Dangerous Prototypes

Hi,
I just wanted to point out that Microchip seem to plan to release a new set of PIC32 chips in DIP 28 Pins packages and other low pins 36 and 44 QFP ...
I discovered the list from Microchip product selector for PIC32

http://www.microchip.com/ParamChartSear ... &pageId=74 (http://www.microchip.com/ParamChartSearch/chart.aspx?branchID=211&mid=10&lang=en&pageId=74)

All those low pins count (<64) are marked as future products, but the datasheet are so complete so I think they will not take to much time to get to market

datasheet: http://ww1.microchip.com/downloads/en/D ... 61168B.pdf (http://ww1.microchip.com/downloads/en/DeviceDoc/61168B.pdf)

Half the speed of the larger cousins (40mhz vs 80) but still nice to have 32bit mips in friendly packages.

Some of the DIP packaged PIC32 chips are in production now and are extremely nicely priced! 8 bit PIC18Fxx is over I think ! really over

They also sport PPS which the larger ones do not.

I've read somewhere (Parallax forums I think) that NXP may be offering their Cortex-M0 (LPC111x ??) in a 28-pin DIL.

That would be killer!

I'd like to see CortexM3 in 28DIL but what I'd really love is to see SPARTAN3 in 28 and 40 pin DIL and also SPARTAN6 with 3k or more slices in 40-pin DIL and some tqfp64 and tqfp100 .. I never figured out why they have to increase the number of pins and move to BGA with number of slices :( ... I'd be perfectly happy with a 100MHz 15000 slices (XC6SLX100 for e.g.) in a 40pin dip :D but no they had to make it 300+ pins bga completely unusable for home use :(

[quote author="arhi"]I'd like to see CortexM3 in 28DIL but what I'd really love is to see SPARTAN3 in 28 and 40 pin DIL and also SPARTAN6 with 3k or more slices in 40-pin DIL and some tqfp64 and tqfp100 .. I never figured out why they have to increase the number of pins and move to BGA with number of slices :( ... I'd be perfectly happy with a 100MHz 15000 slices (XC6SLX100 for e.g.) in a 40pin dip :D but no they had to make it 300+ pins bga completely unusable for home use :([/quote]

i agree, why limit the more powerful devices to higher pin-count packages?

I assume the die on those packages are big so they have room for a lot of pins ... also I assume you can't to GHz+ speeds on regular tqfp but I don't mind having 200MHz peripherals ... even 200MHz for the whole device is ok (would be cool to have GHz internal clock and 200MHz peripherals clock :D ) ... for e.g. ZYNQ, to have arm's inside run on GHz+ and fpga part at 200MHz on the TQFP package ... they said zynq will cost 10-15$ and that it will be available in 1pc quantity, but with it being BGA it's next to impossible to use it :( ... on the other hand, I'm fairly sure if they made a high power device in a DIP40 they would be selling them faster then they can make them

[quote author="TonyD"]I've read somewhere (Parallax forums I think) that NXP may be offering their Cortex-M0 (LPC111x ??) in a 28-pin DIL.[/quote]

Any idea what the port drive strength of LPC110FD20 would be ?  http://http://www.nxp.com/products/microcontrollers/cortex_m0/lpc1100l/LPC1110FD20.html#quickreference lists every other parameter except the port drive strength. Existing NXP LPC111x ports normally provide 2mA drive strength except for one port pin which is 20mA.

Eindhoven, Netherlands and San Jose, California, October 25, 2011 – NXP Semiconductors N.V. (NASDAQ: NXPI) today announced the availability of new low-pin-count package options – SO20, TSSOP20, TSSOP28 and DIP28

ow hello there ARM Cortex in DIP, how are you today:

http://www.nxp.com/news/press-releases/ ... tions.html (http://www.nxp.com/news/press-releases/2011/10/nxp-cortex-m0-microcontrollers-in-high-volume-tssop-and-so-packages-target-8-16-bit-applications.html)

i wonder if this will make front page news :)

@viswesr

http://www.nxp.com/products/microcontro ... l#overview (http://www.nxp.com/products/microcontrollers/cortex_m0/lpc1100l/LPC1110FD20.html#overview)
it states this:
High-current output driver (20 mA) on one pin

That right there is sexy. Send a message via the contact form and they'll post it up.

[quote author="arhi"]I'd like to see CortexM3 in 28DIL[/quote]

TI already offers CortexM3 in DIP packages (originally made by Luminary).

I missed that !!!

@arhi to answer your question as to why You would never see a FPGA or for that matter an applications processor(think OMAP/Blackfin) in a In-Line package is a fairly straightforward one. Those long leads that a DIL package sports start to look line sub-uH inductors at clocks speeds exceeding 100MHz. So, your high speed transitions coupled with this sub uH inductor and the trace capacitance is sufficient to set up resonances.  So the general rule of thumb for  is that for any programmable device running at >= 100Mhz, they go for lead-less package(QFN,QFP,BGA,TQFP). To minimize any parasitic inductance is of paramount importance to maintain good signal integrity and the package houses know this better than anybody else. I hope this makes sense to You!

Cheers,
Ananth

I know that oh too well ... that's why I said 200MHz peripheral clock and higher internal clock. I'd be happy with 40MHz total clock too (and up to 80MHz should work on DIP without a problem and up to 200MHz with SOIC should also work ok) ... thing is, I often need "a lot of gates on not too high speed with low pin count" but "breadboardable" .. (and as it is breadboardable 40MHz is the limit anyhow) and then I have to make a breakout board for fpga, and to break out only few pins I need .. anyhow, I know they will never make it so ..

So, your high speed transitions coupled with this sub uH inductor and the trace capacitance is sufficient to set up resonances. So the general rule of thumb for is that for any programmable device running at >= 100Mhz, they go for lead-less package(QFN,QFP,BGA,TQFP). To minimize any parasitic inductance is of paramount importance to maintain good signal integrity and the package houses know this better than anybody else.

What if we need FPGA to work on frequencies under 100 MHz?...

[quote author="octal"][quote author="arhi"]I'd like to see CortexM3 in 28DIL[/quote]

TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]

What part numbers?  I don't see anything DIP on TI's website, just SOIC.

[quote author="octal"][quote author="arhi"]I'd like to see CortexM3 in 28DIL[/quote]

TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]

I think he meant SSOP, I am not aware of DIP. TSSOP is easier than QFP.

I would like to some ARM TSSOP with USB though... These look like no go and LM3S only has USB on 64 pin LQFP and up.

I actually like dip's for breadboarding ... then when I make pcb I prefer soic/ssop etc as those are faster to solder then dip's .. everything with leads is acceptable (I prever 0.5mm or more pitch but even 0.4mm can be soldered by hand), I hate leadless packages

[quote author="octal"][quote author="arhi"]I'd like to see CortexM3 in 28DIL[/quote]

TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]
Which ones?  The closest I see is 28SOIC for some of the X00 series

Today I've got one PIC32 in DIP28 ;)

I want one! Did your order it as a sample from Microchip?

holy ... they are real :D .. where did you order from ? digi? farnell ? sample from mchip

It was free sample from microchipdirect :)
Ordered 10 days ago
Today got package from Thailand with signature confirmation

Thanks for posting my picture on your blog ;)
2 weeks ago Microchip said that DIP28 will be available for purchase on 09-Dec-2011, but right now it is 21-Dec-2011
Anyway - I'm trying to create "hobbyist-friendly" and "breadboard-ready" kit around this chip as simple as possible with pinouts identical to my nedoCPU-16 (http://http://youtu.be/5MiIuuXSkg8) board
But I've never had a chance to work with PIC32 yet, so may I ask for advice/feedback here?
P.S. Pinouts of nedoCPU family:
X1: A0,A1,A2,A3,A4,A5,+5V,GND (for PIC32 instead of A5 it's +3.3V from on-board voltage regulator)
X2: B0,B1,B2,B3,B4,B5,B6,B7
X3: C0,C1,C2,C3,C4,C5,C6,C7 (for PIC32 it's actually B8...B15, but header is named Cn for compatibility reason)

---

[quote author="Shaos"]
But I've never had a chance to work with PIC32 yet, so may I ask for advice/feedback here?
[/quote]

You can always ask :D (better to open a separate thread for it) question is if we will know the answer :D

[quote author="arhi"][quote author="Shaos"]
But I've never had a chance to work with PIC32 yet, so may I ask for advice/feedback here?
[/quote]

You can always ask :D (better to open a separate thread for it) question is if we will know the answer :D[/quote]

Sorry for "hijacking" this thread ;)
New thread: viewtopic.php?f=56&t=3085 (http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=3085)
P.S. So question is will it be enough to go?

[quote author="Shaos"]
2 weeks ago Microchip said that DIP28 will be available for purchase on 09-Dec-2011, but right now it is 21-Dec-2011
[/quote]

Yesterday they said 23-Dec-2011
Right now it's 26-Dec-2011
So availability date is constantly moving - is it a bad sign?...

Estimated Availability:
03-Jan-2012
:(

[quote author="octal"][quote author="arhi"]I'd like to see CortexM3 in 28DIL[/quote]

TI already offers CortexM3 in DIP packages (originally made by Luminary).[/quote]

Can you cite a part number?  TI's parametric search claims otherwise.

Sadly, the NXP part is also still "in development".  It would be super-amusing if it had a pinout compatible with the 28-pin AVRs (it really should, since the AVR pinout is virtually the only logical layout for a DIP CPU at that sort of speed).

 = Mike

LM3S1xx have soic

DigiKey has 2 of them listed: LM3S101-IRN20-C2 and LM3S102-ERN20-C2T - both are "Non-Stock"...
P.S. The same on ti.com - "No Stock"...

It's like when the ENC28J60 debuted, we'll be lucky to buy them next summer ;)

[quote author="Shaos"]Estimated Availability:
03-Jan-2012
:([/quote]

microchipdirect.com got 870 SPDIPs "In Stock"!
but no prices and no "Add to Card" button for it...
yet?

[quote author="Shaos"][quote author="Shaos"]Estimated Availability:
03-Jan-2012
:([/quote]

microchipdirect.com got 870 SPDIPs "In Stock"!
but no prices and no "Add to Card" button for it...
yet?[/quote]

now they have prices and ability to buy!

Quantity 	USD per Unit
1-25 		2.51
26-99 		2.30
100+ 		2.08 

Wow, that is a really good price compared to 1-off 18Fs. Maybe subsidized? Thanks for the update, I'll post this up.

How many versions of pic32 are available? Any with built in ethernet? Are there any dip package Ethernet microcontrollers out there?

[quote author="moredp"] Are there any dip package Ethernet microcontrollers out there?[/quote]

PIC32MX220F032B
PIC32MX210F016B
PIC24FJ64GB002
PIC24FJ32GB002
18F27J53
18F2550 DIP28
18F4550 DIP40

Arhi,
he is asking about ETHERNET and *NOT* USB DIP chips ;) 
At my knowledge, There are no pic chips with ethernet hardware in DIP package.

:D :D :D :D :D :D :D :D :D  :D :D :D

Answering questions at 05:51AM obviously was not a very good idea :D

[quote author="moredp"]How many versions of pic32 are available? Any with built in ethernet? Are there any dip package Ethernet microcontrollers out there?[/quote]

There are some pic18f controller with builtin 10Mbit/s ethernet and some highend pic32 with 100/10 Mbit/s ethernet. The pic18 can be directly connected to a magjack and the pic32 need some sort of MII/RMII interface.

Here is the microchip part selector. You can choose ethernet and it lists all the mcu with ethernet.

http://www.microchip.com/productselecto ... ector.html (http://www.microchip.com/productselector/MCUProductSelector.html)

[quote author="Sjaak"][quote author="moredp"]How many versions of pic32 are available? Any with built in ethernet? Are there any dip package Ethernet microcontrollers out there?[/quote]

There are some pic18f controller with builtin 10Mbit/s ethernet and some highend pic32 with 100/10 Mbit/s ethernet.  [/quote]

But those are NOT DIP packages !

Your first question was about just ethernet :) But I guess i suffer the same issue as Arhi (don't post when you just flew back from .jp to .nl :)

I guess the signals are too fast for DIP. If you want an all-dip alternative use a enc24 breakout board and a dip pic32 chip.

Hi Sjaak,
The question was from moredp !!!

dont post right after having finished the bottle :D

+ one disadvantage of 18f with ethernet is that you cannot program them as many times as the other ones. For example #twatch (http://http://dangerousprototypes.com/docs/Twatch_v2_network_LCD_hardware_design#Hardware) uses a PIC 18F67J60 uC which can be programmed only 100 times. + it is only SMD, pin count is too high for DIP.

which can be programmed only 100 times.

I have found I get significantly more than this. This is rated for the extremes, at normal room temp and 3.3volts you can probably get 1000. Still not great, but better. I guess (no idea!) that it is due to the flash having to tolerate the extra heat from the Ethernet MAC/PHY, so it is a specific type that burns out faster.

PIC + ENC24 modules seem quite popular. But has anyone tried to use a PIC with one of the silabs CP220x ethernet controllers?

thanks guys. I figured there wasn't a combined ethernet/microcontroller dip package chip. Here is hoping they make a pic32mx5/6/7 in dip package as those are the pic32s with built in ethernet.

Peter sent me the DIP and SOIC size chips. They came today, so I showed them off in today's video blog.

Does anyone here have an idea, why the PIC32MX1 and PIC32MX2 series chips are only rated for 40 MHz while the 3, 4, 5, 6, and 7 series can be clocked with 80 MHz?  Over the weekend I experimented with a PIC32MX220F032B and had absolutely no problem clocking it with 80 MHz.  Because the needed current per MHz is also lower for the 1 and 2 series than for the 3, 4, 5, 6, and 7 series (about 0.5 mA/Mhz compared to about 1 mA/Mhz) it does not even get as warm as the PIC32MX7 I also used.  Maybe it's just a market segmentation strategy from Microchip?

IIRC only the DIP packaged 32mx's are 40MHz rated

No, if you look here http://www.microchip.com/ParamChartSear ... &pageId=74 (http://www.microchip.com/ParamChartSearch/chart.aspx?branchID=211&mid=10&lang=en&pageId=74) you can see, that they are rated for 40 MHz regardless of the package.

yes but they "exist" in dip packages. because they exist in dip packages they limited them to 40mhz (I'm pretty sure they use exactly the same die for dip and soic) .. that's just what I think, I have no proof for that theory :)

Hi

if here are developers from Europe on this forum I would like to let them know that Olimex and Microchip will give away 1000 pcs PIC32MX220F032D development boards on Embedded World exhibition.

The forum do not allow me to post external web links so you can read the full details at olimex .dot. wordpress .dot. com :)

Best regards
Tsvetan

Hi Tsvetan, great to see ppl from Olimex here :) ... I'm interested for EWE but I can't see on the http://olimex.wordpress.com/ (http://olimex.wordpress.com/) where it will be. I see it is 28th February - 1st March. Europe is kinda big, are you organizing it in Sofia or ?

[quote author="arhi"]yes but they "exist" in dip packages. because they exist in dip packages they limited them to 40mhz (I'm pretty sure they use exactly the same die for dip and soic) .. that's just what I think, I have no proof for that theory :)[/quote]

Interesting theory.  IMO it would be quite "wasteful" to declare the whole series to be just 40 MHz because of the possibility to get them in PDIP packages.  If this was the reason, it would make more sense to say in the datasheet 80 MHz, with a remark that for PDIP packages only 40 MHz are reasonable.  However, I *did* use the MX2 in a PDIP package, and it worked flawlessly.  Further, why should the PDIP package limit the internal chip clock frequency?  The internal fast RC oscillator runs at only 8 MHz and the 80 MHz clock are then generated by means of the on chip PLL and pre- and post-divider.

I think that beyond 40MHz the PDIP pins will act like antenas.

Maybe, but as already sayed, the high clock frequency is purely *internal*.  No 80 MHz on any PDIP pin.

No Markus, one of the potential properties of PIC32 series is that they can toggle their pins at each clock cycle, with toggle instruction (1 bit manip). This means that (for 80MHz chips) you can got up to 40MHz freq on any digital pin. In practice, I personally got up to 32MHz (4 months ago) when I does some investigations. In PIC32, unlike ARM chips, you have no AHB (althought recent ARM chips corrected this bottlneck with fast IO and fast AHB).

[quote author="arhi"]Hi Tsvetan, great to see ppl from Olimex here :) ... I'm interested for EWE but I can't see on the where it will be. I see it is 28th February - 1st March. Europe is kinda big, are you organizing it in Sofia or ?[/quote]

Embedded World is in Nuremberg / Germany you can google it as the forum still do not allow me to post URLs :)

Ok, you are right, but I don't understand the argumentation.  As you sayed, it is possible to get *40* MHz on a pin.  Not 80 MHz.  Further, you can also do the fast pin toggling on a PIC32 in a small package with connected PCB traces and would also get "antennas".  And how does AHB (if it stands for Advanced High-performance Bus) relate to this?

However, only MCP would be able to tell the reasoning behind this.  In the meantime I am happy to have my PIC32MX2 running at 80 MHz :)

@olimex, :( I was hoping it's in Sofia, driving from Bgd to Sofia would be fun (I have friends there that I'd visit etc etc), but driving to Nuremberg is a no go :( ... good luck with it I hope you will make a lot of pics and share them here and on that wordpress blog you started :D

@markus, you can overclock them yes (you can overclock other pic mcu's too)... I'm sure they use same die for dip and soic and .. so having "different" spec for dip and mssop would be expensive hence it's better to just make them all 40. Also iirc the testing is done before the chips are packaged so testing is also same for all die's .. again, just a theory :D

Marcus, you got only 40MHz on any pin when your chip is at 80MHz, simply because you need 2 instructions to have a state change:
PinHigh  1 cycle
PinLow  1 cycle

both are done with Toggle instruction
Toggle
Toggle


so at 80MHz, you cannot have more than 40MHz frequency for a rectangular waveform (carré).
With your 40MHz chip you could have 20MHz.

AHB stuff is **not** related to that, I was meaning that AHB is a high speed bus, but on most LPCxxxx ARM7 chips for example working at 72MHz, AHB was unable to go beyond 32MHz, so you could'nt toggle a pin at more than 16 or 18MHz at max.
Now this changed with Fast IO and with new extra high speed AHxxx busses variants where you can go up to 60MHz I think on some high end LPCxxx chips working at 200MHz.

Microchip advertisements for PIC32 were always putting this thing to arg that their choice for a MIPS core is better than an ARM core because the MIPS core was better, consumes less and does not need an AHB like stuff and that it can toggle pins at the frequency of the MCU.

---

[quote author="Markus Gritsch"]And how does AHB (if it stands for Advanced High-performance Bus) relate to this?
[/quote]

That is more relating to how the ARM ic's were designed. I think the AHB is the bus that connects all the low speed peripherals, so the absolute fastest you could send data on the bus to the IO is at the clock of the bus. Since it takes two cycles to toggle the GPIO on an ARM (if I remember right) then you can only go at half the speed of the AHB. I would love to see how the pin toggling at 40 Mhz looks like on a scope, heh. :P

Also, one infuriating thing about the new pic32's is that the PICkit2 does not support them! :( I either have to make a pickit3 clone myself, buy a clone, or try to make the programmer myself. Unfortunately programming them is not easy, I was glancing at how they were programmed. It seems microchip does not want to load the page for my pic32mx220f032b, so I will try to recall this from memory. If you use ICSP then you need to put in a bootloader thing into the ROM first, then put in the program. Also, ICSP in this case is exactly the same as EJTAG except ICSP uses two wires while EJTAG uses four, or to put it more simply, the ICSP gets converted from 2 wire to 4 wire in the PIC and then sent to the core, while EJTAG goes straight to the core.

If I try to make a programmer then I will certainly aim to use EJTAG instead of ICSP because I always wanted to work with JTAG, plus, I would have an actual reason to play with QT!

Microchip forum says that pic32mx is supported by PICKit2 only from PICKit2 Software, not MPLAB (programming only)

P.S. I successfully tested PIC32MX110 in DIP on 80 MHz speed (http://http://dangerousprototypes.com/forum/viewtopic.php?f=56&t=3085&start=27), but if I understand correctly, program from flash will not run on full speed, because flash can't work on such speed, so processor has some kind of wait cycles or something (in my case it performed appox 55 millions instructions per second)

with frequencies over 40MHz you need to configure wait states in order to have communication between cpu and flash flowing.
You need to follow up the datasheet for the "safe" values. I had 32mx work with lower wait state config then datasheet specifies but later on it happened that out of 21 tested mcu's only 3 were able to do so and other 18 couldn't

[quote author="Shaos"](in my case it performed appox 55 millions instructions per second)[/quote]

You can set the required wait stated automatically by calling

    SYSTEMConfigPerformance( SYSCLK );

where SYSCLK is for example 80000000.  PIC32 MCUs have a cache built in, which you can enable or disable.  Depending on the code you are executing, speedup can be dramatic if it fits inside of the cache.  You can also execute code from RAM, which does not have the flash wait states.

good start for 32mx

...
  SYSTEMConfigPerformance(80000000L);
  mOSCSetPBDIV(OSC_PB_DIV_2);
  mJTAGPortEnable(0);
...

This is my code:

    #include <p32xxxx.h>
    #include <plib.h>

    //         internal    8    / 2              * 20               / 1 = 80

    #pragma config FPLLIDIV = DIV_2, FPLLMUL = MUL_20, FPLLODIV = DIV_1, FWDTEN = OFF
    #pragma config POSCMOD = OFF, FNOSC = FRCPLL, FPBDIV = DIV_2

    main()
    {
        // disable JTAG port
        DDPCONbits.JTAGEN = 0;

        SYSTEMConfigPerformance(80000000L);
....................

Also example with running from RAM was not working at all...

[quote author="Shaos"]Also example with running from RAM was not working at all...[/quote]

Did you try this example?
C:Program Files (x86)Microchipmplabc32v2.01examplesc32_examplesram_executionsource

[quote author="Markus Gritsch"][quote author="Shaos"]Also example with running from RAM was not working at all...[/quote]

Did you try this example?
C:Program Files (x86)Microchipmplabc32v2.01examplesc32_examplesram_executionsource[/quote]

Yes, it failed
Probably because I tried 80 MHz
May be I should try 40 MHz instead...

I'd like to pick up on the 40 MHz vs. 80 MHz discussion.  Recently I got three PIC32MX120F032B.  Two of them work fine at 80 MHz but the third one only works up to 72 MHz.  Maybe the reason they are only specified to work up to 40 MHz is so that Microchip can increase the yield (and achieve a lower price) by not having to selectively pick the ones thich work reliably at 80 MHz.