[Xenomai] Troubleshooting Analogy DIO

[Xenomai] Troubleshooting Analogy DIO

[Manu S Madhav]

Hello all,

I am running Ubuntu Precise (12.04) with kernel version 3.4.6 and
xenomai-2.6.2.1. I have a NI PCIe 6259 which I am trying to configure for
Digital I/O. Currently, the code itself is non-RT, I am trying first to
figure out how to do I/O as well as obtain counter / quadrature information
from my DAQ card.

Using Analogy, and with a bit of reading, I was able to get analog input
and output up and running without too much trouble, but digital I/O is
completely evading me. I am just trying to turn on all digital outputs to 1
using synchronous instruction to the DIO subdevice (#2, in my case). Size
of the subdevice is 4 (32 bits).

None of the following attempts worked. From the mailing list archives, I
found a suggestion to compile xenomai libs with --fno-omit-frame-pointer,
and that also didn't help. Any suggestions would be appreciated.


*First try (Using command-line insn_bits)*

********************
insn_bits -v -d analogy0 -s 2 0xffffffff 0xffffffff

insn_bits: device analogy0 opened (fd=0)
insn_bits: basic descriptor retrieved
     subdevices count = 14
     read subdevice index = 0
     write subdevice index = 0
insn_bits: complex descriptor retrieved
insn_bits: selected subdevice index = 2
insn_bits: mask = 0xffffffff
insn_bits: value = 0xffffffff
insn_bits: result = 0x0
********************

*Second try (Using a4l_sync_dio)*

********************
// dsc = is a4l_desc_t , previously configured
 int subd_DIO = 2;

uint32_t mask = 0xffffffff;
uint32_t buf = 0xffffffff;
a4l_sync_dio(&dsc, idx_subd_DIO, &mask, &buf);

********************

Again, even when I am getting a return of 0 (success), I am not measuring a
digital 1 at the physical pins.

*Third try (Using a4l_send_insn)*:

********************
// dsc = is a4l_desc_t , previously configured
int subd_DIO = 2;
unsigned char values[16];

    a4l_insn_t insn;
    insn.type = A4L_INSN_BITS;
    insn.idx_subd = idx_subd_DIO;
    insn.data = values;
    values[0] = 0xff;
    values[1] = 0xff;
    insn.data_size = 2 * sizeof(uint32_t);

    a4l_snd_insn(&dsc, &insn);

********************

Manu S Madhav

Re: [Xenomai] Troubleshooting Analogy DIO

[Gilles Chanteperdrix]

On 05/23/2014 11:43 PM, Manu S Madhav wrote:
> Hello all,
> 
> I am running Ubuntu Precise (12.04) with kernel version 3.4.6 and
> xenomai-2.6.2.1. I have a NI PCIe 6259 which I am trying to configure for
> Digital I/O. Currently, the code itself is non-RT, I am trying first to
> figure out how to do I/O as well as obtain counter / quadrature information
> from my DAQ card.
> 
> Using Analogy, and with a bit of reading, I was able to get analog input
> and output up and running without too much trouble, but digital I/O is
> completely evading me. I am just trying to turn on all digital outputs to 1
> using synchronous instruction to the DIO subdevice (#2, in my case). Size
> of the subdevice is 4 (32 bits).
> 
> None of the following attempts worked. From the mailing list archives, I
> found a suggestion to compile xenomai libs with --fno-omit-frame-pointer,
> and that also didn't help. Any suggestions would be appreciated.

I will let others answer this mail, but:
- the bug with omit-frame-pointer was fixed, I believe, in xenomai
2.6.0, or 2.6.1
- 2.6.2.1 is not the latest version, so, please try 2.6.3 before
reporting an issue that may have been fixed since then.

-- 
                                                                Gilles.

Re: [Xenomai] Troubleshooting Analogy DIO

[Paul Janzen]

Manu S Madhav <manusmad@gmail.com> writes:

> Using Analogy, and with a bit of reading, I was able to get analog input
> and output up and running without too much trouble, but digital I/O is
> completely evading me. I am just trying to turn on all digital outputs to 1
> using synchronous instruction to the DIO subdevice (#2, in my case). Size
> of the subdevice is 4 (32 bits).

Before writing to the DIO device, you have to configure each line that
you want to be a digital output as an output, otherwise it stays a
digital input:

for (i=0; i<num_channels; i++) {
    err = a4l_config_subd(dsc, subd_DIO, A4L_INSN_CONFIG_DIO_OUTPUT, i);
    if (err < 0) [...]
}

-- Paul

Re: [Xenomai] Troubleshooting Analogy DIO

[Manu S Madhav]

Thank you, Paul,

I was under the impression that a4l_config_subd was used for asynchronous
instructions only. DIO is up and running now - thank you!

One more question: Is anyone aware of an analogy implementation for using
the DAQ counter for quadrature encoding? E.g. by using the
A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER
instruction. I saw a couple of threads where someone successfully armed the
counter, but was unable to read information from it. I am going to try and
implement it this weekend, just wondering if there is already a working
solution.

Gilles: I will upgrade to the newest version of Xenomai. Thank you for your
suggestion.

---
Manu Madhav


On Sat, May 24, 2014 at 2:59 PM, Paul Janzen <pcj@xenomai.sez.to> wrote:

> Manu S Madhav <manusmad@gmail.com> writes:
>
> > Using Analogy, and with a bit of reading, I was able to get analog input
> > and output up and running without too much trouble, but digital I/O is
> > completely evading me. I am just trying to turn on all digital outputs
> to 1
> > using synchronous instruction to the DIO subdevice (#2, in my case). Size
> > of the subdevice is 4 (32 bits).
>
> Before writing to the DIO device, you have to configure each line that
> you want to be a digital output as an output, otherwise it stays a
> digital input:
>
> for (i=0; i<num_channels; i++) {
>     err = a4l_config_subd(dsc, subd_DIO, A4L_INSN_CONFIG_DIO_OUTPUT, i);
>     if (err < 0) [...]
> }
>
> -- Paul
>
>
>
>
>

Re: [Xenomai] Troubleshooting Analogy DIO

[Paul Janzen]

Manu S Madhav <manusmad@gmail.com> writes:

> One more question: Is anyone aware of an analogy implementation for
> using the DAQ counter for quadrature encoding? E.g. by using the
> A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER instruction.

I know it works in Comedi, but have not tried it in Analogy.  You may
have to snarf some bit definitions from comedi.

Try INSN_CONFIG, with two data
{ A4L_INSN_CONFIG_SET_COUNTER_MODE, 
  NI_GPCT_COUNTING_MODE_QUADRATURE_X4_BITS |
    NI_GPCT_COUNTING_DIRECTION_HW_UP_DOWN_BITS
}

and then again to arm the counter:
{ A4L_INSN_CONFIG_ARM, NI_GPCT_ARM_IMMEDIATE }

Then you should be able to get the counter value back with
a4l_sync_read. 

A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER appears to be a half-baked
sensoray only thing. 

-- Paul

Re: [Xenomai] Troubleshooting Analogy DIO

[Rosali Pyun]

> To: manusmad@gmail.com
> From: pcj@xenomai.sez.to
> Date: Sat, 24 May 2014 18:46:22 -0700
> CC: xenomai@xenomai.org
> Subject: Re: [Xenomai] Troubleshooting Analogy DIO
> 
> Manu S Madhav <manusmad@gmail.com> writes:
> 
> > One more question: Is anyone aware of an analogy implementation for
> > using the DAQ counter for quadrature encoding? E.g. by using the
> > A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER instruction.
> 
> I know it works in Comedi, but have not tried it in Analogy.  You may
> have to snarf some bit definitions from comedi.
> 
> Try INSN_CONFIG, with two data
> { A4L_INSN_CONFIG_SET_COUNTER_MODE, 
>   NI_GPCT_COUNTING_MODE_QUADRATURE_X4_BITS |
>     NI_GPCT_COUNTING_DIRECTION_HW_UP_DOWN_BITS
> }
> 
> and then again to arm the counter:
> { A4L_INSN_CONFIG_ARM, NI_GPCT_ARM_IMMEDIATE }
> 
> Then you should be able to get the counter value back with
> a4l_sync_read. 
> 
> A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER appears to be a half-baked
> sensoray only thing. 
> 
> -- Paul
> 


I was not able to use the counter for quadrature encoder with NI m-series DAQ card.I ended up using "quadrature clock converter" chip along with the encoder and implemented simple counting functionality.
INSN_CONFIG with {A4L_INSN_CONFIG_SET_COUNTER_MODE, NI_GPCT_COUNTING_MODE_NORMAL_BITS | NI_GPCT_COUNTING_DIRECTION_HW_UP_DOWN_BITS}
Let me know if you have any luck with quadrature encoder mode. 
-Rosali 		 	   		  

Re: [Xenomai] Troubleshooting Analogy DIO

[Manu S Madhav]

Hey Rosali, Paul,

Thanks for the information. I have been trying to do this, using your
instructions as well as gpct_encoder.c from comedi, and also code from
people who have implemented it in Analogy (on 660x DAQs). The attempts have
been a complete bust so far.

I am able to disarm and arm the counter, and this I know because when I
read from counter 0 (using a4l_sync_read), the value is decremented if the
counter is armed, when a +5V input is applied to PFI0. However, the value
keeps decreasing as long as the input is connected, even though there is no
edge. This does not happen when the counter is disarmed, and also the
initial value is 0 when the counter is disarmed and then armed.

This is the only behavior I am able to produce from the counter. I have
tried using both QUADRATURE_X4_BITS and NORMAL_BITS, as well as a
combination of other things. I don't seem to be able to set the mode of the
counter properly.

Curiously, regardless of whether the chan_desc in the instructions say 0 or
1, it is counter 0 which seems to be affected. I cannot do anything with
counter 1, reading it always gives me a value of 0.

Rosali, do you have some sample code I can test? Even with NORMAL_BITS, I
would think the counter should decrement (or increment) only a single time
a +5V input is connected.

Manu S Madhav
Postdoctoral Associate
The Knierim Lab, Zanvyl Krieger Mind/Brain Institute,
LIMBS Lab, Mechanical Engineering Department
Johns Hopkins University
Baltimore, MD, USA 21218


On Tue, May 27, 2014 at 8:56 AM, Rosali Pyun <whimt@hotmail.com> wrote:

>
>
> > To: manusmad@gmail.com
> > From: pcj@xenomai.sez.to
> > Date: Sat, 24 May 2014 18:46:22 -0700
> > CC: xenomai@xenomai.org
> > Subject: Re: [Xenomai] Troubleshooting Analogy DIO
>
> >
> > Manu S Madhav <manusmad@gmail.com> writes:
> >
> > > One more question: Is anyone aware of an analogy implementation for
> > > using the DAQ counter for quadrature encoding? E.g. by using the
> > > A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER instruction.
> >
> > I know it works in Comedi, but have not tried it in Analogy. You may
> > have to snarf some bit definitions from comedi.
> >
> > Try INSN_CONFIG, with two data
> > { A4L_INSN_CONFIG_SET_COUNTER_MODE,
> > NI_GPCT_COUNTING_MODE_QUADRATURE_X4_BITS |
> > NI_GPCT_COUNTING_DIRECTION_HW_UP_DOWN_BITS
> > }
> >
> > and then again to arm the counter:
> > { A4L_INSN_CONFIG_ARM, NI_GPCT_ARM_IMMEDIATE }
> >
> > Then you should be able to get the counter value back with
> > a4l_sync_read.
> >
> > A4L_INSN_CONFIG_GPCT_QUADRATURE_ENCODER appears to be a half-baked
> > sensoray only thing.
> >
> > -- Paul
> >
>
>
> I was not able to use the counter for quadrature encoder with NI m-series
> DAQ card.
> I ended up using "quadrature clock converter" chip along with the encoder
> and implemented simple counting functionality.
>
> INSN_CONFIG with
> {A4L_INSN_CONFIG_SET_COUNTER_MODE,
> NI_GPCT_COUNTING_MODE_NORMAL_BITS |
> NI_GPCT_COUNTING_DIRECTION_HW_UP_DOWN_BITS
> }
>
> Let me know if you have any luck with quadrature encoder mode.
>
> -Rosali
>

Re: [Xenomai] Troubleshooting Analogy DIO

[Paul Janzen]

Manu S Madhav <manusmad@gmail.com> writes:

> counter is armed, when a +5V input is applied to PFI0. However, the value
> keeps decreasing as long as the input is connected, even though there is no
> edge. This does not happen when the counter is disarmed, and also the
> initial value is 0 when the counter is disarmed and then armed.

You are counting ticks of the internal clock (20 MHz in the original
6602), gated with PFI0. You need to CONFIG_SET_CLK_SRC to count an
external "clock" (edges).

> Curiously, regardless of whether the chan_desc in the instructions say 0 or
> 1, it is counter 0 which seems to be affected. I cannot do anything with
> counter 1, reading it always gives me a value of 0.

Counter 1 is one *subdev* up from counter 0, it is not the second
channel of subdev 2. 

-- Paul

Re: [Xenomai] Troubleshooting Analogy DIO

[Rosali Pyun]

> To: manusmad@gmail.com
> From: pcj@xenomai.sez.to
> Date: Tue, 27 May 2014 08:03:36 -0700
> CC: xenomai@xenomai.org
> Subject: Re: [Xenomai] Troubleshooting Analogy DIO
> 
> Manu S Madhav <manusmad@gmail.com> writes:
> 
> > counter is armed, when a +5V input is applied to PFI0. However, the value
> > keeps decreasing as long as the input is connected, even though there is no
> > edge. This does not happen when the counter is disarmed, and also the
> > initial value is 0 when the counter is disarmed and then armed.
> 
> You are counting ticks of the internal clock (20 MHz in the original
> 6602), gated with PFI0. You need to CONFIG_SET_CLK_SRC to count an
> external "clock" (edges).
> 
> > Curiously, regardless of whether the chan_desc in the instructions say 0 or
> > 1, it is counter 0 which seems to be affected. I cannot do anything with
> > counter 1, reading it always gives me a value of 0.
> 
> Counter 1 is one *subdev* up from counter 0, it is not the second
> channel of subdev 2. 
> 
> -- Paul

I am attaching a sample code, which is based on comedi example.
As Paul said, for NI card, counter 0 is subdevice 11 and counter 1 is subdevice 12.
I am not using GATE functionality at the moment.I also found out that P0.6 and P0.7 are UP/DOWN channel for counter 0 and counter 1. (at least for NI PCI 6629 card)
I hope this helps.
-Rosali 		 	   		  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: counter.cpp
Type: text/x-c
Size: 5432 bytes
Desc: not available
URL: <http://www.xenomai.org/pipermail/xenomai/attachments/20140527/e73d559f/attachment.bin>

Re: [Xenomai] Troubleshooting Analogy DIO

[Manu S Madhav]

Thanks again for your suggestions and code samples. I hadn't realized the
two counters were two separate subdevices.

Turns out I was using lsampl_t instead of unsigned int for the instruction
data structure, and that messed up the clock source instruction. Normal
counting works for now, and I can even get by using that, since my encoder
is high resolution (5000 counts / rev). However, quadrature (and
particularly the index bit) would be really useful.

So I set the counter mode to QUADRATURE_X4_BITS | HW_UP_DOWN_BITS, disabled
both gates, and used A4L_INSN_CONFIG_SET_OTHER_SRC to set a,b sources
using, for e.g.:

    int a = 8;
    unsigned int data[3];
    a4l_insn_t insn;

    insn.type = A4L_INSN_CONFIG;
    insn.idx_subd = idx_subd_CNT;
    insn.chan_desc = 0;
    insn.data_size = sizeof(data[0])*3;

    data[0] = A4L_INSN_CONFIG_SET_OTHER_SRC;
    data[1] = NI_GPCT_SOURCE_ENCODER_A;
    data[2] = NI_GPCT_PFI_OTHER_SELECT(a);
    insn.data = data;

    a4l_snd_insn(&dsc, &insn)

Now the counter gives one value for clockwise rotation (e.g. 0) and another
for counterclockwise (e.g. 1). So clearly it is using both a and b inputs
to determine direction of rotation. And this happens regardless of whether
clock source is set or not.

Getting closer - any thoughts?


Manu S Madhav
Postdoctoral Associate
The Knierim Lab, Zanvyl Krieger Mind/Brain Institute,
LIMBS Lab, Mechanical Engineering Department
Johns Hopkins University
Baltimore, MD, USA 21218


On Tue, May 27, 2014 at 11:54 AM, Rosali Pyun <whimt@hotmail.com> wrote:

>
>
> > To: manusmad@gmail.com
> > From: pcj@xenomai.sez.to
> > Date: Tue, 27 May 2014 08:03:36 -0700
>
> > CC: xenomai@xenomai.org
> > Subject: Re: [Xenomai] Troubleshooting Analogy DIO
> >
> > Manu S Madhav <manusmad@gmail.com> writes:
> >
> > > counter is armed, when a +5V input is applied to PFI0. However, the
> value
> > > keeps decreasing as long as the input is connected, even though there
> is no
> > > edge. This does not happen when the counter is disarmed, and also the
> > > initial value is 0 when the counter is disarmed and then armed.
> >
> > You are counting ticks of the internal clock (20 MHz in the original
> > 6602), gated with PFI0. You need to CONFIG_SET_CLK_SRC to count an
> > external "clock" (edges).
> >
> > > Curiously, regardless of whether the chan_desc in the instructions say
> 0 or
> > > 1, it is counter 0 which seems to be affected. I cannot do anything
> with
> > > counter 1, reading it always gives me a value of 0.
> >
> > Counter 1 is one *subdev* up from counter 0, it is not the second
> > channel of subdev 2.
> >
> > -- Paul
>
> I am attaching a sample code, which is based on comedi example.
>
> As Paul said, for NI card, counter 0 is subdevice 11 and counter 1 is
> subdevice 12.
>
> I am not using GATE functionality at the moment.
> I also found out that P0.6 and P0.7 are UP/DOWN channel for counter 0 and
> counter 1. (at least for NI PCI 6629 card)
>
> I hope this helps.
>
> -Rosali
>

Re: [Xenomai] Troubleshooting Analogy DIO

[Rosali Pyun]

> Now the counter gives one value for clockwise rotation (e.g. 0) and another for counterclockwise (e.g. 1). So clearly it is using both a and b inputs to determine direction of > rotation. And this happens regardless of whether clock source is set or not.


> Getting closer - any thoughts?


I also had same experience with my DAQ card.I couldn't make any progress, so I decided to use the clock converter chip.
-Rosali