Microcontrollers with dual (or more) ADC's...

[Skimask]

Looking for more options...

Specifically, I'm wanting an MCU that has more than one ADC module.  NOT multiple ADC inputs to a single ADC module, but more than one ADC module, so two signals can be sampled simultaneously rather than sequentially.

The only thing I've found so far is TI's Stellaris line.  I'm not too keen on those, but if they work, then they'll do.

[kripton2035]

there are lots of simultaneous sampling adc, but with only an adc, it's not part of an mcu
look at TI and analog devices.
I've never seen simultaneous adc inside a mcu.

[andersm]

Some of Microchip's PIC24 and dsPIC33 chips have 2 ADCs.

[nctnico]

NXP has a few controllers in the LPC2000 series with dual ADCs. What do you want to do? Many ARM based microcontrollers have ADCs which can sample at 1Ms/s. If you let such an ADC system run sequentially you'll have 1us delay between the samples.

[Skimask]

@kripton2035 - Just looked over the datasheet for the LM4F120 series.  Two separate ADC modules with the ability to trigger and sample both of them simultaneously.

@andersm - I skipped over the PIC24/33, basically straight to the PIC32MX.

@nctnico - I'll have a look-see.

[kripton2035]

yes - sorry for me an arm is no more a microcontroller, it's bigger than that !
I was only looking in the pic-atmel range !

[TinkeringSteve]

stm32f103 has two ADC units

EDIT:
here's a small not hugely expensive dev board that has it, in case you care (that version does not have many pins, though...)
https://www.olimex.com/Products/ARM/ST/STM32-H103/

[Skimask]

I've got 3 of the Stellaris Launchpad dev boards on the bench.  That's why I looked at those first, but I'm not all that familiar with them...programming/memory map/register set/etc.
Mapping out the pins now to see if I can combine everything together to "fit" to that dev board.

Not very familiar with the STM range either, but looking at the datasheets now.

I'm also going to be wanting to feed 4 PWM modules for "DAC" outputs to give me some sine waves to work with.  With any luck, one of these MCU's will have some sort of DMA module that'll be able to feed the PWM modules.

Lots of stuff to look at...

In case anybody is wondering, this all ties back to the analog signature analyzer I'm building.  Original plan had a "master" PIC doing control/display, and a handful of "slave" PICs to the DAC, ADC, keypad scanning, relay selection, as well as 2 6 channel digital pots (AD5206 types) doing duties as gain changing in the opamps, and a handful of other things.
Just changing the design over to this LM4F120H5QR chip would put every one of those PICs onto the same chip, leaving only a very small handful of external components.  But, I'm still looking at other MCUs...

[TinkeringSteve]

that stm32 also has a 12bit DAC with 2 channels, which you can clock up to 1 MHz I believe (like the ADCs), I guess you won't get that resolution & sample rate with your PWM stuff :-)
And it has DMA, which you can use combind with the ADC and DAC.

And with the 72 MHz & 32bit, 1-cycle hardware mul & hardware div, I reckon you can also have all your PICs replaced by one of the stm32f103's.

[Skimask]

The LM4120 has 12 bit A/D, but no DAC.  Don't need much for frequency at the PWM outputs, maybe 10Khz at the most.
Having a look at the STM range now...

[TinkeringSteve]

In case you like what you see there:

To quick-start with the stm32f10x series, there is the free (of charge) IDE called CooCox CoIDE, the nice thing isthat it lets you select the MCU variant you have (it supports most of the ST cortex m3 & m0 ones, and even some M4), and it will provide you with the correct startup code for it, and a selection of available library components, for e.g. adc, dac, timer etc etc - and list a bunch of example code files for the selected component library, with everything coming right with the IDE.
It's not AAA+ code, but mostly works well enough to demonstrate how it's done ;-)

The IDE is an extremely stripped down version of Eclipse, e.g.. all those crowded  menus are gone.
On their page are instructions how to set up the free ARM compiler for it, it's pretty easy.

[TinkeringSteve]

Btw, why is it that important that you have 2 ADCs and don't use multiplexed inputs ?
I mean, if you scan rather fast, it's almost like using 2 ADCs ?
There are also stm32f100 things for 2..3 bucks or so, which only have 1 ADC and can only do 24 MHz, but it may still be fast enough to capture 2 inputs "sort of at the same time" ...
For that thing there are cheap dev boards from ST, like the stm32vldiscovery for 15 bucks or so, with included debugger (which can also be used to debug external (from the devboard) stm32 ICs by changing 2 jumpers.)

[Leo Bodnar]

I have used multiple 14-bit external SPI ADCs by simply paralleling their SPI CS/CLK/MOSI pins and sticking MISO pins onto a parallel port input.
If you bit-bang SPI clock and read ADC data in parallel you'll have them synced to a few tens of nsec. 
This way you can do synchronous 8 or 16-channel ADC sampling system even on the simplest of the MCUs.

Try MCP4921 or AD stuff.  They are 8 pin parts.

No multiplexing means no settling time, etc.  Additional benefit is they have true differential inputs.

Leo

[Skimask]

It's for an analog signature analyzer, with a digital twist rather than a CRT (like the Huntron Tracker 2000), measuring voltage over current, like the old octopus adapters for an o'scope.
At lower frequencies, it may very well be possible to do 2 successive AD conversions on 2 different channels, and get good results.  But, go to the higher frequencies, and that 2nd AD result will be somewhat skewed.
With a regular ol' PIC (e.g. PIC18F types), about the best you can do is 12us between AD conversions on 2 different channels.  With an input signal freq of 100hz, two successive samples spaced 12us apart are only about ~.43 degrees different.  Jump up to 1Khz, and the same sampling rate ends up getting spaced out by 4.33 degrees (assuming the input signal was a sine wave).  (10Khz =  43.3 degrees differences).

The original plan was to use a couple of PIC12F1822's, sync'd and triggered to take samples at the same time and send the data back when selected as required.  More chips, more pins, more programming...

[Skimask]

@Leo - Got an example part number?
Scratch that...just did a quicky search on Digikey.

MCP3001 was my first hit, and that sounds like it might not be a bad idea.  A small A/D chip to keep the signal input right where it needs to be rather than running a trace/wire halfway across the PCB picking up noise.

[Leo Bodnar]

@Leo - Got an example part number?
Scratch that...just did a quicky search on Digikey.

MCP3001 was my first hit, and that sounds like it might not be a bad idea.  A small A/D chip to keep the signal input right where it needs to be rather than running a trace/wire halfway across the PCB picking up noise.

Yes, that's the idea!  This way isolating analogue and digitals sides is so much easier.
There are tonnes of them around.  E.g. MCP3201 is 12-bit one.  In general, search for 8-pin SPI ADC parts.  Bit resolution, single-ended or differential inputs, max sampling rate is up to you!

Leo

[Niklas]

I think that some of the PICs, and most probably also ARMs, STM32s, AVRs, aimed at motor control have fast ADCs that sample several signals simultaneously. Used for measuring both voltage and current to get the field vectors etc.

[Skimask]

Don't really need a high sample rate..wouldn't hurt though.
Overall idea is to "paint a picture" of a component/circuit rather than get real time results.  e.g. sampling a 1Khz signal at 1Khz would give me the same point on the waveforms every time thru, whereas sampling at an odd rate would grab samples at different places along that waveform and eventually build up a complete picture assisted incorporating a sort of "digital phosphorescence" (is that a phrase?) in software.
Even still, these signals are aren't going to be much more than 10Khz at the high end, but still could get noise injected in there...somehow, someway.

@Niklas - will check into that.  BLDC controller types seem like they would be the likely candidates.

[Kremmen]

Looking for more options...

Specifically, I'm wanting an MCU that has more than one ADC module.  NOT multiple ADC inputs to a single ADC module, but more than one ADC module, so two signals can be sampled simultaneously rather than sequentially.

The only thing I've found so far is TI's Stellaris line.  I'm not too keen on those, but if they work, then they'll do.

I am familiar with Atmel processors, and at least the 8/16 bit ATXMega has the capability to sample 2 channels simultaneously. It is simple enough to apply. Of the 32 bit ones at least UC3C family has a dual ADC. It may not be the most common processor family, but i like the free development environment without any artificial restrictions. And of course for both, the actual toolchain is the familiar GCC we all know and love.

[kripton2035]

simultaneous adc chips :
AD7606_7606-6_7606-4.pdf
AD7608.pdf
AD7654 16B 2CH.pdf
AD7865 14B 4CH.pdf
ads5204 2X10B 40MSP.pdf
ads8361.pdf
ads8364.pdf
MAX11040K-MAX11060.pdf
ths0842 2X8B 40MSP.pdf

[M. András]

http://www.ti.com/lit/ds/symlink/ads1274.pdf , it has a 16 bit part in the same package just adding it to your list

[Corporate666]

Have you checked out the Cypress PSoC? 

[kripton2035]

Have you checked out the Cypress PSoC? 

they have 2 even 3 adc some psoc chips, but none that can do real simultaneous sampling.

[tophathacker]

TI's Stellaris chips are pure awesome .. you should look no further!
Too bad you don't have 1.. or 4 of those on hand...

... tro lolo lo lo

[nctnico]

Too bad for you TI seems to have pulled the plug on the Stellaris devices.