins exactly where I what them. Hey, what a deal.
>
Just a short remark: The free IP cores available aren't necessarily production
quality. The USB 2.0 core available from opencores.org for example is known
to have several show-stopping bugs, I suppose something as complex as a gbit
ethernet controller isn't going to be fully functional either.
Best regards,
Dominic
Reply by ●November 17, 20062006-11-17
--- In l..., "bobtransformer" wrote:
>
> My problem with FPGAs like these is that there is little or
> not analog on them, like A/Ds. Isn't that right ?
> It would be nice if I was wrong. Hope I am and maybe I
> just haven't looked hard enough.
> boB
>
FPGAs with analog blocks are showing up. Actel is selling Fusion
mixed-signal series with configurable analog blocks. Incidentally
Fusion devices also support soft ARM7TDMI core!
But it is not exactly a low cost approach, yet.
elektrknight
> --- In l..., "tsvetanusunov" wrote:
> >
> > > So my thoughts at this point are to roll-my-own SoC. I think that
> is
> > the real solution I need. After discovering how much you can do
> with
> > these and how easy they are, I probably will do a lot more of these
> in
> > the future.
> >
> > good point, but you don't mention what is the cost of the high
> speed
> > USB and Giga Ethernet IPs? I guess 4 digit numbers? and
> Verilog/VHDL
> > guys usually charge more per design than the general micro embedded
> > guys, so if you don't do this yourself but have to subcontract the
> > design the NRE cost will go too high for most of general
> applications
> > IMO, so there is still too much room for the old fashioned MCUs,
> but I
> > may be wrong.
> >
> > Best regards
> > Tsvetan
> > ---
> > PCB prototypes for $26 at http://run.to/pcb
> (http://www.olimex.com/pcb)
> > Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
> > (http://www.olimex.com/dev)
>
>
> My problem with FPGAs like these is that there is little or
> not analog on them, like A/Ds. Isn't that right ?
> It would be nice if I was wrong. Hope I am and maybe I
> just haven't looked hard enough.
> boB
>
FPGAs with analog blocks are showing up. Actel is selling Fusion
mixed-signal series with configurable analog blocks. Incidentally
Fusion devices also support soft ARM7TDMI core!
But it is not exactly a low cost approach, yet.
elektrknight
> --- In l..., "tsvetanusunov" wrote:
> >
> > > So my thoughts at this point are to roll-my-own SoC. I think that
> is
> > the real solution I need. After discovering how much you can do
> with
> > these and how easy they are, I probably will do a lot more of these
> in
> > the future.
> >
> > good point, but you don't mention what is the cost of the high
> speed
> > USB and Giga Ethernet IPs? I guess 4 digit numbers? and
> Verilog/VHDL
> > guys usually charge more per design than the general micro embedded
> > guys, so if you don't do this yourself but have to subcontract the
> > design the NRE cost will go too high for most of general
> applications
> > IMO, so there is still too much room for the old fashioned MCUs,
> but I
> > may be wrong.
> >
> > Best regards
> > Tsvetan
> > ---
> > PCB prototypes for $26 at http://run.to/pcb
> (http://www.olimex.com/pcb)
> > Development boards for ARM, AVR, PIC, MAXQ2000 and MSP430
> > (http://www.olimex.com/dev)
>
Reply by ●November 17, 20062006-11-17
--- In l..., wrote:
>
> Well, here are my thoughts right now. After looking over both
these LPC and STR parts, I am not happy with either of them.
Neither part gives me what I am really after. I started out
planning to use a Cypress CY68013 USB2 controller to get hi-spd USB,
then some other ARM9 MCUs, then went to the STR912 route, and now
looked at LPC2378.
>
> Some of the other posters here made the comment that I was
choosing the wrong part for the job. I agree with them. I can't
find the right part for the job I want to do in one chip. I don't
think it exists.
>
I think you're dead right about these parts. I wouldn't describe
either part as a fast comms processor. I'd describe them as
reasonably fast processors with plenty of comms capabilities: not
the same thing at all. If you need the kind of data throughput
you're talking about, I'd recommend specialist devices (Freescale,
Infineon spring to mind as suppliers, as well as a bunch of
specialist suppliers that typically use MIPS cores).
What I don't understand is why you seemed to think the LPC was
significantly worse than the STR9: as with most of the devices there
are pluses and minuses with each product. Neither is perfect, and
neither is a dog.
> So, I just finished an SoC on an Altera Cyclone FPGA. It was
fantastic. I could create exactly what I needed with soft cores in
terms of bus control, perfs, and CPU. $4 part that does everything
perfectly as I desired. Everything integrated in a single part.
>
> I started thinking about this same approach last night for this
problem, and checking on the available IP cores. It looks very
doable. I can put in both a USB2-480MHz and Enet-1000 which is what
I really wanted from the start. Easy to get external PHYs for
both. Then I can create my own full 32 bit DMA chs between my
external 32 bit bus, external 16 bit bus, and the ENET and USB
exactly how I want it. Virtually all hardware piping, little or no
CPU involvement. Blazing fast. Not only that, I can put all of the
pins exactly where I what them. Hey, what a deal.
>
> So my thoughts at this point are to roll-my-own SoC. I think that
is the real solution I need. After discovering how much you can do
with these and how easy they are, I probably will do a lot more of
these in the future.
>
> Chris.
>
And your development/NRE budget is????
Brendan
>
> Well, here are my thoughts right now. After looking over both
these LPC and STR parts, I am not happy with either of them.
Neither part gives me what I am really after. I started out
planning to use a Cypress CY68013 USB2 controller to get hi-spd USB,
then some other ARM9 MCUs, then went to the STR912 route, and now
looked at LPC2378.
>
> Some of the other posters here made the comment that I was
choosing the wrong part for the job. I agree with them. I can't
find the right part for the job I want to do in one chip. I don't
think it exists.
>
I think you're dead right about these parts. I wouldn't describe
either part as a fast comms processor. I'd describe them as
reasonably fast processors with plenty of comms capabilities: not
the same thing at all. If you need the kind of data throughput
you're talking about, I'd recommend specialist devices (Freescale,
Infineon spring to mind as suppliers, as well as a bunch of
specialist suppliers that typically use MIPS cores).
What I don't understand is why you seemed to think the LPC was
significantly worse than the STR9: as with most of the devices there
are pluses and minuses with each product. Neither is perfect, and
neither is a dog.
> So, I just finished an SoC on an Altera Cyclone FPGA. It was
fantastic. I could create exactly what I needed with soft cores in
terms of bus control, perfs, and CPU. $4 part that does everything
perfectly as I desired. Everything integrated in a single part.
>
> I started thinking about this same approach last night for this
problem, and checking on the available IP cores. It looks very
doable. I can put in both a USB2-480MHz and Enet-1000 which is what
I really wanted from the start. Easy to get external PHYs for
both. Then I can create my own full 32 bit DMA chs between my
external 32 bit bus, external 16 bit bus, and the ENET and USB
exactly how I want it. Virtually all hardware piping, little or no
CPU involvement. Blazing fast. Not only that, I can put all of the
pins exactly where I what them. Hey, what a deal.
>
> So my thoughts at this point are to roll-my-own SoC. I think that
is the real solution I need. After discovering how much you can do
with these and how easy they are, I probably will do a lot more of
these in the future.
>
> Chris.
>
And your development/NRE budget is????
Brendan
Reply by ●November 17, 20062006-11-17
Developing solutions with fpga components usually requires 6 digit budget.
Reading the data specs is not enough as in microcontroller world.
Fpga development needs experience and knowledge which is expensive.
There are software tools that could bring the cost down and most importantly to cut the development time.
I recommend altium designer which is , I believe, years ahead from the competition.
But if someone thinks that fpga is the tool for ALL applications, he is wrong.
Today there are microcontrollers with processing power, I/O count and embedded peripherals offered at unbelievable low prices.
Sometimes when plenty are being offered the task to choose from is a very difficult one and sometimes frustrated.
Regards,
Viron.
----- Original Message ----
From: Brendan Murphy
To: l...
Sent: Friday, November 17, 2006 10:02:23 PM
Subject: [lpc2000] Re: LPC2148 with ethernet
--- In lpc2000@yahoogroups .com, wrote:
>
> Well, here are my thoughts right now. After looking over both
these LPC and STR parts, I am not happy with either of them.
Neither part gives me what I am really after. I started out
planning to use a Cypress CY68013 USB2 controller to get hi-spd USB,
then some other ARM9 MCUs, then went to the STR912 route, and now
looked at LPC2378.
>
> Some of the other posters here made the comment that I was
choosing the wrong part for the job. I agree with them. I can't
find the right part for the job I want to do in one chip. I don't
think it exists.
>
I think you're dead right about these parts. I wouldn't describe
either part as a fast comms processor. I'd describe them as
reasonably fast processors with plenty of comms capabilities: not
the same thing at all. If you need the kind of data throughput
you're talking about, I'd recommend specialist devices (Freescale,
Infineon spring to mind as suppliers, as well as a bunch of
specialist suppliers that typically use MIPS cores).
What I don't understand is why you seemed to think the LPC was
significantly worse than the STR9: as with most of the devices there
are pluses and minuses with each product. Neither is perfect, and
neither is a dog.
> So, I just finished an SoC on an Altera Cyclone FPGA. It was
fantastic. I could create exactly what I needed with soft cores in
terms of bus control, perfs, and CPU. $4 part that does everything
perfectly as I desired. Everything integrated in a single part.
>
> I started thinking about this same approach last night for this
problem, and checking on the available IP cores. It looks very
doable. I can put in both a USB2-480MHz and Enet-1000 which is what
I really wanted from the start. Easy to get external PHYs for
both. Then I can create my own full 32 bit DMA chs between my
external 32 bit bus, external 16 bit bus, and the ENET and USB
exactly how I want it. Virtually all hardware piping, little or no
CPU involvement. Blazing fast. Not only that, I can put all of the
pins exactly where I what them. Hey, what a deal.
>
> So my thoughts at this point are to roll-my-own SoC. I think that
is the real solution I need. After discovering how much you can do
with these and how easy they are, I probably will do a lot more of
these in the future.
>
> Chris.
>
And your development/ NRE budget is????
Brendan
____________________________________________________________________________________
Sponsored Link
$200,000 mortgage for $660/ mo -
30/15 yr fixed, reduce debt -
http://yahoo.ratemarketplace.com
Reading the data specs is not enough as in microcontroller world.
Fpga development needs experience and knowledge which is expensive.
There are software tools that could bring the cost down and most importantly to cut the development time.
I recommend altium designer which is , I believe, years ahead from the competition.
But if someone thinks that fpga is the tool for ALL applications, he is wrong.
Today there are microcontrollers with processing power, I/O count and embedded peripherals offered at unbelievable low prices.
Sometimes when plenty are being offered the task to choose from is a very difficult one and sometimes frustrated.
Regards,
Viron.
----- Original Message ----
From: Brendan Murphy
To: l...
Sent: Friday, November 17, 2006 10:02:23 PM
Subject: [lpc2000] Re: LPC2148 with ethernet
--- In lpc2000@yahoogroups .com, wrote:
>
> Well, here are my thoughts right now. After looking over both
these LPC and STR parts, I am not happy with either of them.
Neither part gives me what I am really after. I started out
planning to use a Cypress CY68013 USB2 controller to get hi-spd USB,
then some other ARM9 MCUs, then went to the STR912 route, and now
looked at LPC2378.
>
> Some of the other posters here made the comment that I was
choosing the wrong part for the job. I agree with them. I can't
find the right part for the job I want to do in one chip. I don't
think it exists.
>
I think you're dead right about these parts. I wouldn't describe
either part as a fast comms processor. I'd describe them as
reasonably fast processors with plenty of comms capabilities: not
the same thing at all. If you need the kind of data throughput
you're talking about, I'd recommend specialist devices (Freescale,
Infineon spring to mind as suppliers, as well as a bunch of
specialist suppliers that typically use MIPS cores).
What I don't understand is why you seemed to think the LPC was
significantly worse than the STR9: as with most of the devices there
are pluses and minuses with each product. Neither is perfect, and
neither is a dog.
> So, I just finished an SoC on an Altera Cyclone FPGA. It was
fantastic. I could create exactly what I needed with soft cores in
terms of bus control, perfs, and CPU. $4 part that does everything
perfectly as I desired. Everything integrated in a single part.
>
> I started thinking about this same approach last night for this
problem, and checking on the available IP cores. It looks very
doable. I can put in both a USB2-480MHz and Enet-1000 which is what
I really wanted from the start. Easy to get external PHYs for
both. Then I can create my own full 32 bit DMA chs between my
external 32 bit bus, external 16 bit bus, and the ENET and USB
exactly how I want it. Virtually all hardware piping, little or no
CPU involvement. Blazing fast. Not only that, I can put all of the
pins exactly where I what them. Hey, what a deal.
>
> So my thoughts at this point are to roll-my-own SoC. I think that
is the real solution I need. After discovering how much you can do
with these and how easy they are, I probably will do a lot more of
these in the future.
>
> Chris.
>
And your development/ NRE budget is????
Brendan
____________________________________________________________________________________
Sponsored Link
$200,000 mortgage for $660/ mo -
30/15 yr fixed, reduce debt -
http://yahoo.ratemarketplace.com
